Local search is widely used to solve combinatorial optimisation problems and to model biological evolution, but the performance of local search algorithms on different kinds of fitness landscapes is poorly understood. Here we introduce a natural approach to modelling fitness landscapes using valued constraints. This allows us to investigate minimal representations (normal forms) and to consider the effects of the structure of the constraint graph on the tractability of local search. First, we show that for fitness landscapes representable by binary Boolean valued constraints there is a minimal necessary constraint graph that can be easily computed. Second, we consider landscapes as equivalent if they allow the same (improving) local search moves; we show that a minimal normal form still exists, but is NP-hard to compute. Next we consider the complexity of local search on fitness landscapes modelled by valued constraints with restricted forms of constraint graph. In the binary Boolean case, we prove that a tree-structured constraint graph gives a tight quadratic bound on the number of improving moves made by any local search; hence, any landscape that can be represented by such a model will be tractable for local search. We build two families of examples to show that both the conditions in our tractability result are essential. With domain size three, even just a path of binary constraints can model a landscape with an exponentially long sequence of improving moves. With a treewidth two constraint graph, even with a maximum degree of three, binary Boolean constraints can model a landscape with an exponentially long sequence of improving moves.

Singleton arc consistency is an important type of local consistency which has been recently shown to solve all constraint satisfaction problems (CSPs) over constraint languages of bounded width. We aim to characterise all classes of CSPs defined by a forbidden pattern that are solved by singleton arc consistency and closed under removing constraints. We identify five new patterns whose absence ensures solvability by singleton arc consistency, four of which are provably maximal and three of which generalise 2-SAT. Combined with simple counter-examples for other patterns, we make significant progress towards a complete classification.

Singleton arc consistency is an important type of local consistency which has been recently shown to solve all constraint satisfaction problems (CSPs) over constraint languages of bounded width. We aim to characterise all classes of CSPs defined by a forbidden pattern that are solved by singleton arc consistency and closed under removing constraints. We identify five new patterns whose absence ensures solvability by singleton arc consistency, four of which are provably maximal and three of which generalise 2-SAT. Combined with simple counter-examples for other patterns, we make significant progress towards a complete classification.

We study methods for transforming valued constraint satisfaction problems (VCSPs) to binary VCSPs. First, we show that the standard dual encoding preserves many aspects of the algebraic properties that capture the computational complexity of VCSPs. Second, we extend the reduction of CSPs to binary CSPs described by Bulín et al. [ Log. Methods Comput. Sci., 11 (2015)] to VCSPs. This reduction establishes that VCSPs over a fixed valued constraint language are polynomial-time equivalent to minimum-cost homomorphism problems over a fixed digraph.

CSP instances are commonly solved by backtracking search combined with constraint propagation. During search, constraint solvers aim to remove any literals (variable-value pair) that can be shown not to be part of any solution. This literal removal, called propagation, is the beating heart of modern constraint solvers. A significant proportion of the runtime of propagating constraint solvers is spent running propagation algorithms. Therefore any mechanism for reducing how frequently propagators are called leads directly to significant performance improvements. One family of popular techniques is dynamic triggering — these techniques aim to avoid invoking a propagator when it would remove no literals. While this technique has been successful in practice, it has not yet been studied theoretically. This paper provides a theoretical framework for understanding when dynamic triggering will be successful. In particular, we prove when a literal deletion does not require a propagator to be executed. To achieve this, we describe supports: a support for a constraint is a set of literals whose presence in a search state ensures that propagating the constraint will not remove any literals. Therefore running the propagator when a literal outside the support is deleted is a waste of time. By characterising supports and giving a definition of dynamic and static supports for the CSP, we provide the framework for a proper analysis. We show how the number of triggers required for different constraints varies widely. For some constraints, dynamic triggering allows very small supports, for others the number of required supports is provably large.

Constraint programming is a natural paradigm for many combinatorial optimisation problems. The complexity of constraint satisfaction for various forms of constraints has been widely-studied, both to inform the choice of appropriate algorithms, and to understand better the boundary between polynomial-time complexity and NP-hardness. In constraint programming it is well-known that any constraint satisfaction problem can be converted to an equivalent binary problem using the so-called dual encoding. Using this standard approach any fixed collection of constraints, of arbitrary arity, can be converted to an equivalent set of constraints of arity at most two. Here we show that this transformation, although it changes the domain of the constraints, preserves all the relevant algebraic properties that determine the complexity. Moreover, we show that the dual encoding preserves many of the key algorithmic properties of the original instance. We also show that this remains true for more general valued constraint languages, where constraints may assign different cost values to different assignments. Hence, we obtain a simple proof of the fact that to classify the computational complexity of all valued constraint languages it suffices to classify only binary valued constraint languages.

The Workflow Satisfiability Problem (WSP) is a problem of practical interest that arises whenever tasks need to be performed by authorized users, subject to constraints defined by business rules. We are required to decide whether there exists a plan - an assignment of tasks to authorized users - such that all constraints are satisfied. It is natural to see the WSP as a subclass of the Constraint Satisfaction Problem (CSP) in which the variables are tasks and the domain is the set of users. What makes the WSP distinctive is that the number of tasks is usually very small compared to the number of users, so it is appropriate to ask for which constraint languages the WSP is fixed-parameter tractable (FPT), parameterized by the number of tasks. This novel approach to the WSP, using techniques from CSP, has enabled us to design a generic algorithm which is FPT for several families of workflow constraints considered in the literature. Furthermore, we prove that the union of FPT languages remains FPT if they satisfy a simple compatibility condition. Lastly, we identify a new FPT constraint language, user-independent constraints, that includes many of the constraints of interest in business processing systems. We demonstrate that our generic algorithm has provably optimal running time O*(2^(klog k)), for this language, where k is the number of tasks.

Discrete optimization problems arise in many different areas and are studied under many different names. In many such problems the quantity to be optimized can be expressed as a sum of functions of a restricted form. Here we present a unifying theory of complexity for problems of this kind. We show that the complexity of a finite-domain discrete optimization problem is determined by certain algebraic properties of the objective function, which we call weighted polymorphisms. We define a Galois connection between sets of rational-valued functions and sets of weighted polymorphisms and show how the closed sets of this Galois connection can be characterized. These results provide a new approach to studying the complexity of discrete optimization. We use this approach to identify certain maximal tractable subproblems of the general problem and hence derive a complete classification of complexity for the Boolean case.

The constraint satisfaction problem (CSP) is a general problem central to computer science and artificial intelligence. Although the CSP is NP-hard in general, considerable effort has been spent on identifying tractable subclasses. The main two approaches consider structural properties (restrictions on the hypergraph of constraint scopes) and relational properties (restrictions on the language of constraint relations). Recently, some authors have considered hybrid properties that restrict the constraint hypergraph and the relations simultaneously. Our key contribution is the novel concept of a CSP pattern and classes of problems defined by forbidden patterns (which can be viewed as forbidding generic sub-problems). We describe the theoretical framework which can be used to reason about classes of problems defined by forbidden patterns. We show that this framework generalises certain known hybrid tractable classes. Although we are not close to obtaining a complete characterisation concerning the tractability of general forbidden patterns, we prove a dichotomy in a special case: classes of problems that arise when we can only forbid binary negative patterns (generic sub-problems in which only disallowed tuples are specified). In this case we show that all (finite sets of) forbidden patterns define either polynomial-time solvable or NP-complete classes of instances.

While several papers exist focusing on the question of election into the Baseball Hall of Fame election, all of them use the same method for building a model: regression analysis. Problems with this include the fact that regressions are continuous functions, and thus have trouble modeling binary problems. While there are some methods that work reasonably well, a logical rule, which can be evaluated to only true or false, seems to be a better option for several reasons. In regression models, since the results are continuous, it is possible to get an answer other than just 1 or 0 for a binary problem. Because of this, an arbitrary cutoff to what is to be considered a positive example must be made. Instead, through the use of genetic algorithms, a logical rule, which can evaluate any example to either true or false (1 or 0) can be found. The rules found by this system are extremely accurate, with training accuracies around 99% and testing accuracies just lower at 97%.

Many researchers in artificial intelligence are beginning to explore the use of soft constraints to express a set of (possibly conflicting) problem requirements. A soft constraint is a function defined on a collection of variables which associates some measure of desirability with each possible combination of values for those variables. However, the crucial question of the computational complexity of finding the optimal solution to a collection of soft constraints has so far received very little attention. In this paper we identify a class of soft binary constraints for which the problem of finding the optimal solution is tractable. In other words, we show that for any given set of such constraints, there exists a polynomial time algorithm to determine the assignment having the best overall combined measure of desirability. This tractable class includes many commonly-occurring soft constraints, such as 'as near as possible' or 'as soon as possible after', as well as crisp constraints such as 'greater than'. Finally, we show that this tractable class is maximal, in the sense that adding any other form of soft binary constraint which is not in the class gives rise to a class of problems which is NP-hard.

Many combinatorial search problems can be expressed as 'constraint satisfaction problems'. This class of problems is known to be NP-hard in general, but a number of restricted constraint classes have been identified which ensure tractability. This paper presents the first general results on combining tractable constraint classes to obtain larger, more general, tractable classes. We give examples to show that many known examples of tractable constraint classes, from a wide variety of different contexts, can be constructed from simpler tractable classes using a general method. We also construct several new tractable classes that have not previously been identified.

Many combinatorial search problems can be expressed as “constraint satisfaction problems” and this class of problems is known to be NP-complete in general. In this paper, we investigate the subclasses that arise from restricting the possible constraint types. We first show that any set of constraints that does not give rise to an NP-complete class of problems must satisfy a certain type of algebraic closure condition. We then investigate all the different possible forms of this algebraic closure property, and establish which of these are sufficient to ensure tractability. As examples, we show that all known classes of tractable constraints over finite domains can be characterized by such an algebraic closure property. Finally, we describe a simple computational procedure that can be used to determine the closure properties of a given set of constraints. This procedure involves solving a particular constraint satisfaction problem, which we call an “indicator problem.”

The paper describes techniques for constructing statistically independent binary sequences with prescribed ratios of zeros and ones. The first construction is a general recursive construction, which forms the sequences from a class of "elementary" sequences. The second construction is a special construction which can be used when the ratio of ones to zeros is expressed in binary notation. The second construction is shown to be optimal in terms of the numbers of input sequences required to construct the desired sequence. The paper concludes with a discussion of how to generate independent "elementary" sequences using simple digital techniques.>

Some recent work [7, 14, 15] in computational learning theory has discussed learning in situations where the teacher is helpful, and can choose to present carefully chosen sequences of labelled examples to the learner. We say a function t in a set H of functions (a hypothesis space) defined on a set X is specified by S***X if the only function in H which agrees with t on S is t itself. The specification number σ(t) of t is the least cardinality of such an S. For a general hypothesis space, we show that the specification number of any hypotheis is at least equal to a parameter from [14] known as the testing dimension of H. We investigate in some detail the specification numbers of hypotheses in the set Hn of linearly separable boolean functions: We present general methods for finding upper bounds on σ(t) and we characterise those t which have largest σ(t). We obtain a general lower bound on the number of examples required and we show that for all nested hypotheses, this lower bound is attained. We prove that for any t ε Hn, there is exactly one set of examples of minimal cardinality (i.e., of cardinality σ(t)) which specifies t. We then discuss those t ε Hn which have limited dependence, in the sense that some of the variables are redundant (i.e., there are irrelevant attributes), giving tight upper and lower bounds on σ(t) for such hypotheses. In the final section of the paper, we address the complexity of computing specification numbers and related parameters.