Lab for Automated Reasoning and Analysis LARA

Available Student Projects in LARA - EPFL Internal Page

© 2016 Laboratory for Automated Reasoning and Analysis. All Rights Reserved. Any reproduction of the parts of this page in printed or electronic form requires a prior written permission from Viktor Kuncak.

These projects are suitable for both BSc and MSc projects (including master's thesis projects) for students in the IC school of EPFL, and are suitable for both computer science and communication systems students. Each project is for one student in a given semester. Students report directly to Prof. Viktor Kuncak, even if PhD students are involved to help with various aspects of the project. The information on whether the project is available or not is also maintained only by Viktor Kuncak, so please contact him if you are interested in the project.

Profile for the candidates

Laboratory for Automated Reasoning and Analysis has BSc and MSc projects available (for PhD program see ). We are looking for students with high grades that are excellent in advanced functional programming and ideally strong in some of the core mathematical methods such as discrete mathematics, numerical analysis, logic, and probabilistic reasoning.

Nature of these projects

Below we list several projects reflecting the current interests of our research group. All projects can be adapted to fit the scope of a semester project (bachelor & master) or master's project.

Many of these project are based on the Leon system for synthesis and verification. We suggest that you get the basic idea of Leon in any case.

Licence for code: Please note that we release software under open source licenses and we request our contributors to agree that EPFL LARA holds copyright of the code that you contribute.

Disclaimer: If you have your own idea, feel free to check with us, but please note that we some times cannot accommodate your own project suggestions even if the student is brilliant and the project is excellent. Moreover, even if you select one of the projects listed below, we reserve the right to not offer you the project. By deciding to talk to us you agree that you will not require detailed justification on why we made our negative decision.

WebAssembly Generation Library for Scala

Write a library for generating WebAssembly code, to be used in Computer Language Processing course as well as for other projects in the community. This library might replace the Cafebabe library:

Robust Front End for Leon Compiled Using scala.js

The goal of this task is to design a reasonably fast front end for the subset of Scala that is relevant for Leon. The front-end should be able to process existing regression test suite for Leon, but will necessarily deviate from scalac or dotty on some examples. In particular, the type system can be simpler and implicit resolution can be more restrictive than in Scala. Optionally, the front end may support omitting type declarations even for certain function parameters, if these can be inferred given a whole program using type inference for parametric types with simple structural subtyping.

Verifying JavaScript

Design a translation of a clean subset of JavaScript including higher-order functions and (mostly immutable) objects into Leon. Map numbers into Double (apply overflow checks as needed), and enable annotations or special function calls to describe types and contracts.

Note that `(f:TObj) ⇒ e` can be encoded as `TObj(x ⇒ e)`.

AST-Based Version Control

Develop a layer on top of git for managing revisions of Leon programs at the level of high-level syntax tree transformations, with conflict resolution doing semantic commutativity checks.

Easily editable and exportable hand-writing presentation videos

Do you know Wandida videos? These famous videos explain concepts using natural drawing, pictures, voice. Although very popular, it is hard to produce them. Their greatest limitation is that they are hard to modify and long to produce, with a lot of post-processing. To simplify this process, we will adopt a vectorized format more suitable to integration with other presentation elements (inline quizz, textual slides, text-to-speech, embedding, hypertext linking, etc.) The student will have define this format, create sample explanation videos, and possibly adapt to let users produce such videos very quickly. As a long-term result this project will enable the creation of much more such videos which are very appreciated by students.

Reversing Embedded Domain Specific Languages (DSL)

To make programming easier, people use languages adapted to a particular domain, in order to write less. For example, when creating web pages from Scala code, one can start to define DSL so that instead of writing `Element(“div”, List(TextElement(“hello ”+“world”)))`, one can simplify write `<.div(“hello ”+“world”)`. We recently wrote an engine which, given a modification of text *on the web page* (ex: hello to bonjour), is able to modify the string from the original expression, whatever the way it was built <.div(“bonjour ”+“world”), even asking questions in the case of ambiguities. The student would push this feature towards having full HTML tree edition, but at the same time reconstructing the smallest DSL expression that can support the change in the output. As an immediate result one will be able to create and modify presentation using reveal.js.

Database filling in Programming by Example

We recently designed a model for modifying program-generated webpages so that it modifies the original program, yielding a very useful programming environnement where both the program and the output can be modified. However, we would like to extract data from the program (e.g; text, svg data, animations, audio…) so that the parsing becomes faster and the data is stored in a structured way. For that, the student will have to come up with creative ways to 1) Store such data in the server 2) Retrieve it from a program 3) Modify this data when the output changes 4) Modify the *structure* of the data if the program change. As an immediate result one will be able to record audio in html slides, or add SVG animations.

Verified Parameterized-Length Bitvectors in Leon

Define and prove correct a library for performing arithmetic (+, -, *, /, %) on modular arithmetic with large integers. Such operations are analogous to Int in that they perform modular arithmetic (unlike the unbounded BigInt), but the number of bits can be user-specified (unlike Int).

Verifying Eliptic Curves using Leon

Verify correctness of implementing group addition operations on elements on elliptic and Edwards curves over certain finite fields (such as those over the prime field of size 2^255-19). See

This project likely requires the previous project on parameterized-length bitvectors.

Verify Smart Contract Code

Verify correctness of electronic contract code. See, for example:

Verify Security Protocol

Propose a way to model security protocols in Leon and perform a case study to verify example protocols.

Forgiving Parser for Leon

Develop a stand-alone parser for Leon programs, which aims to accept almost any sequence of characters as input and interpret it to the best of its knowledge as a Leon program. Optionally also allow syntax supported in other programming languages. A requirement is that a valid Leon program should be interpreted in the same way as in the current front end.

More information:

Libraries and Applications in Leon

Develop and specify additional applications and/or libraries for Leon system, along with their specifications. Examples of applications include parsing algorithms, compilers, email clients, web applications (e.g. a news portal). Examples of libraries include data structures, linear algebra libraries, user interface libraries, as well as sound and graphics processing libraries.

Among possible directions is to use to generate Javascript from Leon. To make such code useful, explore building Leon libraries that interface to Javascript APIs (focusing to the extent possible on stateless APIs) and build Leon specifications and stubs that approximate the behavior of these APIs to enable verification and synthesis.

Improve Leon Web-Based IDE

Develop new IDE functionality and interactive program transformations for Leon. The transformations should subsume refactorings present in e.g. Eclipse IDE, but should also make use of Leon's ability to check correctness conditions to ensure soundness of transformations.

Statistically Biased Code Generation

Develop statistical models for a class of useful Leon programs and use them to build code completion functionality as well as to improve deductive synthesis of Leon.

For more information, please check the notion of a Java language model from the report On Synthesizing Code from Free-Form Queries, and also consult the notion of statistical language model from natural language translation literature (e.g. ).

C Back-End for a Leon Subset

Advance our GenC back for Leon, which generates C code from a subset of Leon, and evaluate it on case studies for embedded code, IoT code, or fragments of OS code functinality.

Improve Our General-Purpose Grammar Parsing Library and

For teaching and research we are developing a library that can parse arbitrary context-free grammars, used in computer language processing course for exercises and the project, and in the future to be used in research. It is currently primarily based on generalizations of CYK algorithm as well as LL(1). Your task would be to introduce GLL, GLR, and Packrat parsing algorithms into this framework. Thanks to deep embedding, our library enables pre-analysis of grammars including the computation of first and follow sets.

Another project in this scope is to further advance the web service.

A version of Scala combinator library interface built on top of this library is also of interest, see e.g. ( ).

Contact us

If one of the projects lists sounds interesting and you believe you are a suitable candide, please do not hesitate to contact Viktor Kuncak, via email to make an appointment. In your email, please include your CV, the list of your grades, and a description of projects that you did at EPFL or outside of EPFL.

Feel free to also talk to the other members of LARA group about their ongoing work and projects, see

projects.txt · Last modified: 2017/04/11 11:43 by vkuncak
© EPFL 2018 - Legal notice