We study Online Convex Optimization (OCO) over a convex set $K\subseteq \mathbb R^d$, where in each round $t$ the learner selects $x_t\in K$ and then observes a convex loss $f_t:K\to[0,1]$, with the goal of minimizing regret to the best fixed decision in hindsight. We introduce a unified probing model that generalizes two recent lines of work: sublinear \emph{best-expert} queries in the experts setting, and pairwise (comparison-based) feedback available every round in OCO. In our framework, the learner has a budget of $k\le T$ \emph{pairwise probes}; on a probed round it may query two points and learn which one has smaller loss. Our main result shows that even a \emph{sublinear and noisy} probe budget can provably improve worst-case regret in the full feedback OCO regime. With $k$ $\delta$-noisy pairwise probes, we obtain: $ {\textup{\textsc{Reg}}}_T \le O\left(\min\left\{\sqrt{dT\ln T},; \frac{dT\ln T}{k|1-2\delta|}\right\}\right) $, which is tight (up to logarithmic factors in $T$) across $T$, $k$ and $\delta$. Specifically regarding the noise parameter $\delta \in [0,1]$, the regret guarantee smoothly degrades as the oracle response approaches a coin flip, i.e., $\delta$ is close to $\frac{1}{2}$. When applying the same techniques to a finite $K$ for the prediction with $d$ experts setting, the resulting rates are instead completely tight in all parameters, including $d$. Our analysis gives a streamlined treatment of pairwise probing in OCO by quantifying the benefit of probing via a variance reduction effect, combined with a second-order (variance-based) analysis of Continuous Exponential Weights.

In the online hitting set problem, sets arrive over time, and the algorithm has to maintain a subset of elements that hit all the sets seen so far. Alon, Awerbuch, Azar, Buchbinder, and Naor (SICOMP 2009) gave an algorithm with competitive ratio O(log n log m) for the (general) online hitting set and set cover problems for m sets and n elements; this is known to be tight for efficient online algorithms. Given this barrier for general set systems, we ask: can we break this double-logarithmic phenomenon for online hitting set/set cover on structured and geometric set systems? We provide an O(log n log log n)-competitive algorithm for the weighted online hitting set problem on set systems with linear shallow-cell complexity, replacing the double-logarithmic factor in the general result by effectively a single logarithmic term. As a consequence of our results we obtain the first bounds for weighted online hitting set for natural geometric set families, thereby answering open questions regarding the gap between general and geometric weighted online hitting set problems.

We consider a multiple-choice mixed packing and covering problem in the online setting: at each timestep \(t\), the algorithm faces a collection of \(K\) choices. Each choice consumes some resources, and gives some benefits; both resources and benefits are \(d\)-dimensional vectors. We would like to make a choice for each timestep, such that we use at most \(B\) units of each resource, and we get at least \(B\) units of each kind of benefit. Among its many applications, this general problem captures the question of load-balancing on unrelated machines, where the choices are allocations of jobs to machines.

The problem of minimizing the total flow time on a single machine is one of the few problems for which we can give an optimal online algorithm: just schedule the job with the shortest remaining processing time (SRPT). However, this requires knowledge of the true running time \(p_j\) of each job \(j\). Azar, Leonardi, and Touitou recently asked: what if we are given estimates \(\hat{p}_j\) for each job, such that the multiplicative error between \(p_j\) and \(\hat{p}_j\) (called the distortion) is at most \(\mu\)? It is easy to construct examples where no algorithm can be \(o(\mu)\) competitive; can we get \(O(\mu)\) competitiveness?

In a linear combinatorial optimization problem, we are given a family F ⊆ 2U of feasible subsets of a ground set U of n elements, and our goal is to find S* = argminS ∈ F ⟨ w,1S ⟩. Traditionally, we are either given the weight vector up-front, or else we are given a value oracle which allows us to evaluate w(S) := ⟨ w, 1S ⟩ for any S ∈ F. We consider the weaker and more robust comparison oracle, which for any two feasible sets S, T ∈ F, reveals only if w(S) is less than/equal to/greater than w(T). We ask: When can we find the optimal feasible set S* = argminS ∈ F w(S) using a small number of comparison queries? If so, when can we do this efficiently?

In the Steiner Forest problem, we are given a graph with edge lengths, and a collection of demand pairs; the goal is to find a subgraph of least total length such that each demand pair is connected in this subgraph. For over twenty years, the best approximation ratio known for the problem was a 2-approximation due to Agrawal, Klein, and Ravi (STOC 1991), despite many attempts to surpass this bound. Finally, in a recent breakthrough, Ahmadi, Gholami, Hajiaghayi, Jabbarzade, and Mahdavi (FOCS 2025) gave a 2-ϵ-approximation, where ϵ ≈ 10-11. In this work, we show how to simplify and extend the work of Ahmadi et al. to obtain an improved 1.994-approximation. We combine some ideas from their work (e.g., an extended run of the moat-growing primal-dual algorithm, and identifying autarkic pairs) with other ideas - submodular maximization to find components to contract, as in the relative greedy algorithms for Steiner tree, and the use of autarkic triples. We hope that our cleaner abstraction will open the way for further improvements.

We revisit the classical problem of minimizing the total flow time of jobs on a single machine in the online setting where jobs arrive over time. It has long been known that the Shortest Remaining Processing Time (SRPT) algorithm is optimal (i.e., 1-competitive) when the job sizes are known upfront [Schrage, 1968]. But in the non-clairvoyant setting where job sizes are revealed only when the job finishes, no algorithm can be constant-competitive [Motwani, Phillips, and Torng, 1994]. We consider the $\varepsilon$-clairvoyant setting, where $\varepsilon \in[0,1]$, and each job’s processing time becomes known once its remaining processing time equals an $\varepsilon$ fraction of its processing time. This captures settings where the system user uses the initial $(1-\varepsilon)$ fraction of a job’s processing time to learn its true length, which it can then reveal to the algorithm. The model was proposed by Yingchareonthawornchai and Torng (2017), and it smoothly interpolates between the clairvoyant setting (when $\varepsilon=1$) and the non-clairvoyant setting (when $\varepsilon=0$). In a concrete sense, we are asking: how much knowledge is required to circumvent the hardness of this problem? We show that a little knowledge is enough, and that a constant competitive algorithm exists for every constant $\varepsilon\gt 0$. More precisely, for all $\varepsilon \in(0,1)$, we present a deterministic $\left\lceil\frac{1}{\varepsilon}\right\rceil$-competitive algorithm, which is optimal for deterministic algorithms. We also present a matching lower bound (up to a constant factor) for randomized algorithms. Our algorithm to achieve this bound is remarkably simple and applies the “optimism in the face of uncertainty” principle. For each job, we form an optimistic estimate of its length, based on the information revealed thus far and run SRPT on these optimistic estimates. The proof relies on maintaining a matching between the jobs in OPT’s queue and the algorithm’s queue, with small prefix expansion. We achieve this by carefully choosing a set of jobs to arrive earlier than their release times without changing the algorithm, and possibly helping the adversary. These early arrivals allow us to maintain structural properties inductively, giving us the tight guarantee.

In this paper, we study the parameterized complexity of local search, whose goal is to find a good nearby solution from the given current solution. Formally, given an optimization problem where the goal is to find the largest feasible subset S of a universe U, the new input consists of a current solution P (not necessarily feasible) as well as an ordinary input for the problem. Given the existence of a feasible solution S^*, the goal is to find a feasible solution as good as S^* in parameterized time f(k)⋅n^O(1), where k denotes the distance |PΔ S^*|. This model generalizes numerous classical parameterized optimization problems whose parameter k is the minimum number of elements removed from U to make it feasible, which corresponds to the case P = U. We apply this model to widely studied Constraint Satisfaction Problems (CSPs), where U is the set of constraints, and a subset U' of constraints is feasible if there is an assignment to the variables satisfying all constraints in U'. We give a complete characterization of the parameterized complexity of all boolean-alphabet symmetric CSPs, where the predicate’s acceptance depends on the number of true literals.

_Length generalization_, the ability of sequence models to generalize to sequences longer than those encountered during training, remains a key challenge for transformers, especially in tasks requiring algorithmic reasoning. Existing theoretical understanding of length generalization is limited, often providing only asymptotic results or focusing on specific problem classes or architectural variants, while empirical approaches frequently rely on ad hoc and often fragile techniques. In this work we introduce a novel framework for analyzing and proving length generalization bounds under specified, verifiable assumptions. A key outcome of the theory is the identification of a natural set of _auxiliary_ tasks, intricately related to the primary task structure, such that strong performance on these auxiliary tasks, alongside the primary task, provably guarantees length generalization within the framework. This motivates a multi-task training procedure that explicitly optimizes performance on both the primary and the identified auxiliary tasks. Empirical evaluations on a variety of synthetic benchmarks known to be challenging for length generalization, including sequence sorting, and reversal, demonstrate that our proposed method yields significant improvements in generalization to substantially longer sequences.

We consider the problem of scheduling $m$ jobs on $n$ unrelated strategic machines to minimize the maximum load of any machine, but the machines are strategic and may misreport processing times to minimize their own load. The pioneering work of Nisan and Ronen gave an $n$-approximate deterministic strategyproof mechanism for this setting, and this was recently shown to be best possible by the breakthrough results of Christodoulou et al. This large approxation guarantee begs the question: how can we avoid these large worst-case results. In this work, we use the powerful framework of algorithms with (machine-learned) predictions to bypass these strong impossibility results. We show how we can predict $O(m+n)$ values to obtain a deterministic strategyproof algorithm whose makespan is within a constant factor of the optimal makespan when the predictions are correct, and $O(n)$ times the optimum no matter how poor the predictions are.

We provide an algorithm with regret $O(C d \log \log T)$ for contextual pricing with $C$ corrupted rounds, improving over the previous bound of $O(d^3 C \log^2(T))$ of Krishnamurthy et al. The result is based on a reduction that calls the uncorrupted algorithm as a black-box, unlike the previous approach that modifies the inner workings of the uncorrupted algorithm. As a result, it leads to a conceptually simpler algorithm. Finally, we provide a lower bound ruling out a $O(C + d\log \log T)$ algorithm. This shows that robustifying contextual pricing is harder than robustifying contextual search with $\epsilon$-ball losses, for which it is possible to design algorithms where corruptions add only an extra additive term $C$ to the regret.

We study the problem of finding the optimal bidding strategy for an advertiser in a multi-platform auction setting. The competition on a platform is captured by a value and a cost function, mapping bidding strategies to value and cost respectively. We assume a diminishing returns property, whereby the marginal cost is increasing in value. The advertiser uses an autobidder that selects a bidding strategy for each platform, aiming to maximize total value subject to budget and return-on-spend constraint. The advertiser has no prior information and learns about the value and cost functions by querying a platform with a specific bidding strategy. Our goal is to design algorithms that find the optimal bidding strategy with a small number of queries.

We initiate the study of centralized algorithms for welfare-maximizing allocation of goods to buyers subject to average-value constraints. We show that this problem is NP-hard to approximate beyond a factor of $\frac{e}{e-1}$, and provide a $\frac{4e}{e-1}$-approximate offline algorithm. For the online setting, we show that no non-trivial approximations are achievable under adversarial arrivals. Under i.i.d. arrivals, we present a polytime online algorithm that provides a constant approximation of the optimal (computationally-unbounded) online algorithm. In contrast, we show that no constant approximation of the ex-post optimum is achievable by an online algorithm.

Even for simple arithmetic tasks like integer addition, it is challenging for Transformers to generalize to longer sequences than those encountered during training. To tackle this problem, we propose *position coupling*, a simple yet effective method that directly embeds the structure of the tasks into the positional encoding of a (decoder-only) Transformer. Taking a departure from the vanilla absolute position mechanism assigning unique position IDs to each of the tokens, we assign the same position IDs to two or more "relevant" tokens; for integer addition tasks, we regard digits of the same significance as in the same position. On the empirical side, we show that with the proposed position coupling, our models trained on 1 to 30-digit additions can generalize up to *200-digit* additions (6.67x of the trained length). On the theoretical side, we prove that a 1-layer Transformer with coupled positions can solve the addition task involving exponentially many digits, whereas any 1-layer Transformer without positional information cannot entirely solve it. We also demonstrate that position coupling can be applied to other algorithmic tasks such as Nx2 multiplication and a two-dimensional task. Our codebase is available at [github.com/HanseulJo/position-coupling](https://github.com/HanseulJo/position-coupling).

We study a setting where agents use no-regret learning algorithms to participate in repeated auctions. Recently, Kolumbus and Nisan [2022a] showed, rather surprisingly, that when bidders participate in second-price auctions using no-regret bidding algorithms, no matter how large the number of interactions $T$ is, the runner-up bidder may not converge to bidding truthfully. Our first result shows that this holds forall deterministictruthful auctions. We also show that the ratio of the learning rates of different bidders can qualitatively affect the convergence of the bidders. Next, we consider the problem of revenue maximization in this environment. In the setting with fully rational bidders, the seminal result of Myerson [1981] showed that revenue can be maximized by using a second-price auction with reserves. We show that, in stark contrast, in our setting with learning bidders, randomized auctions can have strictly better revenue guarantees than second-price auctions with reserves, when $T$ is large enough. To do this, we provide a black-box transformation from any truthful auction $A$ to an auction $A'$ such that: i) all mean-based no-regret learners that participate in $A'$ converge to bidding truthfully, ii) the distance between the allocation rule and the payment rule between $A, A'$ is negligible. Finally, we study revenue maximization in the non-asymptotic regime. We define a notion of auctioneer regret that compares the revenue generated to the revenue of a second price auction with truthful bids. When the auctioneer has to use the same auction throughout the interaction, we show an (almost) tight regret bound of $\tilde{\Theta}(T^{3/4})$. Then, we consider the case where the auctioneer can use different auctions throughout the interaction, but in a way that is oblivious to the bids. For this setting, we show an (almost) tight bound of $\tilde{\Theta}(\sqrt{T})$.

Algorithms with predictions are gaining traction across various domains, as a way to surpass traditional worst-case bounds through (machine-learned) advice. We study the canonical problem of $k$-facility location mechanism design, where the $n$ agents are strategic and might misreport their locations. We receive a prediction for each agent's location, and these predictions are crucially allowed to be only "mostly" and "approximately" correct (MAC for short): a $\delta$-fraction of the predicted locations are allowed to be arbitrarily incorrect, and the remainder of the predictions are required to be correct up to an $\varepsilon$-error. Moreover, we make no assumption on the independence of the errors. Can such "flawed" predictions allow us to beat the current best bounds for strategyproof facility location? We show how natural robustness of the $1$-median (also known as the geometric median) of a set of points leads to an algorithm for single-facility location with MAC predictions. We extend our results to a natural "balanced" variant of the $k$-facility case, and show that without balancedness, robustness completely breaks down even for $k=2$ facilities on a line. As our main result, for this "unbalanced" setting we devise a truthful random mechanism, which outperforms the best known mechanism (with no predictions) by Lu et al.~[2010]. En route, we introduce the problem of "second" facility location, in which the first facility location is already fixed. Our robustness findings may be of independent interest, as quantitative versions of classic breakdown-point results in robust statistics.

In recent years, there has been a surge of interest in the use of machine-learned predictions to bypass worst-case lower bounds for classical problems in combinatorial optimization. So far, the focus has mostly been on online algorithms, where information-theoretic barriers are overcome using predictions about the unknown future. In this paper, we consider the complementary question of using learned information to overcome computational barriers in the form of approximation hardness of polynomial-time algorithms for NP-hard (offline) problems. We show that noisy predictions about the optimal solution can be used to break classical hardness results for maximization problems such as the max-cut problem and more generally, maximization versions of constraint satisfaction problems (CSPs).

Maintaining a maximum bipartite matching online while minimizing augmentations is a well studied problem, motivated by content delivery, job scheduling, and hashing. A breakthrough result of Bernstein, Holm, and Rotenberg (SODA 2018) resolved this problem up to a logarithmic factors. However, to model other problems in scheduling and resource allocation, we may need a richer class of combinatorial constraints (e.g., matroid constraints).

In the stochastic set cover problem (Grandoni et al., FOCS ‘08), we are given a collection S of m sets over a universe U of size N, and a distribution D over elements of U. The algorithm draws n elements one-by-one from D and must buy a set to cover each element on arrival; the goal is to minimize the total cost of sets bought during this process. A universal algorithm a priori maps each element u ∈ U to a set S(u) such that if U ⊆ U is formed by drawing n times from distribution D, then the algorithm commits to outputting S(U). Grandoni et al. gave an O(log mN)-competitive universal algorithm for this stochastic set cover problem.

The online k-taxi problem generalizes the k-server problem, requiring servers to move between source-sink pairs in an n-point metric space, and the cost is the overhead incurred. In the deterministic setting, the problem has a lower bound on the competitiveness of Ω(2k), showing that it is significantly harder than k-server. Randomized algorithms are known with competitiveness O(2k log n) (by Coester and Koutsoupias), (by Buchbinder, Coester and Naor), where Δ is the aspect ratio of the n-point metric space), and O((n log k)2 log n) (by Bubeck, Buchbinder, Coester, and Sellke). The best lower bound known is Ω(log2 k) which is inherited from the k-server problem, obtained in a recent breakthrough by Bubeck, Coester, and Rabani, showing a large gap in our understanding of problems that go slightly beyond the metrical task system framework.

Given a set of points in d-dimensional space, an explainable clustering is one where the clusters are specified by a tree of axis-aligned threshold cuts. Dasgupta et al. (ICML 2020) posed the question of the price of explainability: the worst-case ratio between the cost of the best explainable clusterings to that of the best clusterings.We show that the price of explainability for k medians is at most $1+H_{k-1}$; in fact, we show that the popular Random Thresholds algorithm has exactly this price of explainability, matching the known lower bound constructions. We complement our tight analysis of this particular algorithm by constructing instances where the price of explainability (using any algorithm) is at least $(1-o(1)) \ln k$, showing that our result is best possible, up to lower-order terms. We also improve the price of explainability for the k-means problem to $O(k \ln \ln k)$ from the previous $O(k \ln k)$, considerably closing the gap to the lower bounds of $\Omega(k)$. Finally, we study the algorithmic question of finding the best explainable clustering: We show that explainable k medians and k-means cannot be approximated better than $O(\ln k)$, under standard complexity-theoretic conjectures. This essentially settles the approximability of explainable k-medians and leaves open the intriguing possibility to get significantly better approximation algorithms for k-means than its price of explainability.

We study the classic problem of minimizing the expected total completion time of jobs on m identical machines in the setting where the sizes of the jobs are stochastic. Specifically, the size of each job is a random variable whose distribution is known to the algorithm, but whose realization is revealed only after the job is scheduled. While minimizing the total completion time is easy in the deterministic setting, the stochastic problem has long been notorious: all known algorithms have approximation ratios that either depend on the variances, or depend linearly on the number of machines.

Abstract. This note points out an error in the proof of Theorem 4 in the article “Metric Embedding via Shortest Path Decompositions,” SIAM J. Comput., 51 (2022), pp. 290–314, by the authors, and withdraws the associated claim of Theorem 4.

We develop approximation algorithms for set-selection problems with deterministic constraints, but random objective values, i.e., stochastic probing problems. When the goal is to maximize the objective, approximation algorithms for probing problems are well-studied. On the other hand, few techniques are known for minimizing the objective, especially in the adaptive setting, where information about the random objective is revealed during the set-selection process and allowed to influence it. For minimization problems in particular, incorporating adaptivity can have a considerable effect on performance. In this work, we seek approximation algorithms that compare well to the optimal adaptive policy. We develop new techniques for adaptive minimization, applying them to a few problems of interest. The core technique we develop here is an approximate reduction from an adaptive expectation minimization problem to a set of adaptive probability minimization problems which we call threshold problems. By providing near-optimal solutions to these threshold problems, we obtain bicriteria adaptive policies. We apply this method to obtain an adaptive approximation algorithm for the Min-Element problem, where the goal is to adaptively pick random variables to minimize the expected minimum value seen among them, subject to a knapsack constraint. This partially resolves an open problem raised in [Goel et al., 2010]. We further consider three extensions on the Min-Element problem, where our objective is the sum of the smallest k element-weights, or the weight of the min-weight basis of a given matroid, or where the constraint is not given by a knapsack but by a matroid constraint. For all three of the variations we explore, we develop adaptive approximation algorithms for their corresponding threshold problems, and prove their near-optimality via coupling arguments.

The growing body of work in learning-augmented online algorithms studies how online algorithms can be improved when given access to ML predictions about the future. Motivated by ML models that give a confidence parameter for their predictions, we study online algorithms with predictions that are $\epsilon$-accurate: namely, each prediction is correct with probability (at least) $\epsilon$, but can be arbitrarily inaccurate with the remaining probability. We show that even with predictions that are accurate with a small probability and arbitrarily inaccurate otherwise, we can dramatically outperform worst-case bounds for a range of classical online problems including caching, online set cover, and online facility location. Our main results are an $O(\log(1/\varepsilon))$-competitive algorithm for caching, and a simple $O(1/\varepsilon)$-competitive algorithm for a large family of covering problems, including set cover and facility location, with $\epsilon$-accurate predictions.

We consider three central problems in optimization: the restricted assignment load-balancing problem, the Steiner tree network design problem, and facility location clustering. We consider the online setting, where the input arrives over time, and irrevocable decisions must be made without knowledge of the future. For all these problems, any online algorithm must incur a cost that is approximately $\log |I|$ times the optimal cost in the worst-case, where $|I|$ is the length of the input. But can we go beyond the worst-case? In this work we give algorithms that perform substantially better when a $p$-fraction of the input is given as a sample: the algorithm use this sample to \emph{learn} a good strategy to use for the rest of the input.

We study the problem of solving Packing Integer Programs (PIPs) in the online setting, where columns in [0, 1]d of the constraint matrix are revealed sequentially, and the goal is to pick a subset of the columns that sum to at most B in each coordinate while maximizing the objective. Excellent results are known in the secretary setting, where the columns are adversarially chosen, but presented in a uniformly random order. However, these existing algorithms are susceptible to adversarial attacks: they try to “learn” characteristics of a good solution, but tend to over-fit to the model, and hence a small number of adversarial corruptions can cause the algorithm to fail. In this paper, we give the first robust algorithms for Packing Integer Programs, specifically in the recently proposed Byzantine Secretary framework [BGSZ20]. Our techniques are based on a two-level use of online learning, to robustly learn an approximation to the optimal value, and then to use this robust estimate to pick a good solution. These techniques are general and we use them to design robust algorithms for PIPs in the prophet model as well, specifically in the Prophet-with-Augmentations framework [ISW20]. We also improve known results in the Byzantine Secretary framework: we make the non-constructive results algorithmic and improve the existing bounds for single-item and matroid constraints.

We present a new local-search algorithm for the k-median clustering problem. We show that local optima for this algorithm give a (2.836 + ∊)-approximation; our result improves upon the (3 + ∊)-approximate local-search algorithm of Arya et al. [AGK+01]. Moreover, a computer-aided analysis of a natural extension suggests that this approach may lead to an improvement over the best-known approximation guarantee for the problem. The new ingredient in our algorithm is the use of a potential function based on both the closest and second-closest facilities to each client. Specifically, the potential is the sum over all clients, of the distance of the client to its closest facility, plus (a small constant times) the truncated distance to its second-closest facility. We move from one solution to another only if the latter can be obtained by swapping a constant number of facilities, and has a smaller potential than the former. This refined potential allows us to avoid the bad local optima given by Arya et al. for the local-search algorithm based only on the cost of the solution.

The vector-balancing problem is a fundamental problem in discrepancy theory: given T vectors in [–1, 1]n, find a signing σ(a) ∊ {±1} of each vector a to minimize the discrepancy ‖ Σa σ(a) · a‖∞. This problem has been extensively studied in the static/offline setting. In this paper we initiate its study in the fully-dynamic setting with recourse: the algorithm sees a stream of T insertions and deletions of vectors, and at each time must maintain a low-discrepancy signing, while also minimizing the amortized recourse (the number of times any vector changes its sign) per update. For general vectors, we show algorithms which almost match Spencer's offline discrepancy bound, with O(n polylog T) amortized recourse per update. The crucial idea behind our algorithm is to compute a basic feasible solution to the linear relaxation in a distributed and recursive manner, which helps find a low-discrepancy signing. We bound the recourse using the distributed computation of the basic solution, and argue that only a small part of the instance needs to be re-computed at each update. Since vector balancing has also been greatly studied for sparse vectors, we then give algorithms for low-discrepancy edge orientation, where we dynamically maintain signings for 2-sparse vectors in an n-dimensional space. Alternatively, this can be seen as orienting a dynamic set of edges of an n-vertex graph to minimize the discrepancy, i.e., the absolute difference between in- and out-degrees at any vertex. We present a deterministic algorithm with O(polylog n) discrepancy and O(polylog n) amortized recourse. The core ideas are to dynamically maintain an expander-decomposition with low recourse (using a very simple approach), and then to show that, as the expanders change over time, a natural local-search algorithm converges quickly (i.e., with low recourse) to a low-discrepancy solution. We also give strong lower bounds (with some matching upper bounds) for local-search discrepancy minimization algorithms for vector balancing and edge orientation.

In the k -cut problem, we want to find the lowest-weight set of edges whose deletion breaks a given (multi)graph into k connected components. Algorithms of Karger and Stein can solve this in roughly O ( n 2k ) time. However, lower bounds from conjectures about the k -clique problem imply that Ω ( n (1- o (1)) k ) time is likely needed. Recent results of Gupta, Lee, and Li have given new algorithms for general k -cut in n 1.98k + O(1) time, as well as specialized algorithms with better performance for certain classes of graphs (e.g., for small integer edge weights). In this work, we resolve the problem for general graphs. We show that the Contraction Algorithm of Karger outputs any fixed k -cut of weight α λ k with probability Ω k ( n - α k ), where λ k denotes the minimum k -cut weight. This also gives an extremal bound of O k ( n k ) on the number of minimum k -cuts and an algorithm to compute λ k with roughly n k polylog( n ) runtime. Both are tight up to lower-order factors, with the algorithmic lower bound assuming hardness of max-weight k -clique. The first main ingredient in our result is an extremal bound on the number of cuts of weight less than 2 λ k / k , using the Sunflower lemma. The second ingredient is a fine-grained analysis of how the graph shrinks—and how the average degree evolves—in the Karger process.

We study the problem of embedding shortest-path metrics of weighted graphs into $\ell_p$ spaces. We introduce a new embedding technique based on low-depth decompositions of a graph via shortest paths. The notion of shortest path decomposition (SPD) depth is inductively defined: A (weighed) path graph has SPD depth $1$. General graph has an SPD of depth $k$ if it contains a shortest path whose deletion leads to a graph, each of whose components has SPD depth at most $k-1$. In this paper we give an $O(k^{\min\{\nicefrac{1}{p},\nicefrac{1}{2}\}})$-distortion embedding for graphs of SPD depth at most $k$. This result is asymptotically tight for any fixed $p>1$, while for $p=1$ it is tight up to second order terms. As a corollary of this result, we show that graphs having pathwidth $k$ embed into $\ell_p$ with distortion $O(k^{\min\{\nicefrac{1}{p},\nicefrac{1}{2}\}})$. For $p=1$, this improves over the best previous bound of Lee and Sidiropoulos that was exponential in $k$; moreover, for other values of $p$ it gives the first embeddings whose distortion is independent of the graph size $n$. Furthermore, we use the fact that planar graphs have SPD depth $O(\log n)$ to give a new proof that any planar graph embeds into $\ell_1$ with distortion $O(\sqrt{\log n})$. Our approach also gives new results for graphs with bounded treewidth, and for graphs excluding a fixed minor.

We consider two generalizations of the classical weighted paging problem that incorporate the notion of delayed service of page requests. The first is the (weighted) paging with time windows (\sf PageTW) problem, which is like the classical weighted paging problem except that each page request only needs to be served before a given deadline. This problem arises in many practical applications of online caching, such as the “deadline” I/O scheduler in the Linux kernel and video-on-demand streaming. The second, and more general, problem is the (weighted) paging with delay (\sf PageD) problem, where the delay in serving a page request results in a penalty being added to the objective. This problem generalizes the caching problem to allow delayed service, a line of work that has recently gained traction in online algorithms (e.g., [Y. Emek, S. Kutten, and R. Wattenhofer, Proceedings of the 48 th Annual ACM SIGACT Symposium on Theory of Computing, 2016, pp. 333--344; Y. Azar et al., Proceedings of the 49 th Annual ACM SIGACT Symposium on Theory of Computing, 2017, pp. 551--563; Y. Azar and N. Touitou, Proceedings of the 60 th IEEE Annual Symposium on Foundations of Computer Science, 2019, pp. 60--71]). We give $O(\log k\log n)$-competitive algorithms for both the \sf PageTW and \sf PageD problems on $n$ pages with a cache of size $k$. This significantly improves on the previous best bounds of $O(k)$ for both problems [Y. Azar et al., Proceedings of the 49 th Annual ACM SIGACT Symposium on Theory of Computing, 2017, pp. 551--563]. We also consider the offline \sf PageTW and \sf PageD problems, for which we give $O(1)$-approximation algorithms and prove APX-hardness. These are the first results for the offline problems; even NP-hardness was not known before our work. At the heart of our algorithms is a novel “hitting-set” LP relaxation of the \sf PageTW problem that overcomes the $\Omega(k)$ integrality gap of the natural LP for the problem. To the best of our knowledge, this is the first example of an LP-based algorithm for an online problem with delays/deadlines.

We study the$k$-server problem with time-windows. In this problem, each request$i$arrives at some point$v_{i}$of an$n$-point metric space at time$b_{i}$and comes with a deadline$e_{i}$. One of the$k$servers must be moved to$v_{i}$at some time in the interval [$b_{i}, e_{i}$] to satisfy this request. We give an online algorithm for this problem with a competitive ratio of$\text{poly}\log(n, \Delta)$, where$\Delta$is the aspect ratio of the metric space. Prior to our work, the best competitive ratio known for this problem was$O(k\ \text{poly}\log(n))$given by Azar et al. (STOC 2017). Our algorithm is based on a new covering linear program relaxation for$k$-server on HSTs. This LP naturally corresponds to the min-cost flow formulation of$k$-server, and easily extends to the case of time-windows. We give an online algorithm for obtaining a feasible fractional solution for this LP, and a primal dual analysis framework for accounting the cost of the solution. Together, they yield a new$k$-server algorithm with poly-logarithmic competitive ratio, and extend to the time-windows case as well. Our principal technical contribution lies in thinking of the covering LP as yielding a truncated covering LP at each internal node of the tree, which allows us to keep account of server movements across subtrees. We hope that this LP relaxation and the algorithm/analysis will be a useful tool for addressing$k$-server and related problems.

We give a polynomial-time algorithm for Online-SetCover with a competitive ratio of$O(\log mn)$when the elements are revealed in random order, matching the best possible offline bound of$O(\log n)$when the number of sets$m$is polynomial in the number of elements$n$, and circumventing the$\Omega(\log m \log n)$lower bound known in adversarial order. We also extend the result to solving pure covering IPs when constraints arrive in random order. The algorithm is a multiplicative-weights-based round-and-solve approach we call LearnOrCover. We maintain a coarse fractional solution that is neither feasible nor monotone increasing, but can nevertheless be rounded online to achieve the claimed guarantee (in the random order model). This gives a new offline algorithm for Setcover that performs a single pass through the elements, which may be of independent interest.

This paper considers approximation algorithms for generalized $k$-median problems. This class of problems can be informally described as $k$-median with a constant number of extra constraints, and includes $k$-median with outliers, and knapsack median. Our first contribution is a pseudo-approximation algorithm for generalized $k$-median that outputs a $6.387$-approximate solution, with a constant number of fractional variables. The algorithm builds on the iterative rounding framework introduced by Krishnaswamy, Li, and Sandeep for $k$-median with outliers. The main technical innovation is allowing richer constraint sets in the iterative rounding and taking advantage of the structure of the resulting extreme points. Using our pseudo-approximation algorithm, we give improved approximation algorithms for $k$-median with outliers and knapsack median. This involves combining our pseudo-approximation with pre- and post-processing steps to round a constant number of fractional variables at a small increase in cost. Our algorithms achieve approximation ratios $6.994 + ε$ and $6.387 + ε$ for $k$-median with outliers and knapsack median, respectively. These improve on the best-known approximation ratio $7.081 + ε$ for both problems \cite{DBLP:conf/stoc/KrishnaswamyLS18}.

In the stochastic submodular cover problem, the goal is to select a subset of stochastic items of minimum expected cost to cover a submodular function. Solutions in this setting correspond to a sequential decision process that selects items one by one “adaptively” (depending on prior observations). While such adaptive solutions achieve the best objective, the inherently sequential nature makes them undesirable in many applications. We ask: \emph{how well can solutions with only a few adaptive rounds approximate fully-adaptive solutions?} We consider both cases where the stochastic items are independent, and where they are correlated. For both situations, we obtain nearly tight answers, establishing smooth tradeoffs between the number of adaptive rounds and the solution quality, relative to fully adaptive solutions. Experiments on synthetic and real datasets validate the practical performance of our algorithms, showing qualitative improvements in the solutions as we allow more rounds of adaptivity; in practice, solutions using just a few rounds of adaptivity are nearly as good as fully adaptive solutions.

Consider a random graph model where there is an underlying simple graph G = (V, E), and each edge is sampled independently with probability p ∊ [0, 1]. What is the smallest value of p such that the resulting graph Gp is connected with constant probability? This is a well-studied question for special classes of graphs, such as complete graphs and hypercubes. For instance, when G is the complete graph, we want the connectivity threshold for the Erdős-Rényi G(n, p) model: here the answer is known to be . However, the problem is not well-understood for more general graph classes. We first investigate this connectivity threshold problem for “somewhat dense” graphs. We show that for any and any δ-regular, δ-edge-connected graph G, the random graph Gp for is connected with probability , generalizing upon the case when G is the complete graph. Our proof also bounds the number of approximate mincuts in such a dense graph, which may be of independent interest. Next, for a general graph G with edge connectivity λ, we define an explicit parameter βG ∊ (0, 2 ln n], based on the number of approximate mincuts, and show that there is a sharp transition in the connectivity of G at p = 1 – exp(βG/λ). Moreover, we show that the width of this transition is an additive O(ln λ/λ) term; this improves upon Margulis' classical result bounding the width of the threshold by .

The problem of chasing convex functions is easy to state: faced with a sequence of convex functions f t over d-dimensional Euclidean spaces, the goal of the algorithm is to output a point x t at each time, so that the sum of the function costs f t (x t ), plus the movement costs ||x t − x t − 1 || is minimized. This problem generalizes questions in online algorithms such as caching and the k-server problem. In 1994, Friedman and Linial posed the question of getting an algorithm with a competitive ratio that depends only on the dimension d. In this talk we give an O (d)-competitive algorithm, based on the notion of the Steiner point of a convex body.

The (non-uniform) sparsest cut problem is the following graph-partitioning problem: given a “supply” graph, and demands on pairs of vertices, delete some subset of supply edges to minimize the ratio of the supply edges cut to the total demand of the pairs separated by this deletion. Despite much effort, there are only a handful of nontrivial classes of supply graphs for which constant-factor approximations are known.

We study the problem of chasing convex bodies online: given a sequence of convex bodies the algorithm must respond with points in an online fashion (i.e., is chosen before is revealed). The objective is to minimize the sum of distances between successive points in this sequence. Bubeck et al. (STOC 2019) gave a -competitive algorithm for this problem. We give an algorithm that is -competitive for any sequence of length .

We consider the problem of chasing convex functions, where functions arrive over time. The player takes actions after seeing the function, and the goal is to achieve a small function cost for these actions, as well as a small cost for moving between actions. While the general problem requires a polynomial dependence on the dimension, we show how to get dimension-independent bounds for well-behaved functions. In particular, we consider the case where the convex functions are $\kappa$-well-conditioned, and give an algorithm that achieves an $O(\sqrt \kappa)$-competitiveness. Moreover, when the functions are supported on $k$-dimensional affine subspaces—e.g., when the function are the indicators of some affine subspaces—we get $O(\min(k, \sqrt{k \log T}))$-competitive algorithms for request sequences of length $T$. We also show some lower bounds, that well-conditioned functions require $\Omega(\kappa^{1/3})$-competitiveness, and $k$-dimensional functions require $\Omega(\sqrt{k})$-competitiveness.

In submodular covering problems, we are given a monotone, nonnegative submodular function f:2N→ R+and wish to find the min-cost set S ⊆ N such that f(S)=f(N). When f is a coverage function, this captures Setcover as a special case. We introduce a general framework for solving such problems in a fully-dynamic setting where the function f changes over time, and only a bounded number of updates to the solution (a.k.a. recourse) is allowed. For concreteness, suppose a nonnegative monotone submodular integer-valued function gt is added or removed from an active set G(t)at each time t. If f(t)=Σ(g∈G(t)g) is the sum of all active functions, we wish to maintain a competitive solution to Submodularcover for f(t)as this active set changes, and with low recourse. For example, if each gt is the (weighted) rank function of a matroid, we would be dynamically maintaining a low-cost common spanning set for a changing collection of matroids. We give an algorithm that maintains an O(log(fmax/fmin)) - competitive solution, where fmax, fminare the largest/smallest marginals of f(t). The algorithm guarantees a total recourse of O(log(cmax/cmin)·Σt≤Tgt(N)), where cmax, cminare the largest/smallest costs of elements in N. This competitive ratio is best possible even in the offline setting, and the recourse bound is optimal up to the logarithmic factor. For monotone sub-modular functions that also have positive mixed third derivatives, we show an optimal recourse bound of O(Σt≤Tgt(N)). This structured class includes set-coverage functions, so our algorithm matches the known O(log n)-competitiveness and O(1) recourse guarantees for fully-dynamic Setcover. Our work simultaneously simplifies and unifies previous results, as well as generalizes to a significantly larger class of covering problems. Our key technique is a new potential function inspired by Tsallis entropy. We also extensively use the idea of Mutual Coverage, which generalizes the classic notion of mutual information.

We consider the online carpooling problem: given $n$ vertices, a sequence of edges arrive over time. When an edge $e_t = (u_t, v_t)$ arrives at time step $t$, the algorithm must orient the edge either as $v_t \rightarrow u_t$ or $u_t \rightarrow v_t$, with the objective of minimizing the maximum discrepancy of any vertex, i.e., the absolute difference between its in-degree and out-degree. Edges correspond to pairs of persons wanting to ride together, and orienting denotes designating the driver. The discrepancy objective then corresponds to every person driving close to their fair share of rides they participate in. In this paper, we design efficient algorithms which can maintain polylog$(n,T)$ maximum discrepancy (w.h.p) over any sequence of $T$ arrivals, when the arriving edges are sampled independently and uniformly from any given graph $G$. This provides the first polylogarithmic bounds for the online (stochastic) carpooling problem. Prior to this work, the best known bounds were $O(\sqrt{n \log n})$-discrepancy for any adversarial sequence of arrivals, or $O(\log\!\log n)$-discrepancy bounds for the stochastic arrivals when $G$ is the complete graph. The technical crux of our paper is in showing that the simple greedy algorithm, which has provably good discrepancy bounds when the arriving edges are drawn uniformly at random from the complete graph, also has polylog discrepancy when $G$ is an expander graph. We then combine this with known expander-decomposition results to design our overall algorithm.

In classical secretary problems, a sequence of n elements arrive in a uniformly random order, and we want to choose a single item, or a set of size K. The random order model allows us to escape from the strong lower bounds for the adversarial order setting, and excellent algorithms are known in this setting. However, one worrying aspect of these results is that the algorithms overfit to the model: they are not very robust. Indeed, if a few "outlier" arrivals are adversarially placed in the arrival sequence, the algorithms perform poorly. E.g., Dynkin’s popular 1/e-secretary algorithm is sensitive to even a single adversarial arrival: if the adversary gives one large bid at the beginning of the stream, the algorithm does not select any element at all. We investigate a robust version of the secretary problem. In the Byzantine Secretary model, we have two kinds of elements: green (good) and red (rogue). The values of all elements are chosen by the adversary. The green elements arrive at times uniformly randomly drawn from [0,1]. The red elements, however, arrive at adversarially chosen times. Naturally, the algorithm does not see these colors: how well can it solve secretary problems? We show that selecting the highest value red set, or the single largest green element is not possible with even a small fraction of red items. However, on the positive side, we show that these are the only bad cases, by giving algorithms which get value comparable to the value of the optimal green set minus the largest green item. (This benchmark reminds us of regret minimization and digital auctions, where we subtract an additive term depending on the "scale" of the problem.) Specifically, we give an algorithm to pick K elements, which gets within (1-ε) factor of the above benchmark, as long as K ≥ poly(ε^{-1} log n). We extend this to the knapsack secretary problem, for large knapsack size K. For the single-item case, an analogous benchmark is the value of the second-largest green item. For value-maximization, we give a poly log^* n-competitive algorithm, using a multi-layered bucketing scheme that adaptively refines our estimates of second-max over time. For probability-maximization, we show the existence of a good randomized algorithm, using the minimax principle. We hope that this work will spur further research on robust algorithms for the secretary problem, and for other problems in sequential decision-making, where the existing algorithms are not robust and often tend to overfit to the model.

Sortition is a political system in which decisions are made by panels of randomly selected citizens. The process for selecting a sortition panel is traditionally thought of as uniform sampling without replacement, which has strong fairness properties. In practice, however, sampling without replacement is not possible since only a fraction of agents is willing to participate in a panel when invited, and different demographic groups participate at different rates. In order to still produce panels whose composition resembles that of the population, we develop a sampling algorithm that restores close-to-equal representation probabilities for all agents while satisfying meaningful demographic quotas. As part of its input, our algorithm requires probabilities indicating how likely each volunteer in the pool was to participate. Since these participation probabilities are not directly observable, we show how to learn them, and demonstrate our approach using data on a real sortition panel combined with information on the general population in the form of publicly available survey data.

We study the problem of chasing convex bodies online: given a sequence of convex bodies Kt ⊆ ℝd the algorithm must respond with points xt ϵ Kt in an on-line fashion (i.e., xt is chosen before Kt+1 is revealed). The objective is to minimize the total distance between successive points in this sequence. Recently, Bubeck et al. (STOC 2019) gave a 2O(d)-competitive algorithm for this problem. We give an algorithm that is -competitive for any sequence of length T.

In the submodular cover problem, we are given a monotone submodular function f: 2N → ℝ+, and we want to pick the min-cost set S such that f (S) = f (N). This captures the set cover problem when f is a coverage function. Motivated by problems in network monitoring and resource allocation, we consider the submodular cover problem in an online setting. As a concrete example, suppose at each time t, a nonnegative monotone submodular function gt is given to us. We define as the sum of all functions seen so far. We need to maintain a submodular cover of these submodular functions f(1), f(2), … f(T) in an online fashion; i.e., we cannot revoke previous choices. Formally, at each time t we produce a set St ⊆ N such that f(t)(St) = f(t)(N)—i.e., this set St is a cover—such that St–1 ⊆ St, so previously decisions to pick elements cannot be revoked. (We actually allow more general sequences {f(t)} of submodular functions, but this sum-of-simpler-submodular-functions case is useful for concreteness.) We give polylogarithmic competitive algorithms for this online submodular cover problem. The competitive ratio on an input sequence of length T is O(ln n ln(T · fmax/fmin)), where fmax and fmin are the largest and smallest marginals for functions f(t), and |N| = n. For the special case of online set cover, our competitive ratio matches that of Alon et al. [AAA+09], which are best possible for polynomial-time online algorithms unless NP ⊆ BPP [Kor04]. Since existing offline algorithms for submodular cover are based on greedy approaches which seem difficult to implement online, the technical challenge is to (approximately) solve the exponential-sized linear programming relaxation for submodular cover, and to round it, both in the online setting. Moreover, to get our competitiveness bounds, we define a (seemingly new) generalization of mutual information to general submodular functions, which we call mutual coverage; we hope this will be useful in other contexts.

We consider the (weighted) Paging with Time Windows problem, which is identical to the classical weighted paging problem but where each page request only needs to be served by a given deadline. This problem arises in many practical applications of online caching, such as the deadline I/O scheduler in the Linux kernel and video-on-demand streaming. From a theoretical perspective, this generalizes the caching problem to allow delayed service, a line of work that has recently gained traction in online algorithms (e.g., Emek et al. STOC '16, Azar et al. STOC '17, Azar and Touitou FOCS '19, etc.).

In the k-cut problem, we are given an edge-weighted graph and want to find the least-weight set of edges whose deletion breaks the graph into k connected components. Algorithms due to Karger-Stein and Thorup showed how to find such a minimum k-cut in time approximately O(n 2k−2). The best lower bounds come from conjectures about the solvability of the k-clique problem and a reduction from k-clique to k-cut, and show that solving k-cut is likely to require time Ω(n k ). Our recent results have given special-purpose algorithms that solve the problem in time n 1.98k + O(1), and ones that have better performance for special classes of graphs (e.g., for small integer weights).

We study the stochastic multi-armed bandits problem in the presence of adversarial corruption. We present a new algorithm for this problem whose regret is nearly optimal, substantially improving upon previous work. Our algorithm is agnostic to the level of adversarial contamination and can tolerate a significant amount of corruption with virtually no degradation in performance.

We consider the online problem of scheduling jobs on identical machines, where jobs have precedence constraints. We are interested in the demanding setting where the jobs sizes are not known up-front, but are revealed only upon completion (the non-clairvoyant setting). Such precedence-constrained scheduling problems routinely arise in map-reduce and large-scale optimization. For minimizing the total weighted completion time, we give a constant-competitive algorithm. And for total weighted flow-time, we give an O(1/epsilon^2)-competitive algorithm under (1+epsilon)-speed augmentation and a natural "no-surprises" assumption on release dates of jobs (which we show is necessary in this context). Our algorithm proceeds by assigning virtual rates to all waiting jobs, including the ones which are dependent on other uncompleted jobs. We then use these virtual rates to decide on the actual rates of minimal jobs (i.e., jobs which do not have dependencies and hence are eligible to run). Interestingly, the virtual rates are obtained by allocating time in a fair manner, using a Eisenberg-Gale-type convex program (which we can solve optimally using a primal-dual scheme). The optimality condition of this convex program allows us to show dual-fitting proofs more easily, without having to guess and hand-craft the duals. This idea of using fair virtual rates may have broader applicability in scheduling problems.

In this paper we initiate a systematic study of exact algorithms for some of the well known clustering problems, namely k-MEDIAN and k-MEANS. In k-MEDIAN, the input consists of a set X of n points belonging to a metric space, and the task is to select a subset C ⊆ X of k points as centers, such that the sum of the distances of every point to its nearest center is minimized. In k-MEANS, the objective is to minimize the sum of squares of the distances instead. It is easy to design an algorithm running in time max_{k ≤ n} {n choose k} n^𝒪(1) = 𝒪^*(2ⁿ) (here, 𝒪^*(⋅) notation hides polynomial factors in n). In this paper we design first non-trivial exact algorithms for these problems. In particular, we obtain an 𝒪^*((1.89)ⁿ) time exact algorithm for k-MEDIAN that works for any value of k. Our algorithm is quite general in that it does not use any properties of the underlying (metric) space - it does not even require the distances to satisfy the triangle inequality. In particular, the same algorithm also works for k-Means. We complement this result by showing that the running time of our algorithm is asymptotically optimal, up to the base of the exponent. That is, unless the Exponential Time Hypothesis fails, there is no algorithm for these problems running in time 2^o(n)⋅n^𝒪(1). Finally, we consider the "facility location" or "supplier" versions of these clustering problems, where, in addition to the set X we are additionally given a set of m candidate centers (or facilities) F, and objective is to find a subset of k centers from F. The goal is still to minimize the k-Median/k-Means/k-Center objective. For these versions we give a 𝒪(2ⁿ (mn)^𝒪(1)) time algorithms using subset convolution. We complement this result by showing that, under the Set Cover Conjecture, the "supplier" versions of these problems do not admit an exact algorithm running in time 2^{(1-ε) n} (mn)^𝒪(1).

We study the minimum-cost metric perfect matching problem under online i.i.d arrivals. We are given a fixed metric with a server at each of the points, and then requests arrive online, each drawn independently from a known probability distribution over the points. Each request has to be matched to a free server, with cost equal to the distance. The goal is to minimize the expected total cost of the matching. Such stochastic arrival models have been widely studied for the maximization variants of the online matching problem; however, the only known result for the minimization problem is a tight $O(\log n)$-competitiveness for the random-order arrival model. This is in contrast with the adversarial model, where an optimal competitive ratio of $O(\log n)$ has long been conjectured and remains a tantalizing open question. In this paper, we show improved results in the i.i.d arrival model. We show how the i.i.d model can be used to give substantially better algorithms: our main result is an $O((\log \log \log n)^2)$-competitive algorithm in this model. Along the way we give a $9$-competitive algorithm for the line and tree metrics. Both results imply a strict separation between the i.i.d model and the adversarial and random order models, both for general metrics and these much-studied metrics.

Friedman and Linial [8] introduced the convex body chasing problem to explore the interplay between geometry and competitive ratio in metrical task systems. In convex body chasing, at each time step t ∊ ℕ, the online algorithm receives a request in the form of a convex body Kt ⊂ ℝ and must output a point xt ∊ Kt. The goal is to minimize the total movement between consecutive output points, where the distance is measured in some given norm. This problem is still far from being understood. Recently Bansal et al. [4] gave an 6d(d!)2-competitive algorithm for the nested version, where each convex body is contained within the previous one. We propose a different strategy which is O(d log d)-competitive algorithm for this nested convex body chasing problem. Our algorithm works for any norm. This result is almost tight, given an Ω(d) lower bound for the ℓ∞ norm [8].

We consider the k-server problem on trees and HSTs. We give an algorithm based on Bregman projections. This algorithm has a competitive ratios that match some of the recent results given by Bubeck et al. (STOC 2018), whose algorithm was based on mirror-descent-based continuous dynamics prescribed via a differential inclusion.

Motivated by applications in cloud computing, we study the classical online caching problem for a cache of variable size, where the algorithm pays a maintenance cost that monotonically increases with cache size. This captures not only the classical setting of a fixed cache size, which corresponds to a maintenance cost of 0 for a cache of size at most k and ∞ otherwise, but also other natural settings in the context of cloud computing such as a concave rental cost on cache size. We call this the elastic caching problem. Our results are: (a) a randomized algorithm with a competitive ratio of O(log n) for maintenance cost that is an arbitrary function of cache size, (b) a deterministic algorithm with a competitive ratio of 2 for concave, or more generally submodular maintenance costs, (c) a deterministic n-competitive algorithm when the cost function is any monotone non-negative set function, and (d) a randomized constant-factor approximation algorithm for the offline version of the problem. Our algorithms are based on a configuration LP formulation of the problem, for which our main technical contribution is to maintain online a feasible fractional solution that can be converted to an integer solution using existing rounding techniques.

We study the problem of deleting the smallest set S of vertices (resp. edges) from a given graph G such that the induced subgraph (resp. subgraph) G\S belongs to some class ℋ. We consider the case where graphs in ℋ have treewidth bounded by t, and give a general framework to obtain approximation algorithms for both vertex and edge-deletion settings from approximation algorithms for certain natural graph partitioning problems called k-Subset Vertex Separator and k-Subset Edge Separator, respectively. For the vertex deletion setting, our framework combined with the current best result for k-Subset Vertex Separator, improves approximation ratios for basic problems such as k-Treewidth Vertex Deletion and Planar-ℱ Vertex Deletion. Our algorithms are simpler than previous works and give the first deterministic and uniform approximation algorithms under the natural parameterization. For the edge deletion setting, we give improved approximation algorithms for k-Subset Edge Separator combining ideas from LP relaxations and important separators. We present their applications in bounded-degree graphs, and also give an APX-hardness result for the edge deletion problems.

Given an edge-weighted graph, how many minimum k-cuts can it have? This is a fundamental question in the intersection of algorithms, extremal combinatorics, and graph theory. It is particularly interesting in that the best known bounds are algorithmic: they stem from algorithms that compute the minimum k-cut.

We prove that any graph excluding $K_r$ as a minor can be partitioned into clusters of diameter at most $\Delta$ while removing at most $O(r/\Delta)$ fraction of the edges. This improves over the results of Fakcharoenphol and Talwar, who, building on the work of Klein, Plotkin, and Rao, gave a partitioning that required removing $O(r^2/\Delta)$ fraction of the edges. Our result is obtained by a new approach that relates the topological properties (excluding a minor) of a graph to its geometric properties (the induced shortest path metric). Specifically, we show that techniques used by Andreae in his investigation of the cops and robbers game on graphs excluding a fixed minor can be used to construct padded decompositions of the metrics induced by such graphs. In particular, we get probabilistic partitions with padding parameter $O(r)$ and strong-diameter partitions with padding parameter $O(r^2)$ for $K_r$-minor-free graphs, $O(k)$ for treewidth-$k$ graphs, and $O(\log g)$ for graphs with (Euler) genus $g$.

We present several approximation algorithms for the problem of embedding metric spaces into a line, and into the 2-dimensional plane. Among other results, we give an $O(\sqrt{n})$-approximation algorithm for the problem of finding a line embedding of a metric induced by a given unweighted graph, that minimizes the (standard) multiplicative distortion. We give an improved $\tilde{O}(n^{1/3})$ approximation for the case of metrics induced by unweighted trees.

In the k-cut problem, we are given an edge-weighted graph G and an integer k, and have to remove a set of edges with minimum total weight so that G has at least k connected components. The current best algorithms are an O(n(2-o(1))k) randomized algorithm due to Karger and Stein, and an Õ(n2k) deterministic algorithm due to Thorup. Moreover, several 2-approximation algorithms are known for the problem (due to Saran and Vazirani, Naor and Rabani, and Ravi and Sinha). It has remained an open problem to (a) improve the runtime of exact algorithms, and (b) to get better approximation algorithms. In this paper we show an O(kO(k)n(2Ω/3 + o(1))k)-time algorithm for k-cut. Moreover, we show an (1+ε)-approximation algorithm that runs in time O((k/ε)O(k)nk + O(1)), and a 1.81-approximation in fixed-parameter time O(2O(k(2))poly(n)).

We study the classic bin packing problem in a fully-dynamic setting, where new items can arrive and old items may depart. We want algorithms with low asymptotic competitive ratio while repacking items sparingly between updates. Formally, each item i has a movement cost c_i >= 0, and we want to use alpha * OPT bins and incur a movement cost gamma * c_i, either in the worst case, or in an amortized sense, for alpha, gamma as small as possible. We call gamma the recourse of the algorithm. This is motivated by cloud storage applications, where fully-dynamic bin packing models the problem of data backup to minimize the number of disks used, as well as communication incurred in moving file backups between disks. Since the set of files changes over time, we could recompute a solution periodically from scratch, but this would give a high number of disk rewrites, incurring a high energy cost and possible wear and tear of the disks. In this work, we present optimal tradeoffs between number of bins used and number of items repacked, as well as natural extensions of the latter measure.

We consider the online problem of minimizing weighted flow-time on unrelated machines. Although much is known about this problem in the resource-augmentation setting, these results assume that jobs can be preempted. We give the first constant-competitive algorithm for the non-preemptive setting in the rejection model. In this rejection model, we are allowed to reject an epsilon-fraction of the total weight of jobs, and compare the resulting flow-time to that of the offline optimum which is required to schedule all jobs. This is arguably the weakest assumption in which such a result is known for weighted flow-time on unrelated machines. While our algorithms are simple, we need a delicate argument to bound the flow-time. Indeed, we use the dual-fitting framework, with considerable more machinery to certify that the cost of our algorithm is within a constant of the optimum while only a small fraction of the jobs are rejected.

Suppose a set of requests arrives online: each request gives some value $v_i$ if accepted, but requires using some amount of each of $d$ resources. Our cost is a convex function of the vector of total utilization of these $d$ resources. Which requests should be accept to maximize our profit, i.e., the sum of values of the accepted demands, minus the convex cost? We consider this problem in the random-order a.k.a. secretary model, and show an $O(d)$-competitive algorithm for the case where the convex cost function is also supermodular. If the set of accepted demands must also be independent in a given matroid, we give an $O(d^3 α)$-competitive algorithm for the supermodular case, and an improved $O(d^2α)$ if the convex cost function is also separable. Here $α$ is the competitive ratio of the best algorithm for the submodular secretary problem. These extend and improve previous results known for this problem. Our techniques are simple but use powerful ideas from convex duality, which give clean interpretations of existing work, and allow us to give the extensions and improvements.

In the Steiner Forest problem, we are given a graph and a collection of source-sink pairs, and the goal is to find a subgraph of minimum total length such that all pairs are connected. The problem is APX-Hard and can be 2-approximated by, e.g., the elegant primal-dual algorithm of Agrawal, Klein, and Ravi from 1995. We give a local-search-based constant-factor approximation for the problem. Local search brings in new techniques to an area that has for long not seen any improvements and might be a step towards a combinatorial algorithm for the more general survivable network design problem. Moreover, local search was an essential tool to tackle the dynamic MST/Steiner Tree problem, whereas dynamic Steiner Forest is still wide open. It is easy to see that any constant factor local search algorithm requires steps that add/drop many edges together. We propose natural local moves which, at each step, either (a) add a shortest path in the current graph and then drop a bunch of inessential edges, or (b) add a set of edges to the current solution. This second type of moves is motivated by the potential function we use to measure progress, combining the cost of the solution with a penalty for each connected component. Our carefully-chosen local moves and potential function work in tandem to eliminate bad local minima that arise when using more traditional local moves. Our analysis first considers the case where the local optimum is a single tree, and shows optimality w.r.t. moves that add a single edge (and drop a set of edges) is enough to bound the locality gap. For the general case, we show how to "project" the optimal solution onto the different trees of the local optimum without incurring too much cost (and this argument uses optimality w.r.t. both kinds of moves), followed by a tree-by-tree argument. We hope both the potential function, and our analysis techniques will be useful to develop and analyze local-search algorithms in other contexts.

We consider the problem of makespan minimization: i.e., scheduling jobs on machines to minimize the maximum load. For the deterministic case, good approximations are known even when the machines are unrelated. However, the problem is not well-understood when there is uncertainty in the job sizes. In our setting the job sizes are stochastic, i.e., the size of a job j on machine i is a random variable Xij, whose distribution is known. (Sizes of different jobs are independent of each other.) The goal is to find a fixed assignment of jobs to machines, to minimize the expected makespan—i.e., the expected value of the maximum load over the m machines. For the identical machines special case when the size of a job is the same across all machines, a constant-factor approximation algorithm has long been known. However, the problem has remained open even for the next-harder related machines case. Our main result is a constant-factor approximation for the most general case of unrelated machines. The main technical challenge we overcome is obtaining an efficiently computable lower bound for the optimal solution. We give an exponential-sized LP that we argue gives a strong lower bound. Then we show how to round any fractional solution to satisfy only a small subset of the constraints, which are enough to bound the expected makespan of our solution. We then consider two generalizations. The first is the budgeted makespan minimization problem, where the goal is to minimize the makespan subject to scheduling any subset of jobs whose reward is at least some target reward R. We extend our above result to a constant-factor approximation here using polyhedral properties of the bipartite matching polytope. The second problem is the q-norm minimization problem, where we want to minimize the expected ℓq-norm of the load vectors. Here we give an O(q/ log q)-approximation algorithm using a reduction to the deterministic q-norm problem with side constraints.

In the k-CuT problem, we are given an edge-weighted graph G and an integer k, and have to remove a set of edges with minimum total weight so that G has at least k connected components. Prior work on this problem gives, for all h ∊ [2, k], a (2 – h/k)-approximation algorithm for k-cut that runs in time nO(h). Hence to get a (2 – ε)-approximation algorithm for some absolute constant ε, the best runtime using prior techniques is nO(kε). Moreover, it was recently shown that getting a (2 – ε)-approximation for general k is NP-hard, assuming the Small Set Expansion Hypothesis. If we use the size of the cut as the parameter, an FPT algorithm to find the exact k-Cut is known, but solving the k-CuT problem exactly is W[1]-hard if we parameterize only by the natural parameter of k. An immediate question is: can we approximate k-Cut better in FPT-time, using k as the parameter? We answer this question positively. We show that for some absolute constant ε > 0, there exists a (– ε)-approximation algorithm that runs in time 2O(κ6) · Õ(n4). This is the first FPT algorithm that is parameterized only by k and strictly improves the 2-approximation.

We study the problem of embedding weighted graphs of pathwidth k into ℓp spaces. Our main result is an O(kmin{1p,12})-distortion embedding. For p=1, this is a super-exponential improvement over the best previous bound of Lee and Sidiropoulos. Our distortion bound is asymptotically tight for any fixed p >1.

We consider the maximum independent set problem on sparse graphs with maximum degree $d$. We show that the Lovász $\vartheta$-function based semidefinite program (SDP) has an integrality gap of $ \widetilde{O}(d/\log^{3/2} d)$, improving on the previous best result of $\widetilde{O}(d/\log d)$. This improvement is based on a new Ramsey-theoretic bound on the independence number of $K_r$-free graphs for large values of $r$. We also show that for stronger SDPs, namely, those obtained using poly$\log(d)$ levels of the ${SA}^{+}$ semidefinite hierarchy, the integrality gap reduces to $\widetilde{O}(d/\log^2 d)$. This matches the best unique-games-based hardness result up to lower-order poly$(\log\log d)$ factors. Finally, we give an algorithmic version of this ${SA}^{+}$-based integrality gap result, albeit using $d$ levels of ${SA}^{+}$, via a coloring algorithm of Johansson.

We consider the stochastic unsplittable flow problem: given a graph with edge-capacities, and source-sink pairs with each pair having a size and a value, the goal is to route the pairs unsplittably while respecting edge capacities to maximize the total value of the routed pairs. However, the size of each pair is a random variable and is revealed only after we decide to route that pair. Which pairs should we route, along which paths, and in what order so as to maximize the expected value? We present results for several cases of the problem under the no-bottleneck assumption. We show a logarithmic approximation algorithm for the single-sink problem on general graphs, considerably improving on the prior results of Chawla and Roughgarden which worked for planar graphs. We present an approximation to the stochastic unsplittable flow problem on directed acyclic graphs, within less than a logarithmic factor of the best known approximation in the non-stochastic setting. We present a non-adaptive strategy on trees that is within a constant factor of the best adaptive strategy, asymptotically matching the best results for the non-stochastic unsplittable flow problem on trees. Finally, we give results for the stochastic unsplittable flow problem on general graphs. Our techniques include using edge-confluent flows for the single-sink problem in order to control the interaction between flow-paths, and a reduction from general scheduling policies to "safe" ones (i.e., those guaranteeing no capacity violations), which may be of broader interest.

We study the combinatorial pure exploration problem \textscBest-Set in a stochastic multi-armed bandit game. In an \textscBest-Set instance, we are given $n$ stochastic arms with unknown reward distributions, as well as a family $\mathcal{F}$ of feasible subsets over the arms. Let the weight of an arm be the mean of its reward distribution. Our goal is to identify the feasible subset in $\mathcal{F}$ with the maximum total weight, using as few samples as possible. The problem generalizes the classical best arm identification problem and the top-$k$ arm identification problem, both of which have attracted significant attention in recent years. We provide a novel \textitinstance-wise lower bound for the sample complexity of the problem, as well as a nontrivial sampling algorithm, matching the lower bound up to a factor of $\ln|\mathcal{F}|$. For an important class of combinatorial families (including spanning trees, matchings, and path constraints), we also provide polynomial time implementation of the sampling algorithm, using the equivalence of separation and optimization for convex program, and the notion of approximate Pareto curves in multi-objective optimization (note that $|\mathcal{F}|$ can be exponential in $n$). We also show that the $\ln|\mathcal{F}|$ factor is inevitable in general, through a nontrivial lower bound construction utilizing a combinatorial structure resembling the Nisan-Wigderson design. Our results significantly improve several previous results for several important combinatorial constraints, and provide a tighter understanding of the general \textscBest-Set problem. We further introduce an even more general problem, formulated in geometric terms. We are given $n$ Gaussian arms with unknown means and unit variance. Consider the $n$-dimensional Euclidean space $\mathbb{R}^n$, and a collection $\mathcal{O}$ of disjoint subsets. Our goal is to determine the subset in $\mathcal{O}$ that contains the mean profile (which is the $n$-dimensional vector of the means), using as few samples as possible. The problem generalizes most pure exploration bandit problems studied in the literature. We provide the first nearly optimal sample complexity upper and lower bounds for the problem.

The online (uniform) buy-at-bulk network design problem asks us to design a network, where the edge-costs exhibit economy-of-scale. Previous approaches to this problem used tree-embeddings, giving us randomized algorithms. Moreover, the optimal results with a logarithmic competitive ratio requires the metric on which the network is being built to be known up-front; the competitive ratios then depend on the size of this metric (which could be much larger than the number of terminals that arrive). We consider the buy-at-bulk problem in the least restrictive model where the metric is not known in advance, but revealed in parts along with the demand points seeking connectivity arriving online. For the single sink buy-at-bulk problem, we give a deterministic online algorithm with competitive ratio that is logarithmic in k, the number of terminals that have arrived, matching the lower bound known even for the online Steiner tree problem. In the oblivious case when the buy-at-bulk function used to compute the edge-costs of the network is not known in advance (but is the same across all edges), we give a deterministic algorithm with competitive ratio polylogarithmic in k, the number of terminals. At the heart of our algorithms are optimal constructions for online Light Approximate Shortest-path Trees (LASTs) and spanners, and their variants. We give constructions that have optimal trade-offs in terms of cost and stretch. We also define and give constructions for a new notion of LASTs where the set of roots (in addition to the points) expands over time. We expect these techniques will find applications in other online network-design problems.

Suppose we are given a submodular function f over a set of elements, and we want to maximize its value subject to certain constraints. Good approximation algorithms are known for such problems under both monotone and non-monotone submodular functions. We consider these problems in a stochastic setting, where elements are not all active and we only get value from active elements. Each element e is active independently with some known probability pe, but we don't know the element's status a priori: we find it out only when we probe the element e. Moreover, the sequence of elements we probe must satisfy a given prefix-closed constraint, e.g., matroid, orienteering, deadline, precedence, or any downward-closed constraint. In this paper we study the gap between adaptive and non-adaptive strategies for f being a submodular or a fractionally subadditive (XOS) function. If this gap is small, we can focus on finding good non-adaptive strategies instead, which are easier to find as well as to represent. We show that the adaptivity gap is a constant for monotone and non-monotone submodular functions, and logarithmic for XOS functions of small width. These bounds are nearly tight. Our techniques show new ways of arguing about the optimal adaptive decision tree for stochastic optimization problems.

In this paper, we give new results for the set cover problem in the fully dynamic model. In this model, the set of "active" elements to be covered changes over time. The goal is to maintain a near-optimal solution for the currently active elements, while making few changes in each timestep. This model is popular in both dynamic and online algorithms: in the former, the goal is to minimize the update time of the solution, while in the latter, the recourse (number of changes) is bounded. We present generic techniques for the dynamic set cover problem inspired by the classic greedy and primal-dual offline algorithms for set cover. The former leads to a competitive ratio of O(lognt), where nt is the number of currently active elements at timestep t, while the latter yields competitive ratios dependent on ft, the maximum number of sets that a currently active element belongs to. We demonstrate that these techniques are useful for obtaining tight results in both settings: update time bounds and limited recourse, exhibiting algorithmic techniques common to these two parallel threads of research.

We study the pure exploration problem subject to a matroid constraint (Best-Basis) in a stochastic multi-armed bandit game. In a Best-Basis instance, we are given n stochastic arms with unknown reward distributions, as well as a matroid \mathcalM over the arms. Let the weight of an arm be the mean of its reward distribution. Our goal is to identify a basis of \mathcalM with the maximum total weight, using as few samples as possible. The problem is a significant generalization of the best arm identification problem and the top-k arm identification problem, which have attracted significant attentions in recent years. We study both the exact and PAC versions of Best-Basis, and provide algorithms with nearly-optimal sample complexities for these versions. Our results generalize and/or improve on several previous results for the top-k arm identification problem and the combinatorial pure exploration problem when the combinatorial constraint is a matroid.

We present online algorithms for covering and packing problems with (non-linear) convex objectives. The convex covering problem is defined as: minxϵR+nf(x) s.t. Ax ≥ 1, where f:R+n→ R+is a monotone convex function, and A is an m×n matrix with non-negative entries. In the online version, a new row of the constraint matrix, representing a new covering constraint, is revealed in each step and the algorithm is required to maintain a feasible and monotonically non-decreasing assignment x over time. We also consider a convex packing problem defined as: maxyϵR+mΣj=1myj - g(ATy), where g:R+n→R+is a monotone convex function. In the online version, each variable yj arrives online and the algorithm must decide the value of yj on its arrival. This represents the Fenchel dual of the convex covering program, when g is the convex conjugate of f. We use a primal-dual approach to give online algorithms for these generic problems, and use them to simplify, unify, and improve upon previous results for several applications.

In the aversion k-clustering problem, given a metric space, we want to cluster the points into k clusters. The cost incurred by each point is the distance to the furthest point in its cluster, and the cost of the clustering is the sum of all these per-point-costs. This problem is motivated by questions in generating automatic abstractions of extensive-form games. We reduce this problem to a "local" k-median problem where each facility has a prescribed radius and can only connect to clients within that radius. Our main results is a constant-factor approximation algorithm for the aversion k-clustering problem via the local k-median problem. We use a primal-dual approach; our technical contribution is a non-local rounding step which we feel is of broader interest.

A stochastic probing problem consists of a set of elements whose values are independent random variables. The algorithm knows the distributions of these variables, but not the actual outcomes. The only way to learn the actual outcomes is to probe these elements. However, there are constraints on which set of elements may be probed. (E.g., we may have to travel in some metric to probe elements but have limited time.) These constraints are called outer constraints. We want to develop an algorithm that picks some set of elements to maximize the (expected) value, subject to the picked subset of elements satisfying some other set of constraints, called the inner constraints. In the past, probing problems were studied for the case when both inner and outer constraints were intersections of matroids; these modeled kidney matching and Bayesian auctions applications. One limitation of past work was their reliance on linear-programming-like techniques, which made going beyond matroid-like structures difficult. In this work, we give a very general adaptivity gap result that holds for all prefix-closed outer constraints, as long as the inner constraints are intersections of matroids. The adaptivity gap is O(log n) for any constant number of inner matroid constraints. The prefix-closedness captures most “reasonable” outer constraints, like orienteering, connectivity, and precedence. Based on this we obtain the first approximation algorithms for a number of stochastic probing problems, which have applications, e.g., to path-planning and precedence-constrained scheduling.

In the online Steiner tree problem, a sequence of points is revealed one-by-one: when a point arrives, we only have time to add a single edge connecting this point to the previous ones, and we want to minimize the total length of edges added. Here, a tight bound has been known for two decades: the greedy algorithm maintains a tree whose cost is $O(\log n)$ times the Steiner tree cost, and this is the best possible. But suppose, in addition to the new edge we add, we have time to change a single edge from the previous set of edges: can we do much better? Can we, e.g., maintain a tree that is constant-competitive? We answer this question in the affirmative. We give a primal-dual algorithm that makes only a single swap per step (in addition to adding the edge connecting the new point to the previous ones), and such that the tree's cost is only a constant times the optimal cost. Our dual-based analysis is quite different from previous primal-only analyses. In particular, we give a correspondence between radii of dual balls and lengths of tree edges; since dual balls are associated with points and hence do not move around (in contrast to edges), we can closely monitor the edge lengths based on the dual radii. Showing that these dual radii cannot change too rapidly is the technical heart of the paper and allows us to give a hard bound on the number of swaps per arrival, while maintaining a constant-competitive tree at all times. Previous results for this problem gave an algorithm that performed an amortized constant number of swaps: for each $n$, the number of swaps in the first $n$ steps was $O(n)$. We also give a simpler tight analysis for this amortized case.

We consider the hub label optimization problem, which arises in designing fast preprocessing-based shortest-path algorithms. We give O (log n )-approximation algorithms for the objectives of minimizing the maximum label size (ℓ ∞ -norm) and simultaneously minimizing a constant number of ℓ p -norms. Prior to this, an O (log n )-approximation algorithm was known [Cohen et al. 2003] only for minimizing the total label size (ℓ 1 -norm).

We study the problem of routing in doubling metrics and show how to perform hierarchical routing in such metrics with small stretch and compact routing tables (i.e., with a small amount of routing information stored at each vertex). We say that a metric ( X , d ) has doubling dimension dim(α balls of half its radius. (A doubling metric is one whose doubling dimension dim(G . We show how to perform (1 + τ)-stretch routing on such a metric for any 0 < τ ≤ 1 with routing tables of size at most (α/τ) O (α) log Δlog δ bits with only (α/τ) O (α) log Δ entries , where Δ is the diameter of the graph, and δ is the maximum degree of the graph G ; hence, the number of routing table entries is just τ − O (1) log Δ for doubling metrics. These results extend and improve on those of Talwar (2004). We also give better constructions of sparse spanners for doubling metrics than those obtained from the routing tables earlier; for τ > 0, we give algorithms to construct (1 + τ)-stretch spanners for a metric ( X , d ) with maximum degree at most (2 + 1/τ) O(dim(X)) , matching the results of Das et al. for Euclidean metrics.

Consider the following problem: given a set system ( U , Ω) and an edge-weighted graph G = ( U , E ) on the same universe U , find the set A ∈ Ω such that the Steiner tree cost with terminals A is as large as possible—“which set in Ω is the most difficult to connect up?” This is an example of a max-min problem : find the set A ∈ Ω such that the value of some minimization (covering) problem is as large as possible. In this article, we show that for certain covering problems that admit good deterministic online algorithms, we can give good algorithms for max-min optimization when the set system Ω is given by a p -system or knapsack constraints or both. This result is similar to results for constrained maximization of submodular functions. Although many natural covering problems are not even approximately submodular, we show that one can use properties of the online algorithm as a surrogate for submodularity. Moreover, we give stronger connections between max-min optimization and two-stage robust optimization, and hence give improved algorithms for robust versions of various covering problems, for cases where the uncertainty sets are given by p -systems and knapsack constraints.

We consider a natural online optimization problem set on the real line. The state of the online algorithm at each integer time is a location on the real line. At each integer time, a convex function arrives online. In response, the online algorithm picks a new location. The cost paid by the online algorithm for this response is the distance moved plus the value of the function at the final destination. The objective is then to minimize the aggregate cost over all time. The motivating application is rightsizing power-proportional data centers. We give a 2-competitive algorithm for this problem. We also give a 3-competitive memoryless algorithm, and show that this is the best competitive ratio achievable by a deterministic memoryless algorithm. Finally we show that this online problem is strictly harder than the standard ski rental problem.

We consider the maximum independent set problem on graphs with maximum degree d. We show that the integrality gap of the Lovasz Theta function-based SDP has an integrality gap of O~(d/log3/2 d). This improves on the previous best result of O~(d/log d), and narrows the gap of this basic SDP to the integrality gap of O~(d/log2 d) recently shown for stronger SDPs, namely those obtained using poly log(d) levels of the SA+ semidefinite hierarchy. The improvement comes from an improved Ramsey-theoretic bound on the independence number of Kr-free graphs for large values of r.

In the Steiner Forest problem, we are given terminal pairs si, ti, and need to find the cheapest subgraph which connects each of the terminal pairs together. In 1991, Agrawal, Klein, and Ravi gave a primal-dual constant-factor approximation algorithm for this problem. Until this work, the only constant-factor approximations we know are via linear programming relaxations.

In the online Steiner tree problem, the input is a set of vertices that appear one-by-one, and we have to maintain a Steiner tree on the current set of vertices. The cost of the tree is the total length of edges in the tree, and we want this cost to be close to the cost of the optimal Steiner tree at all points in time. If we are allowed to only add edges, a tight bound of $Θ(\log n)$ on the competitiveness is known. Recently it was shown that if we can add one new edge and make one edge swap upon every vertex arrival, we can maintain a constant-competitive tree online. But what if the set of vertices sees both additions and deletions? Again, we would like to obtain a low-cost Steiner tree with as few edge changes as possible. The original paper of Imase and Waxman had also considered this model, and it gave a greedy algorithm that maintained a constant-competitive tree online, and made at most $O(n^{3/2})$ edge changes for the first $n$ requests. In this paper give the following two results. Our first result is an online algorithm that maintains a Steiner tree only under deletions: we start off with a set of vertices, and at each time one of the vertices is removed from this set: our Steiner tree no longer has to span this vertex. We give an algorithm that changes only a constant number of edges upon each request, and maintains a constant-competitive tree at all times. Our algorithm uses the primal-dual framework and a global charging argument to carefully make these constant number of changes. We then study the natural greedy algorithm proposed by Imase and Waxman that maintains a constant-competitive Steiner tree in the fully-dynamic model (where each request either adds or deletes a vertex). Our second result shows that this algorithm makes only a constant number of changes per request in an amortized sense.

Consider the following edge-orientation problem: edges of a graph appear online one-by-one and they to be directed—given an “orientation”. We want to ensure that the in-degree of each vertex remains low. (This is a simple case of scheduling unit-sized jobs on machines, where each job can only go on one of two machines.) If the edge-orientations we assign are irrevocable, we suffer a significant loss in quality due to online decision-making (as compared to the offline performance). For instance the best online competitive ratio achievable is Θ(log m) for even this toy problem. But what if the decisions are not irrevocable — what if we allow a limited number of reassignments? Can we do much better? We show that indeed we can. For instance, for edge-orientation, we can achieve a constant-competitive load while doing only a constant number of re-orientations per edge (in an amortized sense). For more substantial problems, our results are as follows: For online matching, where the left vertices arrive online and must be matched to the right vertices, we give an algorithm that reassigns the left vertices an (amortized) constant number of times, and maintains a constant factor to the optimal load on the right vertices. We extend this to restricted machine scheduling with arbitrary sized jobs and give an algorithm that maintains load which is O(log log mn) times the optimum, and reassigns each job only an (amortized) constant number of times. Consider a digraph with a single source, where sinks arrive online and want to send unit flow to the source. The goal is to minimize the congestion on the edges. Suppose there is an offline flow such that the total length of the flow paths is F*. We give an algorithm that reroutes flow along O(F*) edges and achieves a O(1)-approximation to the congestion.

We prove that any graph excluding Kr as a minor has can be partitioned into clusters of diameter at most Δ while removing at most O(r/Δ) fraction of the edges. This improves over the results of Fakcharoenphol and Talwar, who building on the work of Klein, Plotkin and Rao gave a partitioning that required to remove O(r2/Δ) fraction of the edges. Our result is obtained by a new approach that relates the topological properties (excluding a minor) of a graph to its geometric properties (the induced shortest path metric). Specifically, we show that techniques used by Andreae in his investigation of the cops and robbers game on graphs excluding a fixed minor, can be used to construct padded decompositions of the metrics induced by such graphs. In particular, we get probabilistic partitions with padding parameter O(r) and strong-diameter partitions with padding parameter O(r2) for Kr-free graphs, O(k) for treewidth-k graphs, and O(log g) for graphs with genus g.

Given a capacitated graph $G = (V,E)$ and a set of terminals $K \subseteq V$, how should we produce a graph $H$ only on the terminals $K$ so that every (multicommodity) flow between the terminals in $G$ could be supported in $H$ with low congestion, and vice versa? (Such a graph $H$ is called a flow sparsifier for $G$.) What if we want $H$ to be a “simple” graph? What if we allow $H$ to be a convex combination of simple graphs? Improving on results of Moitra [Proceedings of the 50th IEEE Symposium on Foundations of Computer Science, IEEE Computer Society, Los Alamitos, CA, 2009, pp. 3--12] and Leighton and Moitra [Proceedings of the 42nd ACM Symposium on Theory of Computing, ACM, New York, 2010, pp. 47--56], we give efficient algorithms for constructing (a) a flow sparsifier $H$ that maintains congestion up to a factor of $O(\frac{\log k}{\log \log k})$, where $k = |K|$; (b) a convex combination of trees over the terminals $K$ that maintains congestion up to a factor of $O(\log k)$; (c) for a planar graph $G$, a convex combination of planar graphs that maintains congestion up to a constant factor. This requires us to give a new algorithm for the 0-extension problem, the first one in which the preimages of each terminal are connected in $G$. Moreover, this result extends to minor-closed families of graphs. Our bounds immediately imply improved approximation guarantees for several terminal-based cut and ordering problems.

Consider the following problem of serving impatient users: we are given a set of customers we would like to serve. We can serve at most one customer in each time step (getting value vi for serving customer i). At the end of each time step, each as-yet-unserved customer i leaves the system independently with probability qi, never to return. What strategy should we use to serve customers to maximize the expected value collected?

Kidney exchanges allow incompatible donor-patient pairs to swap kidneys, but each donation must pass three tests: blood, tissue, and crossmatch. In practice a matching is computed based on the first two tests, and then a single crossmatch test is performed for each matched patient. However, if two crossmatches could be performed per patient, in principle significantly more successful exchanges could take place. In this paper, we ask: If we were allowed to perform two crossmatches per patient, could we harness this additional power optimally and efficiently? Our main result is a polynomial time algorithm for this problem that almost surely computes optimal --- up to lower order terms --- solutions on random large kidney exchange instances.

In the online Steiner tree problem, a sequence of points is revealed one-by-one: when a point arrives, we only have time to add a single edge connecting this point to the previous ones, and we want to minimize the total length of edges added. Here, a tight bound has been known for two decades: the greedy algorithm maintains a tree whose cost is O(log n) times the Steiner tree cost, and this is best possible. But suppose, in addition to the new edge we add, we have time to change a single edge from the previous set of edges: can we do much better? Can we, e.g., maintain a tree that is constant-competitive?

We give a 2-approximation algorithm for the non-uniform Sparsest Cut problem that runs in time nO(k), where k is the treewidth of the graph. This improves on the previous 22k-approximation in time poly(n) 2O(k) due to Chlamtac et al. [18].

A common approach to clustering data is to view data objects as points in a metric space, and then to optimize a natural distance-based objective such as the k -median, k -means, or min-sum score. For applications such as clustering proteins by function or clustering images by subject, the implicit hope in taking this approach is that the optimal solution for the chosen objective will closely match the desired “target” clustering (e.g., a correct clustering of proteins by function or of images by who is in them). However, most distance-based objectives, including those mentioned here, are NP-hard to optimize. So, this assumption by itself is not sufficient, assuming P ≠ NP, to achieve clusterings of low-error via polynomial time algorithms. In this article, we show that we can bypass this barrier if we slightly extend this assumption to ask that for some small constant c , not only the optimal solution, but also all c -approximations to the optimal solution, differ from the target on at most some ϵ fraction of points—we call this (c,ϵ)-approximation-stability . We show that under this condition, it is possible to efficiently obtain low-error clusterings even if the property holds only for values c for which the objective is known to be NP-hard to approximate. Specifically, for any constant c > 1, (c,ϵ) -approximation-stability of k -median or k -means objectives can be used to efficiently produce a clustering of error O (ϵ) with respect to the target clustering, as can stability of the min-sum objective if the target clusters are sufficiently large. Thus, we can perform nearly as well in terms of agreement with the target clustering as if we could approximate these objectives to this NP-hard value.

Given a universe $U$ of $n$ elements and a weighted collection $\mathscr{S}$ of $m$ subsets of $U$, the universal set cover problem is to a priori map each element $u \in U$ to a set $S(u) \in \mathscr{S}$ containing $u$ such that any set $X{\subseteq U}$ is covered by $S(X)=\cup_{u\in XS(u)$. The aim is to find a mapping such that the cost of $S(X)$ is as close as possible to the optimal set cover cost for $X$. (Such problems are also called oblivious or a priori optimization problems.) Unfortunately, for every universal mapping, the cost of $S(X)$ can be $\Omega(\sqrt{n})$ times larger than optimal if the set $X$ is adversarially chosen. In this paper we study the performance on average, when $X$ is a set of randomly chosen elements from the universe: we show how to efficiently find a universal map whose expected cost is $O(\log mn)$ times the expected optimal cost. In fact, we give a slightly improved analysis and show that this is the best possible. We generalize these ideas to weighted set cover and show similar guarantees to (nonmetric) facility location, where we have to balance the facility opening cost with the cost of connecting clients to the facilities. We show applications of our results to universal multicut and disc-covering problems and show how all these universal mappings give us algorithms for the stochastic online variants of the problems with the same competitive factors.

Suppose we would like to know all answers to a set of statistical queries $C$ on a data set up to small error, but we can access the data itself only by using statistical queries. A trivial solution is to exhaustively ask all queries in $C$. In this paper, we investigate how and when we can do better than this naïve approach. We show that the number of statistical queries necessary and sufficient for this task is---up to polynomial factors---equal to the agnostic learning complexity of $C$ in Kearns' statistical query (SQ) model. This gives a complete answer to the question when run-time is not a concern. We then show that the problem can be solved efficiently (allowing arbitrary error on a small fraction of queries) whenever the answers to $C$ can be described by a submodular function. This includes many natural concept classes, such as graph cuts and Boolean disjunctions and conjunctions. These results are interesting not only from a learning theoretic point of view, but also from the perspective of privacy-preserving data analysis. In this context, our second result leads to an algorithm that efficiently releases differentially private answers to all Boolean conjunctions with 1% average error. This presents significant progress on a key open problem in privacy-preserving data analysis. Our first result, on the other hand, gives unconditional lower bounds on any differentially private algorithm that admits a (potentially non--privacy-preserving) implementation using only statistical queries. Not only our algorithms but also most known private algorithms can be implemented using only statistical queries and hence are constrained by these lower bounds. Our result therefore isolates the complexity of agnostic learning in the SQ model as a new barrier in the design of differentially private algorithms.

We consider preemptive online scheduling algorithms to minimize the total weighted/unweighted flow time plus energy for speed-scalable heterogeneous multiprocessors. We show that the well-known priority scheduling algorithms Highest Density First, Weighted Shortest Elapsed Time First, and Weighted Late Arrival Processor Sharing, are not O(1)-speed O(1)-competitive for the objective of weighted flow even in the special case of fixed variable speed processors (aka the related machines setting). This illustrates that scheduling heterogeneous multiprocessors is a different, and algorithmically more challenging problem, than scheduling homogeneous multiprocessors. We then show that a variation of the non-clairvoyant algorithm Late Arrival Processor Sharing coupled with a non-obvious speed scaling algorithm is scalable for the objective of unweighted flow plus energy on speed-scalable multiprocessors. This is the first provably scalable non-clairvoyant algorithm on heterogeneous multiprocessors, even in the related machines setting, for the objective of total (unweighted) flow time.

In the Stochastic Orienteering problem, we are given a metric, where each node also has a job located there with some deterministic reward and a random size. (Think of the jobs as being chores one needs to run, and the sizes as the amount of time it takes to do the chore.) The goal is to adaptively decide which nodes to visit to maximize total expected reward, subject to the constraint that the total distance traveled plus the total size of jobs processed is at most a given budget of B. (I.e., we get reward for all those chores we finish by the end of the day). The (random) size of a job is not known until it is completely processed. Hence the problem combines aspects of both the stochastic knapsack problem with uncertain item sizes and the deterministic orienteering problem of using a limited travel time to maximize gathered rewards located at nodes. In this paper, we present a constant-factor approximation algorithm for the best non-adaptive policy for the Stochastic Orienteering problem. We also show a small adaptivity gap—i.e., the existence of a non-adaptive policy whose reward is at least an Ω(1/ log log B) fraction of the optimal expected reward—and hence we also get an O(log log B)-approximation algorithm for the adaptive problem. Finally we address the case when the node rewards are also random and could be correlated with the waiting time, and give a non-adaptive policy which is an O(log n log B)-approximation to the best adaptive policy on n-node metrics with budget B.

This paper presents parallel algorithms for embedding an arbitrary n-point metric space into a distribution of dominating trees with O(log n) expected stretch. Such embedding has proved useful in the design of many approximation algorithms in the sequential setting. We give a parallel algorithm that runs in O(n2 log n) work and O(log2 n) depth---these bounds are independent of Δ = (maxx,y d(x,y))/(minx≠ y d(x,y)), the ratio of the largest to smallest distance. Moreover, when Δ is exponentially bounded (Δ ≤ 2O(n)), our algorithm can be improved to O(n2) work and O(log2 n) depth. Using these results, we give an RNC O(log k)-approximation algorithm for k-median and an RNC O(log n)-approximation for buy-at-bulk network design. The k-median algorithm is the first RNC algorithm with non-trivial guarantees for arbitrary values of k, and the buy-at-bulk result is the first parallel algorithm for the problem.

The traveling salesman problem (TSP) is a canonical NP-complete problem which is proved by Trevisan [SIAM J. Comput., 30 (2000), pp. 475--485] to be MAX-SNP hard even on high-dimensional Euclidean metrics. To circumvent this hardness, researchers have been developing approximation schemes for „simpler” instances of the problem. For instance, the algorithms of Arora and of Talwar show how to approximate TSP on low-dimensional metrics (for different notions of metric dimension). However, a feature of most current notions of metric dimension is that they are „local”: the definitions require every local neighborhood to be well-behaved. In this paper, we define a global notion of dimension that generalizes the popular notion of doubling dimension, but still allows some small dense regions; e.g., it allows some metrics that contain cliques of size $\sqrt{n}$. Given a metric with global dimension $\dim_{C}$, we give a $(1+\varepsilon)$-approximation algorithm that runs in subexponential time, i.e., in $\exp(O(n^{\delta}\varepsilon^{-4\dim_{C}}))$-time for every constant $0<\delta<1$. As mentioned above, metrics with bounded $\dim_{C}$ may contain metrics of size $O(\sqrt{n})$ on which the TSP problem is hard to approximate to within $(1+\varepsilon)$. Hence, to do better than a running time of $\Omega(\exp\{\sqrt{n}\})$, our algorithms find $O(1)$-approximations to some portions of the tour, and $(1+\varepsilon)$-approximations for other portions, and stitch them together. Moreover, we show that such globally bounded metrics have spanners that preserve distances to arbitrary accuracy and have size $\Theta(n^{1.5})$.

Consider the edge-connectivity survivable network design problem (SNDP): given a graph $G = (V,E)$ with edge-costs, and edge-connectivity requirements $r_{ij} \in \mathbb{Z}_{\geq 0}$ for every pair of vertices $i, j \in V$, find an (approximately) minimum-cost network that provides the required connectivity. While this problem is known to admit good approximation algorithms in the offline case, no algorithms were known for this problem in the online setting. In this paper, we give a randomized $\tilde{O}(r_{\max} \log^3 n)$-competitive online algorithm for this edge-connectivity network design problem that runs in time $O(m^{r_{\max}})$, where $r_{\max} = \max_{ij} r_{ij}$. Our algorithms use the standard embeddings of graphs into random subtrees (i.e., into singly connected subgraphs) as an intermediate step to get algorithms for higher connectivity. As a consequence of using these random embeddings, our algorithms are competitive only against oblivious adversaries. Our results for the online problem give us approximation algorithms that admit strict cost-shares with the same strictness value. This, in turn, implies approximation algorithms for (a) the rent-or-buy version and (b) the (two-stage) stochastic version of the edge-connected network design problem with independent arrivals. If we are in the case when the underlying graph is complete and the edge-costs are metric (i.e., the triangle inequality is satisfied), we improve on our results to give an $O(\log n)$-competitive deterministic online algorithm for the rooted version of the problem, and constant-factor approximation algorithms for the rent-or-buy and stochastic variants of SNDP.

Combinatorial Auctions are a central problem in Algorithmic Mechanism Design: pricing and allocating goods to buyers with complex preferences in order to maximize some desired objective (e.g., social welfare, revenue, or profit). The problem has been well-studied in the case of limited supply (one copy of each item), and in the case of digital goods (the seller can produce additional copies at no cost). Yet in the case of resources -- oil, labor, computing cycles, etc. -- neither of these abstractions is just right: additional supplies of these resources can be found, but at increasing difficulty (marginal cost) as resources are depleted. In this work, we initiate the study of the algorithmic mechanism design problem of combinatorial pricing under increasing marginal cost. The goal is to sell these goods to buyers with unknown and arbitrary combinatorial valuation functions to maximize either the social welfare, or the seller's profit, specifically we focus on the setting of posted item prices with buyers arriving online. We give algorithms that achieve constant factor approximations for a class of natural cost functions - linear, low-degree polynomial, logarithmic - and that give logarithmic approximations for more general increasing marginal cost functions (along with a necessary additive loss). We show that these bounds are essentially best possible for these settings.

In the stochastic knapsack problem, we are given a knapsack of size B, and a set of items whose sizes and rewards are drawn from a known probability distribution. To know the actual size and reward we have to schedule the item-when it completes, we get to know these values. The goal is to schedule the items (possibly making adaptive decisions based on the sizes seen so far) to maximize the expected total reward of items which successfully pack into the knapsack. We know constant-factor approximations when (i) the rewards and sizes are independent, and (ii) we cannot prematurely cancel items after we schedule them. What if either or both assumptions are relaxed? Related stochastic packing problems are the multi-armed bandit (and budgeted learning) problems, here one is given several arms which evolve in a specified stochastic fashion with each pull, and the goal is to (adaptively) decide which arms to pull, in order to maximize the expected reward obtained after B pulls in total. Much recent work on this problem focuses on the case when the evolution of each arm follows a martingale, i.e., when the expected reward from one pull of an arm is the same as the reward at the current state. What if the rewards do not form a martingale? In this paper, we give O(1)-approximation algorithms for the stochastic knapsack problem with correlations and/or cancellations. Extending the ideas developed here, we give O(1)-approximations for MAB problems without the martingale assumption. Indeed, we can show that previously proposed linear programming relaxations for these problems have large integrality gaps. So we propose new time-indexed LP relaxations, using a decomposition and "gap-filling" approach, we convert these fractional solutions to distributions over strategies, and then use the LP values and the time ordering information from these strategies to devise randomized adaptive scheduling algorithms.

This paper presents the design and analysis of a near linear-work parallel algorithm for solving symmetric diagonally dominant (SDD) linear systems. On input an SDD n-by-n matrix A with m non-zero entries and a vector b, our algorithm computes a vector x such that Ax - A+b ≤ ε • A+b in O(m logO(1) n log 1/ε) work and O(m1/3+θ log 1/ε) depth for any fixed θ > 0.

Suppose we would like to know all answers to a set of statistical queries C on a data set up to small error, but we can only access the data itself using statistical queries. A trivial solution is to exhaustively ask all queries in C. Can we do any better? We show that the number of statistical queries necessary and sufficient for this task is---up to polynomial factors---equal to the agnostic learning complexity of C in Kearns' statistical query (SQ)model. This gives a complete answer to the question when running time is not a concern.

We consider two- and multistage versions of stochastic combinatorial optimization problems with recourse: in this framework, the instance for the combinatorial optimization problem is drawn from a known probability distribution $\pi$ and is only revealed to the algorithm over two (or multiple) stages. At each stage, on receiving some more information about the instance, the algorithm is allowed to build some partial solution. Since the costs of elements increase with each passing stage, there is a natural tension between waiting for later stages, to gain more information about the instance, and purchasing elements in earlier stages, to take advantages of lower costs. We provide approximation algorithms for stochastic combinatorial optimization problems (such as the Steiner tree problem, the Steiner network problem, and the vertex cover problem) by means of a simple sampling-based algorithm. In every stage, our algorithm samples the probability distribution of the requirements and constructs a partial solution to serve the resulting sample. We show that if one can construct cost-sharing functions associated with the algorithms used to construct these partial solutions, then this strategy results in provable approximation guarantees for the overall stochastic optimization problem. We also extend this approach to provide an approximation algorithm for the stochastic version of the uncapacitated facility location problem, a problem that does not fit into the simpler framework of our main model.

In the generalized connectivity problem, we are given an edge-weighted graph G = ( V , E ) and a collection D = {( S 1 , T 1 ), …, ( S k , T k )} of distinct demands each demand ( S i , T i ) is a pair of disjoint vertex subsets. We say that a subgraph F of G connects a demand ( S i , T i ) when it contains a path with one endpoint in S i and the other in T i . The goal is to identify a minimum weight subgraph that connects all demands in D . Alon et al. (SODA '04) introduced this problem to study online network formation settings and showed that it captures some well-studied problems such as Steiner forest, facility location with nonmetric costs, tree multicast, and group Steiner tree. Obtaining a nontrivial approximation ratio for generalized connectivity was left as an open problem. We describe the first poly-logarithmic approximation algorithm for generalized connectivity that has a performance guarantee of O (log 2 n log 2 k ). Here, n is the number of vertices in G and k is the number of demands. We also prove that the cut-covering relaxation of this problem has an O (log 3 n log 2 k ) integrality gap. Building upon the results for generalized connectivity, we obtain improved approximation algorithms for two problems that contain generalized connectivity as a special case. For the directed Steiner network problem, we obtain an O ( k 1/2 + ϵ ) approximation which improves on the currently best performance guarantee of Õ ( k 2/3 ) due to Charikar et al. (SODA '98). For the set connector problem, recently introduced by Fukunaga and Nagamochi (IPCO '07), we present a poly-logarithmic approximation; this result improves on the previously known ratio which can be Ω( n ) in the worst case.

When monitoring spatial phenomena with wireless sensor networks, selecting the best sensor placements is a fundamental task. Not only should the sensors be informative, but they should also be able to communicate efficiently. In this article, we present a data-driven approach that addresses the three central aspects of this problem: measuring the predictive quality of a set of sensor locations (regardless of whether sensors were ever placed at these locations), predicting the communication cost involved with these placements, and designing an algorithm with provable quality guarantees that optimizes the NP-hard trade-off. Specifically, we use data from a pilot deployment to build nonparametric probabilistic models called Gaussian Processes (GPs) both for the spatial phenomena of interest and for the spatial variability of link qualities, which allows us to estimate predictive power and communication cost of unsensed locations. Surprisingly, uncertainty in the representation of link qualities plays an important role in estimating communication costs. Using these models, we present a novel, polynomial-time, data-driven algorithm, PSPIEL, which selects Sensor Placements at Informative and communication-Efficient Locations. Our approach exploits two important properties of this problem: submodularity , formalizing the intuition that adding a node to a small deployment can help more than adding a node to a large deployment; and locality , under which nodes that are far from each other provide almost independent information. Exploiting these properties, we prove strong approximation guarantees for our PSPIEL approach. In addition, we show how our placements can be made robust against changes in the environment, and how PSPIEL can be used to plan informative paths for information gathering using mobile robots. We also provide extensive experimental validation of this practical approach on several real-world placement problems, and built a complete system implementation on 46 Tmote Sky motes, demonstrating significant advantages over existing methods.

Traditional efforts for scaling network intrusion detection (NIDS) and intrusion prevention systems (NIPS) have largely focused on a single-vantage-point view. In this paper, we explore an alternative design that exploits spatial, network-wide opportunities for distributing NIDS and NIPS functions. For the NIDS case, we design a linear programming formulation to assign detection responsibilities to nodes while ensuring that no node is overloaded. We describe a prototype NIDS implementation adapted from the Bro system to analyze traffic per these assignments, and demonstrate the advantages that this approach achieves. For NIPS, we show how to maximally leverage specialized hardware (e.g., TCAMs) to reduce the footprint of unwanted traffic on the network. Such hardware constraints make the optimization problem NP-hard, and we provide practical approximation algorithms based on randomized rounding.

Consider the following generalized min-sum set cover or multiple intents re-ranking problem proposed by Azar et al. (STOC 2009). We are given a universe of elements and a collection of subsets, with each set S having a covering requirement of K(S). The objective is to pick one element at a time such that the average covering time of the sets is minimized, where the covering time of a set S is the first time at which K(S) elements from it have been selected. There are two well-studied extreme cases of this problem: (i) when K(S) = 1 for all sets, we get the min-sum set cover problem, and (ii) when K(S) = |S| for all sets, we get the minimum-latency set cover problem. Constant factor approximations are known for both these problems. In their paper, Azar et al. considered the general problem and gave a logarithmic approximation algorithm for it. In this paper, we improve their result and give a simple randomized constant factor approximation algorithm for the generalized min-sum set cover problem.

The group Steiner problem is a classical network design problem where we are given a graph and a collection of groups of vertices, and want to build a min-cost subgraph that connects the root vertex to at least one vertex from each group. What if we wanted to build a subgraph that two-edge-connects the root to each group—that is, for every group g ⊆ V, the subgraph should contain two edge-disjoint paths from the root to some vertex in g? What if we wanted the two edge-disjoint paths to end up at distinct vertices in the group, so that the loss of a single member of the group would not destroy connectivity? In this paper, we investigate tree-embedding techniques that can be used to solve these and other 2-edge-connected network design problems. We illustrate the potential of these techniques by giving poly-logarithmic approximation algorithms for two-edge-connected versions of the group Steiner, connected facility location, buy-at-bulk, and the k-MST problems.

We present efficient differentially private algorithms for learning unions of polygons in the plane (which are not necessarily convex). Our algorithms are $(\alpha,\beta)$--probably approximately correct and $(\varepsilon,\delta)$--differentially private using a sample of size $\tilde{O}\left(\frac{1}{\alpha\varepsilon}k\log d\right)$, where the domain is $[d]\times[d]$ and $k$ is the number of edges in the union of polygons. Our algorithms are obtained by designing a private variant of the classical (nonprivate) learner for conjunctions using the greedy algorithm for set cover.

We give a (2+ε)-speed O(1)-competitive algorithm for scheduling jobs with arbitrary speed-up curves for the l2 norm of flow. We give a similar result for the broadcast setting with varying page sizes.

We consider the problem of embedding a metric into low-dimensional Euclidean space. The classical theorems of Bourgain, and of Johnson and Lindenstrauss say that any metric on n points embeds into an O (log n )-dimensional Euclidean space with O (log n ) distortion. Moreover, a simple “volume” argument shows that this bound is nearly tight: a uniform metric on n points requires nearly logarithmic number of dimensions to embed with logarithmic distortion. It is natural to ask whether such a volume restriction is the only hurdle to low-dimensional embeddings. In other words, do doubling metrics, that do not have large uniform submetrics, and thus no volume hurdles to low dimensional embeddings, embed in low dimensional Euclidean spaces with small distortion? In this article, we give a positive answer to this question. We show how to embed any doubling metrics into O (log log n ) dimensions with O (log n ) distortion. This is the first embedding for doubling metrics into fewer than logarithmic number of dimensions, even allowing for logarithmic distortion. This result is one extreme point of our general trade-off between distortion and dimension: given an n -point metric (V,d) with doubling dimension dim D , and any target dimension T in the range Ω(dim D log log n ) ≤ T ≤ O (log n ), we show that the metric embeds into Euclidean space R T with O (log n √ dim D / T ) distortion.

The k-forest problem is a common generalization of both the k-MST and the dense-k-subgraph problems. Formally, given a metric space on n vertices V , with m demand pairs ⊆ V × V and a “target” k ≤ m , the goal is to find a minimum cost subgraph that connects at least k pairs. In this paper, we give an O (min{√ n ⋅log k ,√ k })-approximation algorithm for k -forest, improving on the previous best ratio of O (min { n 2/3 ,√ m }log n ) by Segev and Segev. We then apply our algsorithm for k -forest to obtain approximation algorithms for several Dial-a-Ride problems. The basic Dial-a-Ride problem is the following: given an n point metric space with m objects each with its own source and destination, and a vehicle capable of carrying at most k objects at any time, find the minimum length tour that uses this vehicle to move each object from its source to destination. We want that the tour be non-preemptive : that is, each object, once picked up at its source, is dropped only at its destination. We prove that an α-approximation algorithm for the k -forest problem implies an O (α⋅log 2 n )-approximation algorithm for Dial-a-Ride. Using our results for k -forest, we get an O (min{√ n ,√ k }⋅log 2 n )-approximation algorithm for Dial-a-Ride. The only previous result known for Dial-a-Ride was an O (√ k log n )-approximation by Charikar and Raghavachari; our results give a different proof of a similar approximation guarantee—in fact, when the vehicle capacity k is large, we give a slight improvement on their results. The reduction from Dial-a-Ride to the k -forest problem is fairly robust, and allows us to obtain approximation algorithms (with the same guarantee) for some interesting generalizations of Dial-a-Ride.

The classical secretary problem studies the problem of selecting online an element (a “secretary”) with maximum value in a randomly ordered sequence. The difficulty lies in the fact that an element must be either selected or discarded upon its arrival, and this decision is irrevocable. Constant-competitive algorithms are known for the classical secretary problems (see, e.g., the survey of Freeman [7]) and several variants. We study the following two extensions of the secretary problem: In the discounted secretary problem, there is a time-dependent “discount” factor d(t), and the benefit derived from selecting an element/secretary e at time t is d(t)·v(e). For this problem with arbitrary (not necessarily decreasing) functions d(t), we show a constant-competitive algorithm when the expected optimum is known in advance. With no prior knowledge, we exhibit a lower bound of , and give a nearly-matching O (log n)-competitive algorithm. In the weighted secretary problem, up to K secretaries can be selected; when a secretary is selected (s)he must be irrevocably assigned to one of K positions, with position k having weight w(k), and assigning object/secretary e to position k has benefit w(k) · v(e). The goal is to select secretaries and assign them to positions to maximize Σe,k w(k) · v(e) · xek where xek is an indicator variable that secretary e is assigned position k. We give constant-competitive algorithms for this problem. Most of these results can also be extended to the matroid secretary case (Babaioff et al. [2]) for a large family of matroids with a constant-factor loss, and an O(log rank) loss for general matroids. These results are based on a reduction from various matroids to partition matroids which present a unified approach to many of the upper bounds of Babaioff et al. These problems have connections to online mechanism design (see, e.g., Hajiaghayi et al. [9]). All our algorithms are monotone, and hence lead to truthful mechanisms for the corresponding online auction problems.

Approximation algorithms for clustering points in metric spaces is a flourishing area of research, with much research effort spent on getting a better understanding of the approximation guarantees possible for many objective functions such as k-median, k-means, and min-sum clustering. This quest for better approximation algorithms is further fueled by the implicit hope that these better approximations also yield more accurate clusterings. E.g., for many problems such as clustering proteins by function, or clustering images by subject, there is some unknown correct “target” clustering and the implicit hope is that approximately optimizing these objective functions will in fact produce a clustering that is close pointwise to the truth. In this paper, we show that if we make this implicit assumption explicit—that is, if we assume that any c-approximation to the given clustering objective Φ is ∊-close to the target—then we can produce clusterings that are O(∊)-close to the target, even for values c for which obtaining a c-approximation is NP-hard. In particular, for k-median and k-means objectives, we show that we can achieve this guarantee for any constant c > 1, and for the min-sum objective we can do this for any constant c > 2. Our results also highlight a surprising conceptual difference between assuming that the optimal solution to, say, the k-median objective is ∊-close to the target, and assuming that any approximately optimal solution is ∊-close to the target, even for approximation factor say c = 1.01. In the former case, the problem of finding a solution that is O(∊)-close to the target remains computationally hard, and yet for the latter we have an efficient algorithm.

We consider the stochastic Steiner forest problem: suppose we were given a collection of Steiner forest instances, and were guaranteed that a random one of these instances would appear tomorrow; moreover, the cost of edges tomorrow will be λ times the cost of edges today. Which edges should we buy today so that we can extend it to a solution for the instance arriving tomorrow, to minimize the expected total cost? While very general results have been developed for many problems in stochastic discrete optimization over the past years, the approximation status of the stochastic Steiner Forest problem has remained open, with previous works yielding constant-factor approximations only for special cases. We resolve the status of this problem by giving a constant-factor primal-dual based approximation algorithm.

Consider the edge-connectivity survivable network design problem: given a graph G = (V,E) with edge-costs, and edge-connectivity requirements rij for every pair of vertices i,j, find an (approximately) minimum-cost network that provides the required connectivity. While this problem is known to admit good approximation algorithms in the offline case, no algorithms were known for this problem in the online setting.

Given a metric M=(V,d), a graph G=(V,E) is a t-spanner for M if every pair of nodes in V has a “short” path (i.e., of length at most t times their actual distance) between them in the spanner. Furthermore, this spanner has a hop diameter bounded by D if every pair of nodes has such a short path that also uses at most D edges. We consider the problem of constructing sparse (1+ε)-spanners with small hop diameter for metrics of low doubling dimension. In this paper, we show that given any metric with constant doubling dimension k and any 0<ε<1, one can find (1+ε)-spanner for the metric with nearly linear number of edges (i.e., only O(nlog * n+n ε −O(k)) edges) and constant hop diameter; we can also obtain a (1+ε)-spanner with linear number of edges (i.e., only n ε −O(k) edges) that achieves a hop diameter that grows like the functional inverse of Ackermann’s function. Moreover, we prove that such tradeoffs between the number of edges and the hop diameter are asymptotically optimal.

We consider the problem of embedding finite metrics with slack: We seek to produce embeddings with small dimension and distortion while allowing a (small) constant fraction of all distances to be arbitrarily distorted. This definition is motivated by recent research in the networking community, which achieved striking empirical success at embedding Internet latencies with low distortion into low-dimensional Euclidean space, provided that some small slack is allowed. Answering an open question of Kleinberg, Slivkins, and Wexler [in Proceedings of the 45th IEEE Symposium on Foundations of Computer Science, 2004], we show that provable guarantees of this type can in fact be achieved in general: Any finite metric space can be embedded, with constant slack and constant distortion, into constant-dimensional Euclidean space. We then show that there exist stronger embeddings into $\ell_1$ which exhibit gracefully degrading distortion: There is a single embedding into $\ell_1$ that achieves distortion at most $O(\log\frac{1}{\epsilon})$ on all but at most-1.5pt an $\epsilon$ fraction of distances simultaneously for all $\epsilon>0$. We extend this with distortion1pt $O(\log\frac{1}{\epsilon})^{1/p}$ to maps into general $\ell_p$, $p\geq1$, for several classes of metrics, including those with bounded doubling dimension and those arising from the shortest-path metric of a graph with an excluded minor. Finally, we show that many of our constructions are tight and give a general technique to obtain lower bounds for $\epsilon$-slack embeddings from lower bounds for low-distortion embeddings.

Given a universe U of n elements and a weighted collection l of m subsets of U, the universal set cover problem is to a-priori map each element u epsi U to a set S(u) epsi l containing u, so that X sube U is covered by S(X)=UuepsiXS(u). The aim is finding a mapping such that the cost of S(X) is as close as possible to the optimal set-cover cost for X. (Such problems are also called oblivious or a-priori optimization problems.) Unfortunately, for every universal mapping, the cost of S(X) can be Omega(radicn) times larger than optimal if the set X is adversarially chosen. In this paper we study the performance on average, when X is a set of randomly chosen elements from the universe: we show how to efficiently find a universal map whose expected cost is O(log mn) times the expected optimal cost. In fact, we give a slightly improved analysis and show that this is the best possible. We generalize these ideas to weighted set cover and show similar guarantees to (non-metric) facility location, where we have to balance the facility opening cost with the cost of connecting clients to the facilities. We show applications of our results to universal multi-cut and disc-covering problems, and show how all these universal mappings give us stochastic online algorithms with the same competitive factors.

In many optimization problems, a solution can be viewed as ascribing a ``cost\'\' to each client, and the goal is to optimize some aggregation of the per-client costs. We often optimize some $L_p$-norm (or some other symmetric convex function or norm) of the vector of costs---though different applications may suggest different norms to use. Ideally, we could obtain a solution that optimizes several norms simultaneously. In this paper, we examine approximation algorithms that simultaneously perform well on all norms, or on all $L_p$ norms. A natural problem in this framework is the $L_p$ Set Cover problem, which generalizes \textsc{Set Cover} and \textsc{Min-Sum Set Cover}. We show that the greedy algorithm \emph{simultaneously gives a $(p + \ln p + O(1))$-approximation for all $p$, and show that this approximation ratio is optimal up to constants} under reasonable complexity-theoretic assumptions. We additionally show how to use our analysis techniques to give similar results for the more general \emph{submodular set cover}, and prove some results for the so-called \emph{pipelined set cover} problem. We then go on to examine approximation algorithms in the ``all-norms\'\' and the ``all-$L_p$-norms\'\' frameworks more broadly, and present algorithms and structural results for other problems such as $k$-facility-location, TSP, and average flow-time minimization, extending and unifying previously known results.

In this article, we study metrics of negative type , which are metrics ( V , d) such that √d is an Euclidean metric; these metrics are thus also known as ℓ 2 -squared metrics. We show how to embed n -point negative-type metrics into Euclidean space ℓ 2 with distortion D = O (log 3/4 n ). This embedding result, in turn, implies an O (log 3/4 k )-approximation algorithm for the Sparsest Cut problem with nonuniform demands. Another corollary we obtain is that n -point subsets of ℓ 1 embed into ℓ 2 with distortion O (log 3/4 n ).

Consider the following classical network design problem: a set of terminals T = { t i } wishes to send traffic to a root r in an n -node graph G = ( V , E ). Each terminal t i sends d i units of traffic and enough bandwidth has to be allocated on the edges to permit this. However, bandwidth on an edge e can only be allocated in integral multiples of some base capacity u e and hence provisioning k × u e bandwidth on edge e incurs a cost of ⌈k⌉ times the cost of that edge. The objective is a minimum-cost feasible solution. This is one of many network design problems widely studied where the bandwidth allocation is governed by side constraints: edges can only allow a subset of cables to be purchased on them or certain quality-of-service requirements may have to be met. In this work, we show that this problem and, in fact, several basic problems in this general network design framework cannot be approximated better than Ω(log log n ) unless NP ⊆ DTIME ( n O (log log log n ) ), where | V | = n . In particular, we show that this inapproximability threshold holds for (i) the Priority-Steiner Tree problem, (ii) the (single-sink) Cost-Distance problem, and (iii) the single-sink version of an even more fundamental problem, Fixed Charge Network Flow. Our results provide a further breakthrough in the understanding of the level of complexity of network design problems. These are the first nonconstant hardness results known for all these problems.

Static expression experiments analyze samples from many individuals. These samples are often snapshots of the progression of a certain disease such as cancer. This raises an intriguing question: Can we determine a temporal order for these samples? Such an ordering can lead to better understanding of the dynamics of the disease and to the identification of genes associated with its progression. In this paper we formally prove, for the first time, that under a model for the dynamics of the expression levels of a single gene, it is indeed possible to recover the correct ordering of the static expression datasets by solving an instance of the traveling salesman problem (TSP). In addition, we devise an algorithm that combines a TSP heuristic and probabilistic modeling for inferring the underlying temporal order of the microarray experiments. This algorithm constructs probabilistic continuous curves to represent expression profiles leading to accurate temporal reconstruction for human data. Applying our method to cancer expression data we show that the ordering derived agrees well with survival duration. A classifier that utilizes this ordering improves upon other classifiers suggested for this task. The set of genes displaying consistent behavior for the determined ordering are enriched for genes associated with cancer progression.

In many applications, one has to actively select among a set of expensive observa- tions before making an informed decision. Often, we want to select observations which perform well when evaluated with an objective function chosen by an adver- sary. Examples include minimizing the maximum posterior variance in Gaussian Process regression, robust experimental design, and sensor placement for outbreak detection. In this paper, we present the Submodular Saturation algorithm, a sim- ple and efficient algorithm with strong theoretical approximation guarantees for the case where the possible objective functions exhibit submodularity, an intuitive diminishing returns property. Moreover, we prove that better approximation al- gorithms do not exist unless NP-complete problems admit efficient algorithms. We evaluate our algorithm on several real-world problems. For Gaussian Process regression, our algorithm compares favorably with state-of-the-art heuristics de- scribed in the geostatistics literature, while being simpler, faster and providing theoretical guarantees. For robust experimental design, our algorithm performs favorably compared to SDP-based algorithms.

We present constant-factor approximation algorithms for several widely-studied NP-hard optimization problems in network design, including the multicommodity rent-or-buy, virtual private network design, and single-sink buy-at-bulk problems. Our algorithms are simple and their approximation ratios improve over those previously known, in some cases by orders of magnitude. We develop a general analysis framework to bound the approximation ratios of our algorithms. This framework is based on a novel connection between random sampling and game-theoretic cost sharing.

When monitoring spatial phenomena with wireless sensor networks, selecting the best sensor placements is a fundamental task. Not only should the sensors be informative, but they should also be able to communicate efficiently. In this paper, we present a data-driven approach that addresses the three central aspects of this problem: measuring the predictive quality of a set of sensor locations (regardless of whether sensors were ever placed at these locations), predicting the communication cost involved with these placements, and designing an algorithm with provable quality guarantees that optimizes the NP-hard tradeoff. Specifically, we use data from a pilot deployment to build non-parametric probabilistic models called Gaussian Processes (GPs)both for the spatial phenomena of interest and for the spatial variability of link qualities, which allows us to estimate predictive power and communication cost of unsensed locations. Surprisingly, uncertainty in the representation of link qualities plays an important role in estimating communication costs. Using these models, we present a novel, polynomial-time, data-driven algorithm, pSPIEL, which selects Sensor Placements at Informative and cost-Effective Locations. Our approach exploits two important properties of this problem: submodularity, formalizing the intuition that adding a node to a small deployment can help more than adding a node to a large deployment; and locality, under which nodes that are far from each other provide almost independent information. Exploiting these properties, we prove strong approximation guarantees for our pSPIEL approach. We also provide extensive experimental validation of this practical approach on several real-world placement problems, and built a complete system implementation on 46 Tmote Sky motes, demonstrating significant advantages over existing methods.

A quorum system over a universe of logical elements is a collection of subsets (quorums) of elements, any two of which intersect. In numerous distributed algorithms, the elements of the universe reside on the nodes of a physical network and the participating nodes access the system by contacting every element in some quorum, potentially causing the added network congestion induced by these quorum accesses to play a limiting factor in the performance of the algorithm.In this paper we initiate the study of algorithms to place universe elements on the nodes of a physical network so as to minimize the network congestion that results from quorum accesses, while also ensuring that no physical node is overloaded by access requests from clients. We consider two models, one in which communication routes can be chosen arbitrarily and one in which they are fixed in advance. We show that in either model, the optimal congestion (with respect to the load constraints) cannot be approximated to any factor (unless P=NP). However, we show that at most doubling the load on nodes allows us to achieve a congestion that is close to this optimal value. We also shed some light on the extent to which element migration can reduce congestion in this context.

We show that the shortest-path metric of any k-outerplanar graph, for any fixed k, can be approximated by a probability distribution over tree metrics with constant distortion and hence also embedded into $\ell_1$ with constant distortion. These graphs play a central role in polynomial time approximation schemes for many NP-hard optimization problems on general planar graphs and include the family of weighted $k\times n$ planar grids. This result implies a constant upper bound on the ratio between the sparsest cut and the maximum concurrent flow in multicommodity networks for k-outerplanar graphs, thus extending a theorem of Okamura and Seymour [J. Combin. Theory Ser. B, 31 (1981), pp. 75-81] for outerplanar graphs, and a result of Gupta et al. [Combinatorica, 24(2004), pp. 233-269] for treewidth-2 graphs. In addition, we obtain improved approximation ratios for k-outerplanar graphs on various problems for which approximation algorithms are based on probabilistic tree embeddings. We conjecture that these embeddings for k-outerplanar graphs may serve as building blocks for $\ell_1$ embeddings of more general metrics.

We consider the problem of embedding finite metrics with slack: we seek to produce embeddings with small dimension and distortion while allowing a (small) constant fraction of all distances to be arbitrarily distorted. This definition is motivated by recent research in the networking community, which achieved striking empirical success at embedding Internet latencies with low distortion into low-dimensional Euclidean space, provided that some small slack is allowed. Answering an open question of Kleinberg, Slivkins, and Wexler (2004), we show that provable guarantees of this type can in fact be achieved in general: any finite metric can be embedded, with constant slack and constant distortion, into constant-dimensional Euclidean space. We then show that there exist stronger embeddings into /spl lscr//sub 1/ which exhibit gracefully degrading distortion: these is a single embedding into /spl lscr//sub 1/ that achieves distortion at most O(log 1//spl epsi/) on all but at most an /spl epsi/ fraction of distances, simultaneously for all /spl epsi/ > 0. We extend this with distortion O(log 1//spl epsi/)/sup 1/p/ to maps into general /spl lscr//sub p/, p /spl ges/ 1 for several classes of metrics, including those with bounded doubling dimension and those arising from the shortest-path metric of a graph with an excluded minor. Finally, we show that many of our constructions are tight, and give a general technique to obtain lower bounds for /spl epsi/-slack embeddings from lower bounds for low-distortion embeddings.

A quorum system is a family of sets (themselves called quorums), each pair of which intersect. In many distributed algorithms, the basic unit accessed by a client is a quorum of nodes. Such algorithms are used for applications such as mutual exclusion, data replication, and dissemination of information. However, accessing spread-out quorums causes access delays that we would like to minimize. Furthermore, every member of the quorum incurs processing load to handle quorum accesses by clients.In this paper we study the problem of placing quorums in a physical network so as to minimize the delay that clients incur by accessing quorums, and while respecting each physical node's capacity (in terms of the load of client requests it can handle). We provide approximation algorithms for this problem for two natural measures of delay (the max-delay and total-delay). All our algorithms ensure that each node's load is within a constant factor of its capacity, and minimize delay to within a constant factor of the optimal delay for all capacity-respecting solutions. We also provide better approximations for several well-known quorum systems.

Real-world networks often need to be designed under uncertainty, with only partial information and predictions of demand available at the outset of the design process. The field of stochastic optimization deals with such problems where the forecasts are specified in terms of probability distributions of future data. In this paper, we broaden the set of models as well as the techniques being considered for approximating stochastic optimization problems. For example, we look at stochastic models where the cost of the elements is correlated to the set of realized demands, and risk-averse models where upper bounds are placed on the amount spent in each of the stages. These generalized models require new techniques, and our solutions are based on a novel combination of the primal-dual method truncated based on optimal LP relaxation values, followed by a tree-rounding stage. We use these to give constant-factor approximation algorithms for the stochastic Steiner tree and single sink network design problems in these generalized models.

Several combinatorial optimization problems choose elements to minimize the total cost of constructing a feasible solution that satisfies requirements of clients. In the Steiner Tree problem, for example, edges must be chosen to connect terminals (clients); in Vertex Cover, vertices must be chosen to cover edges (clients); in Facility Location, facilities must be chosen and demand vertices (clients) connected to these chosen facilities. We consider a stochastic version of such a problem where the solution is constructed in two stages: Before the actual requirements materialize, we can choose elements in a first stage. The actual requirements are then revealed, drawn from a pre-specified probability distribution π thereupon, some more elements may be chosen to obtain a feasible solution for the actual requirements. However, in this second (recourse) stage, choosing an element is costlier by a factor of σ> 1. The goal is to minimize the first stage cost plus the expected second stage cost.We give a general yet simple technique to adapt approximation algorithms for several deterministic problems to their stochastic versions via the following method. First stage: Draw σ independent sets of clients from the distribution π and apply the approximation algorithm to construct a feasible solution for the union of these sets. Second stage: Since the actual requirements have now been revealed, augment the first-stage solution to be feasible for these requirements. We use this framework to derive constant factor approximations for stochastic versions of Vertex Cover, Steiner Tree and Uncapacitated Facility Location for arbitrary distributions π in one fell swoop. For special (product) distributions, we obtain additional and improved results. Our techniques adapt and use the notion of strict cost-shares introduced in [5].

A new packet routing model proposed by the Internet Engineering Task Force is MultiProtocol Label Switching, or MPLS [B. Davie and Y. Rekhter, MPLS: Technology and Applications, Morgan Kaufmann (Elsevier), New York, 2000]. Instead of each router's parsing the packet network layer header and doing its lookups based on that analysis (as in much of conventional packet routing), MPLS ensures that the analysis of the header is performed just once. The packet is then assigned a stack of labels, where the labels are usually much smaller than the packet headers themselves. When a router receives a packet, it examines the label at the top of the label stack and makes the decision of where the packet is forwarded based solely on that label. It can pop the top label off the stack if it so desires, and can also push some new labels onto the stack, before forwarding the packet. This scheme has several advantages over conventional routing protocols, the two primary ones being (a) reduced amount of header analysis at intermediate routers, which allows for faster switching times, and (b) better traffic engineering capabilities and hence easier handling of quality of service issues. However, essentially nothing is known at a theoretical level about the performance one can achieve with this protocol, or about the intrinsic trade-offs in its use of resources. This paper initiates a theoretical study of MPLS protocols, and routing algorithms and lower bounds are given for a variety of situations. We first study the routing problem on the line, a case which is already nontrivial, and give routing protocols whose trade-offs are close to optimality. We then extend our results for paths to trees, and thence onto more general graphs. These routing algorithms on general graphs are obtained by finding a tree cover of a graph, i.e., a small family of subtrees of the graph such that, for each pair of vertices, one of the trees in the family contains an (almost-)shortest path between them. Our results show tree covers of logarithmic size for planar graphs and graphs with bounded separators, which may be of independent interest.

The doubling constant of a metric space (X, d) is the smallest value /spl lambda/ such that every ball in X can be covered by /spl lambda/ balls of half the radius. The doubling dimension of X is then defined as dim (X) = log/sub 2//spl lambda/. A metric (or sequence of metrics) is called doubling precisely when its doubling dimension is bounded. This is a robust class of metric spaces which contains many families of metrics that occur in applied settings. We give tight bounds for embedding doubling metrics into (low-dimensional) normed spaces. We consider both general doubling metrics, as well as more restricted families such as those arising from trees, from graphs excluding a fixed minor, and from snowflaked metrics. Our techniques include decomposition theorems for doubling metrics, and an analysis of a fractal in the plane according to T. J. Laakso (2002). Finally, we discuss some applications and point out a central open question regarding dimensionality reduction in L/sub 2/.

We study the multicommodity rent-or-buy problem, a type of network design problem with economies of scale. In this problem, capacity on an edge can be rented, with cost incurred on a per-unit of capacity basis, or bought, which allows unlimited use after payment of a large fixed cost. Given a graph and a set of source-sink pairs, we seek a minimum-cost way of installing sufficient capacity on edges so that a prescribed amount of flow can be sent simultaneously from each source to the corresponding sink. The first constant-factor approximation algorithm for this problem was recently given by Kumar et al.; however, this algorithm and its analysis are both quite complicated, and its performance guarantee is extremely large. In this paper, we give a conceptually simple 12-approximation algorithm for this problem. Our analysis of this algorithm makes crucial use of cost sharing, the task of allocating the cost of an object to many users of the object in a "fair" manner. While techniques from approximation algorithms have recently yielded new progress on cost sharing problems, our work is the first to show the converse - those ideas from cost sharing can be fruitfully applied in the design and analysis of approximation algorithms.

Multiprotocol label switching or MPLS technology is being increasingly deployed by several of the largest Internet service providers to solve problems such as traffic engineering and to offer IP services like virtual private networks (VPNs). In MPLS, the analysis of the packet (network layer) header is performed just once, and each packet is assigned a stack of labels, which is examined by subsequent routers when making forwarding decisions. Despite the fact that MPLS is becoming widespread on the Internet, we know essentially very little about the performance one can achieve with it, and about the intrinsic trade-offs in its use of resources. In this paper, we undertake a comprehensive study of the label size versus stack depth trade-off for MPLS routing protocols on lines and trees. We show that in addition to LSP tunneling, label stacks can also be used to dramatically reduce the number of labels required for setting up MPLS LSPs in a network. Based on this observation, we develop routing algorithms and prove lower bounds for two basic problems: (1) fixed label routing: given a fixed number of labels, we want to minimize the stack depth, and (2) fixed stack routing: given a bound on the stack depth, we want to minimize the number of labels used. Our simulation results validate our approach, demonstrating that our novel protocols enable MPLS routing on large trees with few labels and small stack sizes. Thus, our MPLS routing algorithms are applicable to a number of practical scenarios involving the provisioning of VPNs and multicast trees.

MPLS (MultiProtocol Label Switching) is a new technology proposed by the IETF [4,10] for network routing, and is being increasingly deployed by the largest Internet service providers. The MPLS technology differs from conventional network protocols in a crucial way: instead of reading the entire packet header at all switching points, the analysis of the packet header is done just once, when the packet header is assigned a stack of labels, and thenceforth, each switching point or router just gets to look at the label at the top of the stack (and the ingress edge), and based only on this information, it has to make a decision about the next-hop node [17,16]. In another departure from conventional routing and in particular from IP source routing, where the entire packet route is explicitly put in the header and popped off along the route, the router can not only pop the top label, it can push other labels on top of the stack.The two parameters of interest in designing MPLS routing protocols are the number of labels used, and the depth of the stack used for routing. Clearly, both cannot be simultaneously minimized, and there is often an interesting trade-off between label size and stack depth: it is obvious that if k labels are used, one must have a stack depth of logk n.In fact, it was not known whether this bound could be achieved even for trees; the best stack depth previously achieved with a constant number of labels was O(log2 n). In this paper, we show that one can indeed get asymptotically optimal upper bounds and route on trees using k labels and a maximum stack depth of O(logk n), and that this trade-off can be achieved using a simpler and more intuitive protocol than the one given in [9]. These tree-routing ideas are then shown to give better routing protocols for planar graphs as well. In particular, we show how to route along near-shortest paths (of length at most (1 + e) times the shortest-path) with O(log2 n/e) labels and logarithmic stack depth. We also apply them to graphs with smaller separators, including most large ISP networks.

We give simple and easy-to-analyze randomized approximation algorithms for several well-studied NP-hard network design problems. Our algorithms improve over the previously best known approximation ratios. Our main results are the following. We give a randomized 3.55-approximation algorithm for the connected facility location problem. The algorithm requires three lines to state, one page to analyze, and improves the best-known performance guarantee for the problem. We give a 5.55-approximation algorithm for virtual private network design. Previously, constant-factor approximation algorithms were known only for special cases of this problem. We give a simple constant-factor approximation algorithm for the single-sink buy-at-bulk network design problem. Our performance guarantee improves over what was previously known, and is an order of magnitude improvement over previous combinatorial approximation algorithms for the problem.

We present the first constant factor approximation algorithm for network design with multiple commodities and economies of scale. We consider the rent-or-buy problem, a type of multicommodity buy-at-bulk network design in which there are two ways to install capacity on any given edge. Capacity can be rented, with cost incurred on a per-unit of capacity basis, or bought, which allows unlimited use after payment of a large fixed cost. Given a graph and a set of source-sink pairs, we seek a minimum-cost way of installing sufficient capacity on edges so that a prescribed amount of flow can be sent simultaneously from each source to the corresponding sink. Recent work on buy-at-bulk network design has concentrated on the special case where all sinks are identical; existing constant factor approximation algorithms for this special case make crucial use of the assumption that all commodities ship flow to the same sink vertex and do not obviously extend to the multicommodity rent-or-buy problem. Prior to our work, the best heuristics for the multicommodity rent-or-buy problem achieved only logarithmic performance guarantees and relied on the machinery of relaxed metrical task systems or of metric embeddings. By contrast, we solve the network design problem directly via a novel primal-dual algorithm.

The study of the effect of priced information on basic algorithmic problems was initiated by M. Charikar et al. (2000). The authors continue the study of sorting and selection in the priced comparison model, i.e., when each comparison has an associated cost, and answer some of the open problems suggested by Charikar et al. If the comparison costs are allowed to be arbitrary, we show that one cannot get good approximation ratios. A different way to assign costs is based on the idea that one can distill out an intrinsic value for each item being compared, such that the cost of comparing two elements is some "well-behaved" or "structured" function of their values. We feel that most practical applications will have some structured cost property. The authors study the problems of sorting and selection (which includes finding the maximum and the median) in the structured cost model. We get a variety of approximation results for these problems, depending on the restrictions we put on the structured costs. We show that it is possible to get much improved results with the structured cost model than the case when we do not have any assumptions on comparison costs.

MultiProtocol Label Switching (MPLS) is a routing model proposed by the IETF for the Internet, and is becoming widely popular. In this paper, we initiate a theoretical study of the routing model, and give routing algorithms and lower bounds in a variety of situations. We first study the routing problems on the line. We then build up our results from paths through trees to more general graphs. The basic technique to go to general graphs is that of finding a tree cover, which is a small set of subtrees of the graph such that for each pair of vertices, one of the trees contains a shortest (or near-shortest) path between them. The concept of tree covers appears to have many interesting applications.

Consider a setting in which a group of nodes, situated in a large underlying network, wishes to reserve bandwidth on which to support communication. Virtual private networks (VPNs) are services that support such a construct; rather than building a new physical network on the group of nodes that must be connected, bandwidth in the underlying network is reserved for communication within the group, forming a virtual “sub-network.”

In a traditional classification problem, we wish to assign labels from a set to each of objects so that the labeling is consistent with some observed data that includes pairwise relationships among the objects. Kleinberg and Tardos recently formulated a general classification problem of this type, the “metric labeling problem”, and gave an approximation algorithm for it. The algorithm is based on solving a linear programming relaxation of a natural integer program and then randomized rounding. In this paper we consider an important case of the metric labeling problem, in which the metric is the truncated linear metric. This is a natural non-uniform and robust metric, and it arises in a number of applications. We give a combinatorial 4-approximation algorithm for this metric. Our algorithm is a natural local search method, where the local steps are based on minimum cut computations in an appropriately constructed graph. Our method extends previous work by Boykov, Veksler and Zabih on more restricted classes of metrics.

Motivated by many recent algorithmic applications, the paper aims to promote a systematic study of the relationship between the topology of a graph and the metric distortion incurred where the graph is embedded into l/sub 1/ space. The main results are: 1. Explicit constant-distortion embeddings of all series parallel graphs, and all graphs with bounded Euler number. These are thus the first natural families known to have constant distortion (strictly greater than 1). Using the above embeddings, we obtain algorithms to approximate the sparsest cut in such graphs to within a constant factor. 2) A constant-distortion embedding of outerplanar graphs into the restricted class of l/sub 1/-metrics known as "dominating tree metrics". We also show a lower bound of /spl Omega/(log n) on the distortion for embeddings of series-parallel graphs into (distributions over) dominating tree metrics. This shows, surprisingly, that such metrics approximate distances very poorly even for families of graphs with low tree width, and excludes the possibility of using them to explore the finer structure of l/sub 1/-embeddability.

We consider embedding metrics induced by trees into Euclidean spaces with a restricted number of dimensions. We show that any weighted tree T with n vertices and L leaves can be embedded into d -dimensional Euclidean space with Õ (L 1/(d-1) ) distortion. Furthermore, we exhibit an embedding with almost the same distortion which can be computed efficiently. This distortion substantially improves the previous best upper bound of \\tilde O (n 2/d ) and almost matches the best known lower bound of Ω(L 1/d ) .