Faites connaître cette offre !
Reference : UMR5506-ANDROM-001
Workplace : MONTPELLIER
Date of publication : Thursday, November 07, 2019
Type of Contract : FTC Scientist
Contract Period : 3 months
Expected date of employment : 1 February 2020
Proportion of work : Full time
Remuneration : between 2 097,64 et 2 674,49 € dependning on the experience
Desired level of education : 5-year university degree
Experience required : Indifferent
Ingénieur d'études en informatique : a work on statistical tests of random number generators, and development of software improving the quality of the (pseudo)random bits produced by physical generators of randomness.
Generation of random bits is a classical problem known in the context of pseudo-random generators and also in the context of truly random physical processes (some electronic devices that produce random bits use an unpredictable physical noise or intrinsically nondeterministic quantum phenomena). However, the quality of the physical generators of the random bits remains badly founded and poorly tested. The first objective of this project is to study the validity and quality of several practical random number generators.
When we discuss the quality of random or pseudo-random generators, we have to use randomness tests, such as the open source Diehard and Dieharder tests, NIST tests, and so on. The second objective of the project is an inventory and revision of statistical tests for random and pseudo-random generators. Currently, some tests are used without a strong theoretical foundation. We suggest to improve the quality of statistical tests and develop new techniques of "whitening," which is used to improve the quality of non-ideal sources of random bits.
The proposed contract involves three types of activity:
A. Correcting errors in the current version of the open source battery of randomness test Dieharder.
B. Extension of the battery of the tests included in Dieharder. In particular, we suppose to add tests based on the Kolmogorov--Smirnov statistics for two independent samples.
C. Numerical experiments with physical generators of random numbers: evaluating the quality of these generators with help of the revised battery of tests Dieharder.
These jobs involve programming in C (revision of the existing code of Dieharder and writing new extensions for it). The contract does not involve much theoretical research, but the work requires understanding of the mathematical background behind the used statistical algorithms.
We are looking for a candidate with a fair knowledge of mathematics (probability theory, linear algebra from the standard university courses) and computer science, with a degree of Master or equivalent (PhD is not necessary). The candidate must have sufficient programming skills to use, debug, and improve the existing open source "randomness tests" (typically written in C). The knowledge of English is necessary; knowledge of French is an advantage.
The work will be carried out at the Laboratory of Computer Science, Robotics and Microelectronics of Montpellier (LIRMM, Montpellier) in the team ESCAPE, in the framework of the ANR project RaCAF (ANR-15- CE40-0016-01).
contract for 3 months with a possible extension
We talk about it on Twitter!