Current

The seminar is held on Tuesdays from 11 to 12 and is organised by Sabrina Kunzweiler and Maxime Bombar. Unless stated otherwise it takes place in room 2 of IMB. To get announcements, you can subscribe to the mailing list of the Bordeaux number theory seminars. Last minute changes may appear first on the IMB website.

• 2026-03-17
  11:00
  Salle 2
  Alix Barraud (CANARI)
  Dual of AG Codes from Hirzebruch Surfaces
  Duals of AG (Algebraic Geometry) codes on curves are well-understood and can easily be expressed as AG codes on the same curves. However, next to nothing is known for duals of AG codes on surfaces, only that they can be expressed as sums of AG codes on the same surfaces, but the proof is not constructive. As a consequence, only a few duals of AG codes on surfaces are known, and their minimum distances remain difficult to estimate. The aim of this talk is to give an explicit form for the duals of AG codes from Hirzebruch surfaces. After a visual description of these surfaces and their rational points, I will describe their AG codes before presenting how we manage to compute their duals.
• 2026-03-24
  11:00
  Salle 2
  Mahshid Riahinia (ENS)
  Post-Quantum Public-Key Pseudorandom Correlation Functions for Oblivious Transfer
  Public-Key Pseudorandom Correlation Functions (PK-PCFs) are functions that generate pseudorandom correlated strings. These correlations can then be used to speed up secure computation protocols. Recent works have made significant progress building PK-PCFs using group-based assumptions, however, these assumptions do not hold up against quantum attackers. Much less is known about PK-PCFs in the post-quantum regime. In this talk, I will introduce an efficient lattice-based PK-PCF for oblivious transfer (OT) correlations. At the heart of our result lie several technical contributions that might be of independent interest. In particular, we introduce the first efficient lattice-based constrained pseudorandom functions for low-degree polynomials, from a new but natural “secret-power” variant of ring learning with errors (ring LWE) assumption.