Sampling algorithms play a pivotal role in probabilistic AI. However, verifying if a sampler program indeed samples from the claimed distribution is a notoriously hard problem. Provably correct testers like Barbarik,Teq,Flash, Cubeprobe for testing of different kinds of samplers were proposed only in the last few years. All these testers focus on the worst-case efficiency, and do not support verification of samplers over infinite domains, a case occurring frequently in Astronomy, Finance, Network Security etc. In this work, we design the first tester of samplers with instance-dependent efficiency, allowing us to test samplers over natural numbers. Our tests are developed via a novel distance estimation algorithm between an unknown and a known probability distribution using an 'interval conditioning' framework. The core technical contribution is a new connection with probability mass estimation of a continuous distribution. The practical gains are also substantial—our experiments establish up to 1000× speedup over state-of-the-art testers.

We study the sample complexity of conditional independence testing. In this problem, given i.i.d. samples from a discrete distribution $P_{ABC}$, the goal is to distinguish whether $A$ and $C$ are conditionally independent with respect to $B$, i.e., $P_{ABC}=P_{A|B}P_BP_{C|B}$, or whether $A$ and $C$ are conditionally dependent, $\Delta(P_{ABC},P_{A|B}P_BP_{C|B})\geq \varepsilon$ for some fixed threshold $\varepsilon$ and distance measure $\Delta$. We are interested in the cases where $\Delta$ is either the $\ell_1$ distance or the KL-divergence. The study for the case of $\ell_1$ distance was initiated by (Canonne et al., STOC 2018), and the KL-divergence was recently studied by (Seyfried et al., COLT 2025). Both works design algorithms whose sample complexities scale sublinearly in the dimensions of the subsystems, and showed tight lower bounds in some parameter regimes. While Canonne et al. derived partial lower bounds for the remaining regimes as well, the problem of fully resolving the sample complexity in all parameters remained open. In this work, we settle these open questions and prove optimal sample complexity lower bounds for both of these problems, thereby completely settling the sample complexities up to polylogarithmic factors.

Over the last three decades, function testing has been extensively studied over Boolean, finite fields, and discrete settings. However, to encode the real-world applications more succinctly, function testing over the reals (where the domain and range, both are reals) is of prime importance. Recently, there have been some works in the direction of testing for algebraic representations of such functions: the work by Fleming and Yoshida (ITCS 20), Arora, Kelman, and Meir (SOSA 25) on linearity testing and the work of Arora, Bhattacharyya, Fleming, Kelman, and Yoshida (SODA 23) for testing low-degree polynomials. Our work follows the same avenue, wherein we study three well-studied sparse representations of functions, over the reals, namely (i) k-linearity, (ii) k-sparse, low-degree polynomials, and (iii) k-juntas. In this setting, given approximate query access to some f:ℝⁿ → ℝ, we want to decide if the function satisfies some property of interest, or if it is far from all functions that satisfy the property. Here, the distance is measured in the 𝓁₁-metric, under the assumption that we are drawing samples from the Standard Gaussian distribution. We present efficient testers and Ω(k) lower bounds for testing each of these three properties.

We initiate the study of quantum property testing in sparse directed graphs, and more particularly in the unidirectional model, where the algorithm is allowed to query only the outgoing edges of a vertex. In the classical unidirectional model, the problem of testing $k$-star-freeness, and more generally $k$-source-subgraph-freeness, is almost maximally hard for large $k$. We prove that this problem has almost quadratic advantage in the quantum setting. Moreover, we show that this advantage is nearly tight, by showing a quantum lower bound using the method of dual polynomials on an intermediate problem for a new, property testing version of the $k$-collision problem that was not studied before. To illustrate that not all problems in graph property testing admit such a quantum speedup, we consider the problem of $3$-colorability in the related undirected bounded-degree model, when graphs are now undirected. This problem is maximally hard to test classically, and we show that also quantumly it requires a linear number of queries.

We investigate the sample complexity of mutual information and conditional mutual information testing. For conditional mutual information testing, given access to independent samples of a triple of random variables $(A, B, C)$ with unknown distribution, we want to distinguish between two cases: (i) $A$ and $C$ are conditionally independent, i.e., $I(A:C|B) = 0$, and (ii) $A$ and $C$ are conditionally dependent, i.e., $I(A:C|B) \geq \varepsilon$ for some threshold $\varepsilon$. We establish an upper bound on the number of samples required to distinguish between the two cases with high confidence, as a function of $\varepsilon$ and the three alphabet sizes. We conjecture that our bound is tight and show that this is indeed the case in several parameter regimes. For the special case of mutual information testing (when $B$ is trivial), we establish the necessary and sufficient number of samples required up to polylogarithmic terms. Our technical contributions include a novel method to efficiently simulate weakly correlated samples from the conditionally independent distribution $P_{A|B} P_{C|B} P_B$ given access to samples from an unknown distribution $P_{ABC}$, and a new estimator for equivalence testing that can handle such correlated samples, which might be of independent interest.

This work establishes a novel link between the problem of PAC-learning high-dimensional graphical models and the task of (efficient) counting and sampling of graph structures, using an online learning framework. The problem of efficiently counting and sampling graphical structures, such as spanning trees and acyclic orientations, has been a vibrant area of research in algorithms. We show that this rich algorithmic foundation can be leveraged to develop new algorithms for learning high-dimensional graphical models. We present the first efficient algorithm for (both realizable and agnostic) learning of Bayes nets with a chordal skeleton. In particular, we present an algorithm that, given integers $k,d > 0$, error parameter $\varepsilon > 0$, an undirected chordal graph $G$ on $n$ vertices, and sample access to a distribution $P^\ast$ on $[k]^n$; (1) returns a Bayes net $\widehat{P}$ with skeleton $G$ and indegree $d$, whose KL-divergence from $P^\ast$ is at most $\varepsilon$ more than the optimal KL-divergence between $P^\ast$ and any Bayes net with skeleton $G$ and indegree $d$, (2) uses $\widetilde{O}(n^3k^{d+1}/\varepsilon^2)$ samples from $P^\ast$ and runs in time $\mathrm{poly}(n,k,\varepsilon^{-1})$ for constant $d$. Prior results in this spirit were for only for trees ($d=1$, tree skeleton) via Chow-Liu, and in the realizable setting for polytrees (arbitrary $d$ but tree skeleton). Thus, our result significantly extends the state-of-the-art in learning Bayes net distributions. We also establish new results for learning tree and polytree distributions.

The framework of distribution testing is currently ubiquitous in the field of property testing. In this model, the input is a probability distribution accessible via independently drawn samples from an oracle. The testing task is to distinguish a distribution that satisfies some property from a distribution that is far in some distance measure from satisfying it. The task of tolerant testing imposes a further restriction, that distributions close to satisfying the property are also accepted. This work focuses on the connection between the sample complexities of non-tolerant testing of distributions and their tolerant testing counterparts. When limiting our scope to label-invariant (symmetric) properties of distributions, we prove that the gap is at most quadratic, ignoring poly-logarithmic factors. Conversely, the property of being the uniform distribution is indeed known to have an almost-quadratic gap. When moving to general, not necessarily label-invariant properties, the situation is more complicated, and we show some partial results. We show that if a property requires the distributions to be non-concentrated, that is, the probability mass of the distribution is sufficiently spread out, then it cannot be non-tolerantly tested with$o(\sqrt {n})$many samples, where n denotes the universe size. Clearly, this implies at most a quadratic gap, because a distribution can be learned (and hence tolerantly tested against any property) using$\mathcal {O}(n)$many samples. Being non-concentrated is a strong requirement on properties, as we also prove a close to linear lower bound against their tolerant tests. Apart from the case where the distribution is non-concentrated, we also show if an input distribution is very concentrated, in the sense that it is mostly supported on a subset of size s of the universe, then it can be learned using only$\mathcal {O}(s)$many samples. The learning procedure adapts to the input, and works without knowing s in advance.

Samplers are the backbone of the implementations of any randomized algorithm. Unfortunately, obtaining an efficient algorithm to test the correctness of samplers is very hard to find. Recently, in a series of works, testers like Barbarik, Teq, Flash for testing of some particular kinds of samplers, like CNF-samplers and Horn-samplers, were obtained. However, their techniques have a significant limitation because one can not expect to use their methods to test for other samplers, such as perfect matching samplers or samplers for sampling linear extensions in posets. In this paper, we present a new testing algorithm that works for such samplers and can estimate the distance of a new sampler from a known sampler (say, the uniform sampler). Testing the identity of distributions is the heart of testing the correctness of samplers. This paper's main technical contribution is developing a new distance estimation algorithm for distributions over high-dimensional cubes using the recently proposed subcube conditioning sampling model. Given subcube conditioning access to an unknown distribution P, and a known distribution Q defined over an n-dimensional Boolean hypercube, our algorithm CubeProbeEst estimates the variation distance between P and Q within additive error using subcube conditional samples from P. Following the testing-via-learning paradigm, we also get a tester that distinguishes between the cases when P and Q are close or far in variation distance with high probability using subcube conditional samples. This estimation algorithm CubeProbeEst in the subcube conditioning sampling model helps us to design the first tester for self-reducible samplers. The correctness of the tester is formally proved. Moreover, we implement CubeProbeEst to test the quality of three samplers for sampling linear extensions in posets.

Sampling over combinatorial spaces is a fundamental problem in artificial intelligence with a wide variety of applications. Since state-of-the-art techniques heavily rely on heuristics whose rigorous analysis remains beyond the reach of current theoretical tools, the past few years have witnessed interest in the design of techniques to test the quality of samplers. The current state-of-the-art techniques, $\mathsf{Barbarik}$ and $\mathsf{Barbarik2}$, focuses on the cases where combinatorial spaces are encoded as Conjunctive Normal Form (CNF) formulas. While CNF is a general-purpose form, often techniques rely on exploiting specific representations to achieve speedup. Of particular interest are Horn clauses, which form the basis of the logic programming tools in AI. In this context, a natural question is whether it is possible to design a tester that can determine the correctness of a given Horn sampler. The primary contribution of this paper is an affirmative answer to the above question. We design the first tester, $\mathsf{Flash}$, which tests the correctness of a given Horn sampler: given a specific distribution $\mathcal{I}$ and parameters $\eta$, $\varepsilon$, and $\delta$, the tester $\mathsf{Flash}$ correctly (with probability at least $ 1-\delta$) distinguishes whether the underlying distribution of the Horn-sampler is “$\varepsilon$-close” to $\mathcal{I}$ or “$\eta$-far” from $\mathcal{I}$ by sampling only $\widetilde{\mathcal{O}}(\mathsf{tilt}^3/(\eta - \varepsilon)^4)$ samples from the Horn-sampler, where the $\mathsf{tilt}$ is the ratio of the maximum and the minimum (non-zero) probability masses of $\mathcal{I}$. We also provide a prototype implementation of $\mathsf{Flash}$ and test three state-of-the-art samplers on a set of benchmarks.

Distribution testing is a central part of property testing, with applications to various research areas, such as computational and statistical learning, information theory, and probabilistic program checking. The original distribution testing model relies on samples drawn independently from the distribution to be tested. However, when the distribution in question is over the $n$-dimensional Hamming cube $\left\{0,1\right\}^{n}$ for a large $n$, even reading a few samples is infeasible. To address this, Goldreich and Ron [ITCS 2022] have defined a model called the \emph{huge object model}, in which the samples may only be queried in a few places.For any sample/query model, the following three questions are considered fundamental: {\bf (i)} understand what classes of objects can be “learned \emph{easily}", {\bf (ii)} characterize {\em testable properties}, that is, properties that can be tested in the given sample/query model using a constant number of samples/queries, and {\bf (iii)} understand the {\em gap} between {\em adaptive} and {\em non-adaptive} query/sample complexities.In this work, we study these questions for the huge object model for distribution testing. To do so, we initiate a study of a general class of distribution properties that are invariant under a permutation of the indices of the vectors in $\left\{0,1\right\}^{n}$, while still not being necessarily fully symmetric as per the definition used in traditional distribution testing.We prove that every distribution over $\left\{0,1\right\}^{n}$ whose support has a bounded VC-dimension can be efficiently learned up to a permutation. The number of queries made by the algorithm depends only on the VC-dimension of the support of the distribution and is independent of $n$. This gives efficient testers for index-invariant distribution properties that admit a global VC-dimension bound. To complement this result, we argue that satisfying only index-invariance or only a VC-dimension bound is insufficient to guarantee a tester whose query complexity is independent of $n$. Moreover, we prove that the dependency of the sample and query complexities of our tester on the VC-dimension is essentially tight. As a second part of this work, we address the question of thenumber of queries required for non-adaptive testing. We show that it can be at most quadratic in the numberof queries required for an adaptive tester in the case of index-invariant properties. This contrasts with the tight (easily provable) exponential gap between adaptive and non-adaptive testers for general non-index-invariant properties. Finally, we provide an index-invariant property for which the quadratic gap between adaptive and non-adaptive query complexities for testing is almost tight.

The framework of distribution testing is currently ubiquitous in the field of property testing. In this model, the input is a probability distribution accessible via independently drawn samples from an oracle. The testing task is to distinguish a distribution that satisfies some property from a distribution that is far from satisfying it in the $\ell_1$ distance. The task of tolerant testing imposes a further restriction, that distributions close to satisfying the property are also accepted. This work focuses on the connection of the sample complexities of non-tolerant ("traditional") testing of distributions and tolerant testing thereof. When limiting our scope to label-invariant (symmetric) properties of distribution, we prove that the gap is at most quadratic. Conversely, the property of being the uniform distribution is indeed known to have an almost-quadratic gap. When moving to general, not necessarily label-invariant properties, the situation is more complicated, and we show some partial results. We show that if a property requires the distributions to be non-concentrated, then it cannot be non-tolerantly tested with $o(\sqrt{n})$ many samples, where $n$ denotes the universe size. Clearly, this implies at most a quadratic gap, because a distribution can be learned (and hence tolerantly tested against any property) using $\mathcal{O}(n)$ many samples. Being non-concentrated is a strong requirement on the property, as we also prove a close to linear lower bound against their tolerant tests. To provide evidence for other general cases (where the properties are not necessarily label-invariant), we show that if an input distribution is very concentrated, in the sense that it is mostly supported on a subset of size $s$ of the universe, then it can be learned using only $\mathcal{O}(s)$ many samples. The learning procedure adapts to the input, and works without knowing $s$ in advance.

Bipartite testing has been a central problem in the area of property testing since its inception in the seminal work of Goldreich, Goldwasser, and Ron. Though the non-tolerant version of bipartite testing has been extensively studied in the literature, the tolerant variant is not well understood. In this paper, we consider the following version of tolerant bipartite testing problem: Given two parameters ε, δ ∈ (0,1), with δ > ε, and access to the adjacency matrix of a graph G, we have to decide whether G can be made bipartite by editing at most ε n² entries of the adjacency matrix of G, or we have to edit at least δ n² entries of the adjacency matrix to make G bipartite. In this paper, we prove that for δ = (2+Ω(1))ε, tolerant bipartite testing can be decided by performing 𝒪̃(1/ε³) many adjacency queries and in 2^𝒪̃(1/ε) time complexity. This improves upon the state-of-the-art query and time complexities of this problem of 𝒪̃(1/ε⁶) and 2^𝒪̃(1/ε²), respectively, due to Alon, Fernandez de la Vega, Kannan and Karpinski, where 𝒪̃(⋅) hides a factor polynomial in log (1/ε).

The graph isomorphism distance between two graphs G_u and G_k is the fraction of entries in the adjacency matrix that has to be changed to make G_u isomorphic to G_k. We study the problem of estimating, up to a constant additive factor, the graph isomorphism distance between two graphs in the query model. In other words, if G_k is a known graph and G_u is an unknown graph whose adjacency matrix has to be accessed by querying the entries, what is the query complexity for testing whether the graph isomorphism distance between G_u and G_k is less than γ₁ or more than γ₂, where γ₁ and γ₂ are two constants with 0 ≤ γ₁ < γ₂ ≤ 1. It is also called the tolerant property testing of graph isomorphism in the dense graph model. The non-tolerant version (where γ₁ is 0) has been studied by Fischer and Matsliah (SICOMP'08). In this paper, we prove a (interesting) connection between tolerant graph isomorphism testing and tolerant testing of the well studied Earth Mover’s Distance (EMD). We prove that deciding tolerant graph isomorphism is equivalent to deciding tolerant EMD testing between multi-sets in the query setting. Moreover, the reductions between tolerant graph isomorphism and tolerant EMD testing (in query setting) can also be extended directly to work in the two party Alice-Bob communication model (where Alice and Bob have one graph each and they want to solve tolerant graph isomorphism problem by communicating bits), and possibly in other sublinear models as well. Testing tolerant EMD between two probability distributions is equivalent to testing EMD between two multi-sets, where the multiplicity of each element is taken appropriately, and we sample elements from the unknown multi-set with replacement. In this paper, our (main) contribution is to introduce the problem of {(tolerant) EMD testing between multi-sets (over Hamming cube) when we get samples from the unknown multi-set without replacement} and to show that this variant of tolerant testing of EMD is as hard as tolerant testing of graph isomorphism between two graphs. {Thus, while testing of equivalence between distributions is at the heart of the non-tolerant testing of graph isomorphism, we are showing that the estimation of the EMD over a Hamming cube (when we are allowed to sample without replacement) is at the heart of tolerant graph isomorphism.} We believe that the introduction of the problem of testing EMD between multi-sets (when we get samples without replacement) opens an entirely new direction in the world of testing properties of distributions.