PhD positions

The following PhD subject are proposed by ITI IRMIA++ members for PhD contracts starting in september/october 2024.

In order to apply, please send to iti-irmiapp[at]unistra.fr:

  • the subject you are applying for (Please note: applications that do not specify the chosen subject will not be considered)
  • your resume
  • a cover letter
  • full transcript of Master's degree grades
  • names and contact information of references.

Deadline for application reception : April 20th, 2024


If you are interested in a subject which is not in the subjects list, please contact directly the researchers and team you want to work with.

 

If you start a PhD in an ITI IRMIA++ team, we can offer you a financial help for your installation !
More information on the dedicated page.

Subject: Efficient Data Compression within the Task-Based Method for High-Performance Applications

Supervision

Stéphane Genaud (ICube, Strasbourg)
Philippe Helluy (IRMA, Strasbourg)
Bérenger Bramas  (ICube, Strasbourg)

Laboratory and team

ICube, Strasbourg - Team "ICPS"

Subject description

The PhD thesis will focus on improving the efficiency of high-performance applications through a sophisticated data compression approach. It will improve compression and decompression techniques for better performance on CPUs and GPUs, and introduce the automatic integration of data compression within task-based applications by modifying runtime systems such as StarPU. The research will evaluate the impact of compression on various numerical applications, in order to demonstrate its effectiveness and identify its limitations. The thesis will attempt to produce a predictive model for autonomously deciding on the use and configuration of compression, while taking into account the accuracy of simulations despite the potential loss of data due to compression.

Related mathematical skills

Linear algebra
Numerical Analysis
HPC, CUDA, C++

 

Subject: Wheeled operads in algebra, topology, and numerical methods

Supervision

Vladimir Dotsenko (RMA, Strasbourg)
Najib Idrissi (IMJ-PRG, Université Paris Cité)

Laboratory and team

IRMA, Strasbourg - Team "Algèbre, représentations, topologie"

Subject description

An operad is an algebraic notion that formalizes properties of multilinear operations with several arguments. A wheeled operad additionally encodes traces (or, in the analytic context, divergences). This thesis will rely on recent work of the supervisor to explore applications of wheeled operads in a range of theoretical and applies problems.

Related mathematical skills

The selected candidate will be sufficiently fluent in basics of category theory and homotopical algebra, and will have written their master thesis on one of these subjects. Expertise in combinatorics is desirable but not essential.

Subject: Disease detection and quantification on a digitized urban tree

Supervision

Franck HETROY-WHEELER (ICube, Strasbourg)
Rémi ALLEGRE (ICube, Strasbourg)
Vincent VIGON (IRMA, Strasbourg)

Laboratory and team

ICube, Strasbourg - Team “IGG"
IRMA, Strasbourg - Team “Probabilités"

Subject description

Design of a frugal deep learning approach to detect and quantify diseases on an urban tree 3D point cloud.

Related mathematical skills

Excellent knowledge of deep learning main architectures
Digital geometry

Subject: Applications of Rabinowitz Floer homology in Hamiltonian dynamics

Supervision

Alexandru Oancea (IRMA, Strasbourg)

Laboratory and team

IRMA, Strasbourg - Team "Geometry"

Subject description

Rabinowitz Floer homology is an invariant of exact contact embeddings. It was shown recently that it carries the structure of a graded Frobenius algebra. This project aims to apply this powerful new algebraic structure to the study of Hamiltonian dynamics.

Related mathematical skills

Strong background in geometry and topology

Subject: Procedural generation of multi-channel textures based on random fields

Supervision

Basile Sauvage (ICube, Strasbourg)

Laboratory and team

ICube, Strasbourg - Team “IGG"

Subject description

The PhD candidate will investigate new models of stochastic fields dedicated to the specificities of multi-channel textures. He/she will lay the mathematical foundations, design novel algorithms to generate the textures, and implement them on graphics boards. The analysis of and comparison to real world examples is also an objective.

Related mathematical skills

Analysis (integration, derivation), probabilities and statictics (random variables).
Bonus : stochastic processes, Fourier transform.

 

Subject: Deep-learning decoding of star formation in high redshift galaxies

Supervision

Florent Renaud (ObAS, Strasbourg)

Laboratory and team

Observatoire Astronomique de Strasbourg, Strasbourg

Subject description

The physical conditions for star formation in galaxies in the early Universe (at high redshift) are still poorly known. The goal of this project is to use machine-learning algorithms to post- process cosmological simulations in order to help the physical interpretation of observations.

Related mathematical skills

-

Subject: Making Coq Proofs More Reliable and More Easily Reusable: Applications to Mathematics

Supervision

Nicolas Magaud (ICube, Strasbourg)

Laboratory and team

ICube, Strasbourg - Team "IGG"

Subject description

Proof assistants like Coq or Lean are increasingly popular to help mathematicians carry out proofs of the results they conjecture. However, formal proofs remain highly technical and are difficult to reuse. In this thesis, we propose to design and implement new tools to improve the robustness, the inter-operability and the reuse of formal proof developments. We hope this will make proving new results formally easier.

Related mathematical skills

Foundations of Mathematics
Logic and Type Theory
Applications are possible in various fields of mathematics including combinatorics, algebraic and/or differential geometry and applied mathematics.

 

Subject: Three-dimensional surfaces preserving their geometric and topological properties under a discretization process

Supervision

Etienne Baudrier and Etienne Le Quentrec (ICube, Strasbourg)

Laboratory and team

ICube, Strasbourg - Team "IMAGeS"

Subject description

Our research aims to establish a correspondence between continuous and discrete representations of objects, crucial for geometric operations in image processing. To achieve this, the thesis objective is to extend the class of Curves with Locally Bounded Total Curvature developed in 2D to 3D. This study involves establishing local topological properties on surfaces to demonstrate results on the topology of their discretization and subsequently comparing the global geometric quantities of the continuous surface with its discretization in 3D.

Related mathematical skills

Bachelor level in geometry and topology.
Programming skill useful but not required.

INRIA
UFR de mathématique et d'informatique
Faculté de physique et ingénierie
ICUBE
IRMA
Observatoire astronomique de Strasbourg