Synthesizing data structure transformations from input-output examples

Synthesizing intended programs from user-specified input-output examples, also known as Programming by Examples (PBE), is a challenging problem in program synthesis, and has been applied to a wide range of domains. A key challenge in PBE is to efficiently discover a user-intended program in the search space that can be exponentially large. In this work, we propose a method for automatic synthesis of functional programs on list manipulation, by using offline-trained Recurrent Neural Network (RNN) models to guide the program search. We adopt miniKanren, an embedded domain-specific language for flexible relational programming, as an underlying top-down deductive search engine of candidate programs that are consistent with input-output examples. Our approach targets an easy and effective integration of deep learning techniques in making better PBE systems and combines two technical ideas on generating diverse training dataset and designing rich feature embeddings of probable subproblems for synthesis generated by deductive search. The offline-learned model is then used in PBE to guide the top-down deductive search with specific strategies. To practically manipulate data structures of lists, our method synthesizes functional programs with popular higher-order combinators including $\texttt {map}$ , $\texttt {foldl}$ and $\texttt {foldr}$ . We have implemented our method and evaluated it with challenging program synthesis tasks on list manipulation. The experiments show promising results on the performance of our method compared to related state-of-the-art inductive synthesizers.

Program Synthesis, which is the task of discovering programs that realize user intent, can be useful in several scenarios: enabling people with no programming background to develop utility programs, helping regular programmers automatically discover tricky/mundane details, program understanding, discovery of new algorithms, and even teaching.

We present a new general-purpose synthesis technique for generating programs from input-output examples. Our method, called metric program synthesis, relaxes the observational equivalence idea (used widely in bottom-up enumerative synthesis) into a weaker notion of observational similarity, with the goal of reducing the search space that the synthesizer needs to explore. Our method clusters programs into equivalence classes based on an expert-provided distance metric and constructs a version space that compactly represents “approximately correct” programs. Then, given a “close enough” program sampled from this version space, our approach uses a distance-guided repair algorithm to find a program that exactly matches the given input-output examples. We have implemented our proposed metric program synthesis technique in a tool called SyMetric and evaluate it in three different domains considered in prior work. Our evaluation shows that SyMetric outperforms other domain-agnostic synthesizers that use observational equivalence and that it achieves results competitive with domain-specific synthesizers that are either designed for or trained on those domains.

We present a new framework and associated synthesis algorithms for program synthesis over noisy data, i.e., data that may contain incorrect/corrupted input-output examples. This framework is based on an extension of finite tree automata called {\em weighted finite tree automata}. We show how to apply this framework to formulate and solve a variety of program synthesis problems over noisy data. Results from our implemented system running on problems from the SyGuS 2018 benchmark suite highlight its ability to successfully synthesize programs in the face of noisy data sets, including the ability to synthesize a correct program even when every input-output example in the data set is corrupted.

This paper presents an algorithm for synthesizing recursive functions that process algebraic datatypes. It is founded on proof-theoretic techniques that exploit both type information and input–output examples to prune the search space. The algorithm uses refinement trees, a data structure that succinctly represents constraints on the shape of generated code. We evaluate the algorithm by using a prototype implementation to synthesize more than 40 benchmarks and several non-trivial larger examples. Our results demonstrate that the approach meets or outperforms the state-of-the-art for this domain, in terms of synthesis time or attainable size of the generated programs.

In this thesis, we explore techniques for the development and verification of programs in a high-level, expressive, and safe programming language. Our programs can express problems over unbounded domains and over recursive and mutable data structures. We present an implementation language flexible enough to build interesting and useful systems. We mostly maintain a core shared language for the specifications and the implementation, with only a few extensions specific to expressing the specifications. Extensions of the core shared language include imperative features with state and side effects, which help when implementing efficient systems. Our language is a subset of the Scala programming language. Once verified, programs can be compiled and executed using the existing Scala tools. We present algorithms for verifying programs written in this language. We take a layer-based approach, where we reduce, at each step, the program to an equivalent program in a simpler language. We first purify functions by transforming away mutations into explicit return types in the functions' signatures. This step rewrites all mutations of data structures into cloning operations. We then translate local state into a purely functional code, hence eliminating all traces of imperative programming. The final language is a functional subset of Scala, on which we apply verification. We integrate our pipeline of translations into Leon, a verifier for Scala. We verify the core functional language by using an algorithm already developed inside Leon. The program is encoded into equivalent first-order logic formulas over a combination of theories and recursive functions. The formulas are eventually discharged to an external SMT solver. We extend this core language and the solving algorithm with support for both infinite-precision integers and bit-vectors. The algorithm takes into account the semantics gap between the two domains, and the programmer is ultimately responsible to use the proper type to represent the data. We build a reusable interface for SMT-LIB that enables us to swap solvers transparently in order to validate the formulas emitted by Leon. We experiment with writing solvers in Scala; they could offer both a better and safer integration with the rest of the system. We evaluate the cost of using a higher-order language to implement such solvers, traditionally written in C/C++. Finally, we experiment with the system by building fully working and verified applications. We rely on the intersection of many features including higher-order functions, mutable data structures, recursive functions, and nondeterministic environment dependencies, to build concise and verified applications.

We look at program synthesis where the aim is to automatically synthesize a controller that operates on data structures and from which a concrete program can be easily derived. We do not aim at a fully-automatic process or tool that produces a program meeting a given specification of the program’s behaviour. Rather, we aim at the design of a clear and well-founded approach for supporting programmers at the design and implementation phases. Concretely, we first show that a program synthesis task can be modeled as a generalized planning problem. This is done at an abstraction level where the involved data structures are seen as black-boxes that can be interfaced with actions and observations, the first corresponding to the operations and the second to the queries provided by the data structure. The abstraction level is high enough to capture intuitive and common assumptions as well as general and simple strategies used by programmers, and yet it contains sufficient structure to support the automated generation of concrete solutions (in the form of controllers). From such controllers and the use of standard data structures, an actual program in a general language like C++ or Python can be easily obtained. Then, we discuss how the resulting generalized planning problem can be reduced to an LTL synthesis problem, thus making available any LTL synthesis engine for obtaining the controllers. We illustrate the effectiveness of the approach on a series of examples.

The ability to automatically discover a program consistent with a given user intent (specification) is the holy grail of Computer Science. While significant progress has been made on the so-called problem of Program Synthesis, a number of challenges remain; particularly for the case of synthesizing richer and larger programs. This is in large part due to the difficulty of search over the space of programs. In this paper, we argue that the above-mentioned challenge can be tackled by learning synthesizers automatically from a large amount of training data. We present a first step in this direction by describing our novel synthesis approach based on two neural architectures for tackling the two key challenges of Learning to understand partial input-output specifications and Learning to search programs. The first neural architecture called the Spec Encoder computes a continuous representation of the specification, whereas the second neural architecture called the Program Generator incrementally constructs programs in a hypothesis space that is conditioned by the specification vector. The key idea of the approach is to train these architectures using a large set of (spec,P) pairs, where P denotes a program sampled from the DSL L and spec denotes the corresponding specification satisfied by P. We demonstrate the effectiveness of our approach on two preliminary instantiations. The first instantiation, called Neural FlashFill, corresponds to the domain of string manipulation programs similar to that of FlashFill. The second domain considers string transformation programs consisting of composition of API functions. We show that a neural system is able to perform quite well in learning a large majority of programs from few input-output examples. We believe this new approach will not only dramatically expand the applicability and effectiveness of Program Synthesis, but also would lead to the coming together of the Program Synthesis and Machine Learning research disciplines.

Multimodal program synthesis, which leverages different types of user input to synthesize a desired program, is an attractive way to scale program synthesis to challenging settings; however, it requires integrating noisy signals from the user, like natural language, with hard constraints on the program's behavior. This paper proposes an optimal neural synthesis approach where the goal is to find a program that satisfies user-provided constraints while also maximizing the program's score with respect to a neural model. Specifically, we focus on multimodal synthesis tasks in which the user intent is expressed using a combination of natural language (NL) and input-output examples. At the core of our method is a top-down recurrent neural model that places distributions over abstract syntax trees conditioned on the NL input. This model not only allows for efficient search over the space of syntactically valid programs, but it allows us to leverage automated program analysis techniques for pruning the search space based on infeasibility of partial programs with respect to the user's constraints. The experimental results on a multimodal synthesis dataset (StructuredRegex) show that our method substantially outperforms prior state-of-the-art techniques in terms of accuracy and efficiency, and finds model-optimal programs more frequently.

Neural models excel at extracting statistical patterns from large amounts of data, but struggle to learn patterns or reason about language from only a few examples. In this work, we ask: Can we learn explicit rules that generalize well from only a few examples? We explore this question using program synthesis. We develop a synthesis model to learn phonology rules as programs in a domain-specific language. We test the ability of our models to generalize from few training examples using our new dataset of problems from the Linguistics Olympiad, a challenging set of tasks that require strong linguistic reasoning ability. In addition to being highly sample-efficient, our approach generates human-readable programs, and allows control over the generalizability of the learnt programs.

Inductive functional programming (IFP) is a research field extending from software science to artificial intelligence that deals with functional program synthesis based on generalization from ambiguous specifications, usually given as input-output example pairs. Currently, the approaches to IFP can be categorized into two general groups: the analytical approach that is based on analysis of the input-output example pairs, and the generate-and-test approach that is based on generation and testing of many candidate programs. The analytical approach shows greater promise for application to greater problems because the search space is restricted by the given example set, but it requires much more examples written in orderto yield results that reflect the user's intention, which is bothersome and causes the algorithm to slow down. On the other hand, the generate-and-test approach does not require long description of input-output examples, but does not restrict the search space using the example set. This paper proposes a new approach taking the best of the two, called "analytically-generate-and-test approach", which is based on analytical generation and testing of many program candidates. For generating many candidate programs, the proposed system uses a new variant of I GOR II, the exemplary analytical inductive functional programming algorithm. This new system preserves the efficiency features of analytical approaches, while minimizing the possibility of generating unintended programs even when using fewer input-output examples.

Automatic software generation based on some specification is known as program synthesis. Most existing approaches formulate program synthesis as a search problem with discrete parameters. In this paper, we present a novel formulation of program synthesis as a continuous optimization problem using an evolutionary approach, known as Covariance Matrix Adaptation Evolution Strategy. We then propose several mapping schemes to convert the continuous formulation into actual programs and propose different restart policies for the evolutionary approach. This is the first work that demonstrates the feasibility of continuous approach in synthesizing complex programs, not just simple toy programs. We compare our system, Genesys , to several recent program synthesis techniques (in both discrete and continuous domains). We find that Genesys synthesizes more programs within a fixed time budget than those existing schemes. For example, for programs of length 10, Genesys synthesizes 28% more programs than those existing schemes within the same time budget.

Context. The problem of automated development of evaluation programs for the neuroevolution of augmenting topologies. Neuroevolution algorithms apply mechanisms of mutation, recombination, and selection to find neural networks with behavior that satisfies the conditions of a certain formally defined problem. An example of such a problem is finding a neural network that implements a certain digital logic.
 Objective. The goal of the work is the automated design and generation of an evaluation program for a sample neuroevolution problem (binary multiplexer).
 Method. The methods and tools of Glushkov’s algebra of algorithms and hyperscheme algebra are applied for the parameterdriven generation of a neuroevolution evaluation program for a binary multiplexer. Glushkov’s algebra is the basis of the algorithmic language intended for multilevel structural design and documentation of sequential and parallel algorithms and programs in a form close to a natural language. Hyperschemes are high-level parameterized specifications intended for solving a certain class of problems. Setting parameter values and subsequent interpretation of hyperschemes allows obtaining algorithms adapted to specific conditions of their use.
 Results. The facilities of hyperschemes were implemented in the developed integrated toolkit for the automated design and synthesis of programs. Based on algorithm schemes, the system generates programs in a target programming language. The advantage of the system is the possibility of describing algorithm schemes in a natural-linguistic form. An experiment was conducted consisting in the execution of the generated program for the problem of evaluating a binary multiplexer on a distributed cloud platform. The multiplexer example is included in SharpNEAT, an open-source framework that implements the genetic neuroevolution algorithm NEAT for the .NET platform. The parallel distributed implementation of the SharpNEAT was proposed in the previous work of the authors.
 Conclusions. The conducted experiments demonstrated the possibility of the developed distributed system to perform evaluations on 64 cloud clients-executors and obtain an increase in 60–100% of the maximum capabilities of a single-processor local implementation.

We present an enumerative program synthesis framework called component-based refactoring that can refactor “direct” style code that does not use library components into equivalent “combinator” style code that does use library components. This framework introduces a sound but incomplete technique to check the equivalence of direct code and combinator code called equivalence by canonicalization that does not rely on input-output examples or logical specifications. Moreover, our approach can repurpose existing compiler optimizations, leveraging decades of research from the programming languages community. We instantiated our new synthesis framework in two contexts: (i) higher-order functional combinators such as map and filter in the staticallytyped functional programming language Elm and (ii) high-performance numerical computing combinators provided by the NumPy library for Python. We implemented both instantiations in a tool called Cobbler and evaluated it on thousands of real programs to test the performance of the component-based refactoring framework in terms of execution time and output quality. Our work offers evidence that synthesis-backed refactoring can apply across a range of domains without specification beyond the input program.

A key challenge in example-based program synthesis is the gigantic search space of programs. To address this challenge, various work proposed to use abstract interpretation to prune the search space. However, most of existing approaches have focused only on forward abstract interpretation, and thus cannot fully exploit the power of abstract interpretation. In this paper, we propose a novel approach to inductive program synthesis via iterative forward-backward abstract interpretation. The forward abstract interpretation computes possible outputs of a program given inputs, while the backward abstract interpretation computes possible inputs of a program given outputs. By iteratively performing the two abstract interpretations in an alternating fashion, we can effectively determine if any completion of each partial program as a candidate can satisfy the input-output examples. We apply our approach to a standard formulation, syntax-guided synthesis (SyGuS), thereby supporting a wide range of inductive synthesis tasks. We have implemented our approach and evaluated it on a set of benchmarks from the prior work. The experimental results show that our approach significantly outperforms the state-of-the-art approaches thanks to the sophisticated abstract interpretation techniques.