In constraint-based local search, an assignment to the search variables is improved upon by an iterative procedure that replaces the current assignment with a similar assignment. The latter is selected by a heuristic that assesses the qualities of a subset of all similar assignments, where the quality of such assignments is determined via a process called invariant graph propagation. Since, typically, many similar assignments are considered in every iteration, invariant graph propagation must be as efficient as possible. Since invariant graph propagation is independent of the selection heuristic, any comparison between different invariant graph propagation styles under different selection heuristics can be misleading. In this paper, we describe and compare both theoretically and empirically the throughput of several invariant graph propagation styles, and give criteria when one style or another is to be used.

In large neighbourhood search (LNS), an incumbent initial solution is incrementally improved by selecting a subset of the variables, called the freeze set, and fixing them to their values in the incumbent solution, while a value for each remaining variable is found and assigned via solving (such as constraint programming-style propagation and search). Much research has been performed on finding generic and problem-specific LNS selection heuristics that select freeze sets that lead to high-quality solutions. In constraint-based local search (CBLS), the relations between the variables via the constraints are fundamental and well-studied, as they capture dependencies of the variables. In this paper, we apply these ideas from CBLS to the LNS context, presenting the novel dependency curation scheme, which exploits them to find a low-cardinality set of variables that the freeze set of any selection heuristic should be a subset of. The scheme often improves the overall performance of generic selection heuristics. Even when the scheme is used with a naive generic selection heuristic that selects random freeze sets, the performance is competitive with more elaborate generic selection heuristics.

The optimization of sawing processes in the wood industry is critical for maximizing efficiency and profitability. The introduction of computerized tomography scanners provides sawmill operators with three-dimensional internal models of logs, which can be used to assess value and yield more accurately. We present a methodology for solving the sawing optimization problem employing a flexible sawing scheme that allows greater flexibility in cutting logs into products while considering product quality classes influenced by wane defects. The methodology has two phases: preprocessing and optimization. In the preprocessing phase, two alternative algorithms are given that generate and evaluate the potential sawing positions of products by considering the 3D surface of the log, product size requirements, and product quality classes. In the optimization phase, a maximum set-packing problem is solved for the preprocessed data using mixed-integer programming (MIP), aiming to obtain a feasible cut pattern that maximizes value yield. This is implemented in a system named FlexSaw, which takes advantage of parallel computation during the preprocessing phase and utilizes a MIP solver during the optimization phase. The proposed sawing methods are evaluated on the Swedish Pine Stem Bank. Additionally, FlexSaw is compared with an existing tool that utilizes cant sawing. Results demonstrate the superiority of flexible sawing. While the practical feasibility of implementing a flexible way of sawing logs is constrained by the limitations of current sawmill machinery, the potential increase in yield promotes the exploration of alternative machinery in the wood industry.

Constraint programming is a widely used paradigm to solve combinatorial problems. High-level constraint modelling languages, such as MiniZinc, GAMS, OPL, and AMPL, encourage the separation of instance parameters and models, where the number of variables and constraints depend on values of instance parameters. This paper investigates an extension of treewidth suitable for studying the complexity high-level constraint models.

Finite automata with counters are often used to specify families of constraints. The addition of counters provides more power than membership in regular languages that is possible with finite automata. All available propagation algorithms for counter automata based constraints maintain only bounds consistency on counter variables, and although it is possible to maintain domain consistency it can be computationally very costly. In this paper we give an algorithm that decides when maintaining bounds consistency for an automata with a single counter implies domain consistency.

Large-neighbourhood search (LNS) improves an initial solution, hence it is not directly applicable to satisfaction problems. In order to use LNS in a constraint programming (CP) framework to solve satisfaction problems, we usually soften some hard-to-satisfy constraints by replacing them with penalty-function constraints. LNS is then used to reduce their penalty to zero, thus satisfying the original problem. However, this can give poor performance as the penalties rarely cause propagation and therefore do not drive each CP search, and by extension the LNS search, towards satisfying the replaced constraints until very late. Our key observation is that entirely replacing a constraint is often overkill, as the propagator for the replaced constraint could have performed some propagation without causing backtracking. We propose the notion of a non-failing propagator, which is subsumed just before causing a backtrack. We show that, by only making a few changes to an existing CP solver, any propagator can be made non-failing without modifying its code. Experimental evaluation shows that non-failing propagators, when used in conjunction with penalties, can greatly improve LNS performance compared to just having penalties. This allows us to leverage the power of the many sophisticated propagators that already exist in CP solvers, in order to use LNS for solving hard satisfaction problems and for finding initial solutions to hard-to-satisfy optimisation problems.

Using constraint programming (CP) to explore a local-search neighbourhood was first tried in the mid 1990s. The advantage is that constraint propagation can quickly rule out uninteresting neighbours, sometimes greatly reducing the number actually probed. However, a CP model of the neighbourhood has to be handcrafted from the model of the problem: this can be difficult and tedious. That research direction appears abandoned since large-neighbourhood search (LNS) and constraint-based local search (CBLS) arose as alternatives that seem easier to use. Recently, the notion of declarative neighbourhood was added to the technology-independent modelling language MiniZinc, for use by any backend to MiniZinc, but currently only used by a CBLS backend. We demonstrate that declarative neighbourhoods are indeed technology-independent by using the old idea of CP-based neighbourhood exploration: we explain how to encode automatically a declarative neighbourhood into a CP model of the neighbourhood. This enables us to lift any CP solver into a local-search backend to MiniZinc. Our prototype is competitive with CP, CBLS, and LNS backends to MiniZinc.

A black-box local-search backend to a solving-technology-independent modelling language, such as MiniZinc, automatically infers from the structure of a declarative model for a satisfaction or optimisation problem a combination of a neighbourhood, heuristic, and meta-heuristic. These ingredients are then provided to a local-search solver, but are manually designed in a handcrafted local-search algorithm. However, such a backend can perform poorly due to model structure that is inappropriate for local search, for example when it considers moves modifying only variables that represent auxiliary information. Towards overcoming such inefficiency, we propose compound-move generation, an extension to local-search solvers that uses a complete-search solver in order to augment moves modifying non-auxiliary variables so that they also modify auxiliary ones. Since compound-move generation is intended to be applied to such models, we discuss how to identify them. We present several refinements of compound-move generation and show its very positive impact on several third-party models. This helps reduce the unavoidable gap between black-box local search and local-search algorithms crafted by experts.

Author(s): Pearson, Justin | Advisor(s): Hespanha, Joao P | Abstract: The performance of a control system is often limited by constraints on timing, bandwidth, and energy. This dissertation explores the trade-offs between constraints on these resources, the control system performance, and the system to be controlled.We begin by considering a networked control system in which the sensor sends its measurements to the controller over a limited-bandwidth communications channel. We explore the observation that the absence of communication nevertheless conveys information --- i.e., nothing communication-worthy occurred. This suggests that energy (or other resources consumed by communication) could be saved using the timing of messages to transmit information, rather than the normal practice of transmitting data in the messages themselves. We develop a framework to explore this idea and derive a condition for the existence of a stabilizing controller that captures the trade-off between bandwidth, resource consumption, and the unstable eigenvalues of the linear system to be controlled. A surprising result is that if this condition is satisfied, then one may design a stabilizing controller that consumes resources at an arbitrarily small rate, provided one has access to a sufficiently precise clock. In an extreme example, a large amount of data is encoded into the precise transmission time of a single bit, and the receiver decodes this data from the time the bit is received. This result quantifies the trade-off between bandwidth and time as resources for transmitting information.Next, we use our framework to analyze a family of event-based controllers. We show that these controllers can stabilize a system while consuming resources at a rate that is within 2.5 times the theoretically-minimum rate. These event-based controllers are intuitive and easy to implement, and our stability condition quantifies the cost (in additional required communication resources) that a control engineer pays for the convenience of implementing an event-based controller instead of the relatively more complicated controllers from the first section that use the theoretically-minimum communication rate.A takeaway from these results is that networked and distributed control systems can benefit from precise timing. However, even non-networked systems can benefit from precise timing. We explore this by developing a control architecture that allows a controller running on a non-real-time operating system to run with a high degree of determinacy, even when the OS task scheduler suspends the control task. The architecture employs a small microprocessor to be used as a real-time processor that runs independently from the OS and buffers sensor measurement and actuator commands. We implement this on a Beaglebone Black single-board computer and demonstrate that this architecture can significantly improve a controller's performance in the presence of OS preemption.

The aim of solver-independent modelling is to create a model of a satisfaction or optimisation problem independent of a particular technology. This avoids early commitment to a solving technology and allows easy comparison of technologies. MiniZinc is a solver-independent modelling language, supported by CP, MIP, SAT, SMT, and constraint-based local search (CBLS) backends. Some technologies, in particular CP and CBLS, require not only a model but also a search strategy. While backends for these technologies offer default search strategies, it is often beneficial to include in a model a user-specified search strategy for a particular technology, especially if the strategy can encapsulate knowledge about the problem structure. This is complex since a local-search strategy (comprising a neighbourhood, a heuristic, and a meta-heuristic) is often tightly tied to the model. Hence we wish to use the same language for specifying the model and the local search. We show how to extend MiniZinc so that one can attach a fully declarative neighbourhood specification to a model, while maintaining the solver-independence of the language. We explain how to integrate a model-specific declarative neighbourhood with an existing CBLS backend for MiniZinc.

An important task of testing a telecommunication protocol consists in analysing logs. The goal of log analysis is to check that the timing and the content of transmitted messages comply with specification. In order to perform such checks, protocols can be described using a constraint modelling language. In this paper we focus on a complex protocol where some messages can be delayed. Simply introducing variables for possible delays for all messages in the constraint model can drastically increase the complexity of the problem. However, some delays can be calculated, but this calculation is difficult to do by hand and to justify. We present an industrial application of the Coq proof assistant to prove a property of a 4G protocol and validate a constraint model. By using interactive theorem proving we derived constraints for message delays of the protocol and found missing constraints in the initial model.

Strings are extensively used in modern programming languages and constraints over strings of unknown length occur in a wide range of real-world applications such as software analysis and verification, testing, model checking, and web security. Nevertheless, practically no constraint programming solver natively supports string constraints. We introduce string variables and a suitable set of string constraints as builtin features of the MiniZinc modelling language. Furthermore, we define an interpreter for converting a MiniZinc model with strings into a FlatZinc instance relying only on integer variables. This conversion is obtained via rewrite rules, and does not require any extension of the existing FlatZinc specification. This provides a user-friendly interface for modelling combinatorial problems with strings, and enables both string and non-string solvers to actually solve such problems.

Integer time series are often subject to constraints on the aggregation of the features of all occurrences of some pattern within the series. For example, the number of inflexions may be constrained, or the sum of the peak maxima, or the minimum of the valley widths. Many time-series constraints can be described by transducers. The output alphabet of such a transducer consists of symbols that denote the phases of identifying the maximal occurrences of a pattern. It was recently shown how to synthesise automatically a constraint propagator and a constraint checker from such a transducer, which however has to be designed manually from a pattern. Here we define a large class of patterns, present an algorithm for automatically generating a low-level transducer from such a high-level pattern, and prove it correct. This class covers all 20 patterns of the Time-Series Constraint Catalogue, which can now be automatically extended at will.

Graphical user interfaces are important components of today's software. User interfaces often require checking correctness of user interactions. In web applications such checks can be a part of the JavaScript code. User interfaces in web applications can evolve, some elements can be removed and new elements can be added. To check JavaScript code covers all possible incorrect scenarios in user interactions in web application, constraint programming is used. We use the MiniZinc constraint modelling language to model incorrect user behaviour and to convert JavaScript code into a constraint model. Then we perform an equivalence check to find deviations in JavaScript code. The approach was applied to design user interface of an industrial software product.

We present the design and implementation of bounded length sequence (BLS) variables for a CP solver. The domain of a BLS variable is represented as the combination of a set of candidate lengths and a sequence of sets of candidate characters. We show how this representation, together with requirements imposed by propagators, affects the implementation of BLS variables for a copying CP solver, most importantly the closely related decisions of data structure, domain restriction operations, and propagation events. The resulting implementation outperforms traditional bounded-length string representations for CP solvers, which use a fixed-length array of candidate characters and a padding symbol.

First this report presents a restricted set of finite transducers used to synthesise structural time-series constraints described by means of a multi-layered function composition scheme. Second it provides the corresponding synthesised catalogue of structural time-series constraints where each constraint is explicitly described in terms of automata with registers.

A checker for a constraint on a variable sequence can often be compactly specified by an automaton, possibly with accumulators, that consumes the sequence of values taken by the variables; such an automaton can also be used to decompose its specified constraint into a conjunction of logical constraints. The inference achieved by this decomposition in a CP solver can be boosted by automatically generated implied constraints on the accumulators, provided the latter are updated in the automaton transitions by linear expressions. Automata with non-linear accumulator updates can be automatically synthesised for a large family of time-series constraints. In this paper, we describe and evaluate extensions to those techniques. First, we improve the automaton synthesis to generate automata with fewer accumulators. Second, we decompose a constraint specified by an automaton with accumulators into a conjunction of linear inequalities, for use by a MIP solver. Third, we generalise the implied constraint generation to cover the entire family of time-series constraints. The newly synthesised automata for time-series constraints outperform the old ones, for both the CP and MIP decompositions, and the generated implied constraints boost the inference, again for both the CP and MIP decompositions. We evaluate CP and MIP solvers on a prototypical application modelled using time-series constraints.

Integer time series are often subject to constraints on the aggregation of the integer features of all occurrences of some pattern within the series. For example, the number of inflexions may be constrained, or the sum of the peak maxima, or the minimum of the peak widths. It is currently unknown how to maintain domain consistency efficiently on such constraints. We propose parametric ways of systematically deriving glue constraints, which are a particular kind of implied constraints, as well as aggregation bounds that can be added to the decomposition of time-series constraints [5]. We evaluate the beneficial propagation impact of the derived implied constraints and bounds, both alone and together.