arXiv subjects

Maciej Bendkowski

Publications and source records attributed to Maciej Bendkowski.

14 recordsLinked to original sources

arXiv2022

Automatic compile-time synthesis of entropy-optimal Boltzmann samplers

We present a famework for the automatic compilation of multi-parametric Boltzmann samplers for algebraic data types in Haskell. Our framework uses Template Haskell to synthesise efficient, entropy-optimal samplers generating random instances of user-declared algebraic data types. Users can control the outcome distribution through a pure, declarative interface. For instance, users can control the mean size and constructor frequencies of generated objects. We illustrate the effectiveness of our framework through a prototype generic-boltzmann-brain library showing that it is possible to control thousands of different parameters in systems of tens of thousands of ADTs. Our prototype framework synthesises Boltzmann samplers capable of rapidly generating random objects of sizes in the millions.

cs.SE↗

arXiv2021

Tuning as convex optimisation: a polynomial tuner for multi-parametric combinatorial samplers

Combinatorial samplers are algorithmic schemes devised for the approximate- and exact-size generation of large random combinatorial structures, such as context-free words, various tree-like data structures, maps, tilings, RNA molecules. They can be adapted to combinatorial specifications with additional parameters, allowing for a more flexible control over the output profile of parametrised combinatorial patterns. One can control, for instance, the number of leaves, profile of node degrees in trees or the number of certain sub-patterns in generated strings. However, such a flexible control requires an additional and nontrivial tuning procedure. Using techniques of convex optimisation, we present an efficient tuning algorithm for multi-parametric combinatorial specifications. Our algorithm works in polynomial time in the system description length, the number of tuning parameters, the number of combinatorial classes in the specification, and the logarithm of the total target size. We demonstrate the effectiveness of our method on a series of practical examples, including rational, algebraic, and so-called Pólya specifications. We show how our method can be adapted to a broad range of less typical combinatorial constructions, including symmetric polynomials, labelled sets and cycles with cardinality lower bounds, simple increasing trees or substitutions. Finally, we discuss some practical aspects of our prototype tuner implementation and provide its benchmark results.

math.CO↗

arXiv2021

A note on the asymptotic expressiveness of ZF and ZFC

We investigate the asymptotic densities of theorems provable in Zermelo-Fraenkel set theory ZF and its extension ZFC including the axiom of choice. Assuming a canonical De Bruijn representation of formulae, we construct asymptotically large sets of sentences unprovable within ZF, yet provable in ZFC. Furthermore, we link the asymptotic density of ZFC theorems with the provable consistency of ZFC itself. Consequently, if ZFC is consistent, it is not possible to refute the existence of the asymptotic density of ZFC theorems within ZFC. Both these results address a recent question by Zaionc regarding the asymptotic equivalence of ZF and ZFC.

math.LO↗

arXiv2020

How to generate random lambda terms?

We survey several methods of generating large random lambda-terms, focusing on their closed and simply-typed variants. We discuss methods of exact- and approximate-size generation, as well as methods of achieving size-uniform and non-uniform outcome distributions.

math.CO↗

arXiv2019

On the enumeration of closures and environments with an application to random generation

Environments and closures are two of the main ingredients of evaluation in lambda-calculus. A closure is a pair consisting of a lambda-term and an environment, whereas an environment is a list of lambda-terms assigned to free variables. In this paper we investigate some dynamic aspects of evaluation in lambda-calculus considering the quantitative, combinatorial properties of environments and closures. Focusing on two classes of environments and closures, namely the so-called plain and closed ones, we consider the problem of their asymptotic counting and effective random generation. We provide an asymptotic approximation of the number of both plain environments and closures of size $n$. Using the associated generating functions, we construct effective samplers for both classes of combinatorial structures. Finally, we discuss the related problem of asymptotic counting and random generation of closed environemnts and closures.

cs.LO↗

arXiv2018

Towards the average-case analysis of substitution resolution in $λ$-calculus

Substitution resolution supports the computational character of $β$-reduction, complementing its execution with a capture-avoiding exchange of terms for bound variables. Alas, the meta-level definition of substitution, masking a non-trivial computation, turns $β$-reduction into an atomic rewriting rule, despite its varying operational complexity. In the current paper we propose a somewhat indirect average-case analysis of substitution resolution in the classic $λ$-calculus, based on the quantitative analysis of substitution in $λ\upsilon$, an extension of $λ$-calculus internalising the $\upsilon$-calculus of explicit substitutions. Within this framework, we show that for any fixed $n \geq 0$, the probability that a uniformly random, conditioned on size, $λ\upsilon$-term $\upsilon$-normalises in $n$ normal-order (i.e. leftmost-outermost) reduction steps tends to a computable limit as the term size tends to infinity. For that purpose, we establish an effective hierarchy $\left(\mathscr{G}_n\right)_n$ of regular tree grammars partitioning $\upsilon$-normalisable terms into classes of terms normalising in $n$ normal-order rewriting steps. The main technical ingredient in our construction is an inductive approach to the construction of $\mathscr{G}_{n+1}$ out of $\mathscr{G}_n$ based, in turn, on the algorithmic construction of finite intersection partitions, inspired by Robinson's unification algorithm. Finally, we briefly discuss applications of our approach to other term rewriting systems, focusing on two closely related formalisms, i.e. the full $λ\upsilon$-calculus and combinatory logic.

cs.LO↗

arXiv2018

Statistical properties of lambda terms

We present a quantitative, statistical analysis of random lambda terms in the de Bruijn notation. Following an analytic approach using multivariate generating functions, we investigate the distribution of various combinatorial parameters of random open and closed lambda terms, including the number of redexes, head abstractions, free variables or the de Bruijn index value profile. Moreover, we conduct an average-case complexity analysis of finding the leftmost-outermost redex in random lambda terms showing that it is on average constant. The main technical ingredient of our analysis is a novel method of dealing with combinatorial parameters inside certain infinite, algebraic systems of multivariate generating functions. Finally, we briefly discuss the random generation of lambda terms following a given skewed parameter distribution and provide empirical results regarding a series of more involved combinatorial parameters such as the number of open subterms and binding abstractions in closed lambda terms.

math.CO↗

arXiv2018

Combinatorics of explicit substitutions

$λ\upsilon$ is an extension of the $λ$-calculus which internalises the calculus of substitutions. In the current paper, we investigate the combinatorial properties of $λ\upsilon$ focusing on the quantitative aspects of substitution resolution. We exhibit an unexpected correspondence between the counting sequence for $λ\upsilon$-terms and famous Catalan numbers. As a by-product, we establish effective sampling schemes for random $λ\upsilon$-terms. We show that typical $λ\upsilon$-terms represent, in a strong sense, non-strict computations in the classic $λ$-calculus. Moreover, typically almost all substitutions are in fact suspended, i.e. unevaluated, under closures. Consequently, we argue that $λ\upsilon$ is an intrinsically non-strict calculus of explicit substitutions. Finally, we investigate the distribution of various redexes governing the substitution resolution in $λ\upsilon$ and investigate the quantitative contribution of various substitution primitives.

cs.LO↗

arXiv2017

Polynomial tuning of multiparametric combinatorial samplers

Boltzmann samplers and the recursive method are prominent algorithmic frameworks for the approximate-size and exact-size random generation of large combinatorial structures, such as maps, tilings, RNA sequences or various tree-like structures. In their multiparametric variants, these samplers allow to control the profile of expected values corresponding to multiple combinatorial parameters. One can control, for instance, the number of leaves, profile of node degrees in trees or the number of certain subpatterns in strings. However, such a flexible control requires an additional non-trivial tuning procedure. In this paper, we propose an efficient polynomial-time, with respect to the number of tuned parameters, tuning algorithm based on convex optimisation techniques. Finally, we illustrate the efficiency of our approach using several applications of rational, algebraic and Pólya structures including polyomino tilings with prescribed tile frequencies, planar trees with a given specific node degree distribution, and weighted partitions.

math.CO↗

arXiv2017

Random generation of closed simply-typed $λ$-terms: a synergy between logic programming and Boltzmann samplers

A natural approach to software quality assurance consists in writing unit tests securing programmer-declared code invariants. Throughout the literature a great body of work has been devoted to tools and techniques automating this labour-intensive process. A prominent example is the successful use of randomness, in particular random typeable $λ$-terms, in testing functional programming compilers such as the Glasgow Haskell Compiler. Unfortunately, due to the intrinsically difficult combinatorial structure of typeable $λ$-terms no effective uniform sampling method is known, setting it as a fundamental open problem in the random software testing approach. In this paper we combine the framework of Boltzmann samplers, a powerful technique of random combinatorial structure generation, with today's Prolog systems offering a synergy between logic variables, unification with occurs check and efficient backtracking. This allows us to develop a novel sampling mechanism able to construct uniformly random closed simply-typed $λ$-terms of up size 120. We apply our techniques to the generation of uniformly random closed simply-typed normal forms and design a parallel execution mechanism pushing forward the achievable term size to 140. Under consideration in Theory and Practice of Logic Programming (TPLP).

cs.LO↗

arXiv2016

Combinatorics of $λ$-terms: a natural approach

We consider combinatorial aspects of $λ$-terms in the model based on de Bruijn indices where each building constructor is of size one. Surprisingly, the counting sequence for $λ$-terms corresponds also to two families of binary trees, namely black-white trees and zigzag-free ones. We provide a constructive proof of this fact by exhibiting appropriate bijections. Moreover, we identify the sequence of Motzkin numbers with the counting sequence for neutral $λ$-terms, giving a bijection which, in consequence, results in an exact-size sampler for the latter based on the exact-size sampler for Motzkin trees of Bodini et alli. Using the powerful theory of analytic combinatorics, we state several results concerning the asymptotic growth rate of $λ$-terms in neutral, normal, and head normal forms. Finally, we investigate the asymptotic density of $λ$-terms containing arbitrary fixed subterms showing that, inter alia, strongly normalising or typeable terms are asymptotically negligible in the set of all $λ$-terms.

cs.LO↗

arXiv2016

On the likelihood of normalisation in combinatory logic

We present a quantitative basis-independent analysis of combinatory logic. Using a general argument regarding plane binary trees with labelled leaves, we generalise the results of David et al. and Bendkowski et al. to all Turing-complete combinator bases proving, inter alia, that asymptotically almost no combinator is strongly normalising nor typeable. We exploit the structure of recently discovered normal-order reduction grammars showing that for each positive $n$, the set of $\mathbf{S} \mathbf{K}$-combinators reducing in $n$ normal-order reduction steps has positive asymptotic density in the set of all combinators. Our approach is constructive, allowing us to systematically find new asymptotically significant fractions of normalising combinators. We show that the density of normalising combinators cannot be less than $34\%$, improving the previously best lower bound of approximately $3\%$. Finally, we present some super-computer experimental results, conjecturing that the density of normalising combinators is close to $85\%$.

cs.LO↗

arXiv2016

Normal-order reduction grammars

We present an algorithm which, for given $n$, generates an unambiguous regular tree grammar defining the set of combinatory logic terms, over the set $\{S,K\}$ of primitive combinators, requiring exactly $n$ normal-order reduction steps to normalize. As a consequence of Curry and Feys's standardization theorem, our reduction grammars form a complete syntactic characterization of normalizing combinatory logic terms. Using them, we provide a recursive method of constructing ordinary generating functions counting the number of $S K$-combinators reducing in $n$ normal-order reduction steps. Finally, we investigate the size of generated grammars, giving a primitive recursive upper bound.

cs.LO↗

arXiv2016

A natural counting of lambda terms

We study the sequences of numbers corresponding to lambda terms of given sizes, where the size is this of lambda terms with de Bruijn indices in a very natural model where all the operators have size 1. For plain lambda terms, the sequence corresponds to two families of binary trees for which we exhibit bijections. We study also the distribution of normal forms, head normal forms and strongly normalizing terms. In particular we show that strongly normalizing terms are of density 0 among plain terms.

cs.LO↗