Clément Pernet

Results 44 issues of Clément Pernet

Currently the "examples" folder plays two roles: 1. demonstrate simple examples of usage of the code 2. provide binaries for solving a single problem, such as Det, Rank etc This...

In particular in order to have fullCRA for rational solver with rational matrix. Suggestion: the scan through the rational matrix should compute d, the lcm of the denominators on the...

As illustrated here https://trac.sagemath.org/ticket/21579#comment:13, the impossible can happen. In this example, once in while, the charpoly over Z is incorrect, because early termination of the Chinese remainder algorithm finished after...

It offers `setEntry` and `getEntry` methods, calling the `setEntry` and `getEntry` of the template argument (supposed to be a blackbox). Hence it can not be used to build a transpose...

Presently, in every solution, there is a specialization for `BlasMatrix` but none for `BlasSubMatrix` As a consequence, when calling e.g. `det` on submatrix, the fallback used is the templated Matrix...

enhancement

As a follow up on #148: in the setting of a CRT solve, if the matrix mod p is not full rank, the system can be either - inconsistent: then...

Motivation: use rational chinese remaindering with polynomials. So far `rational-cra2.h` only proposes to work with generic vectors (not knowing their base field), but `varprec-cra-early-multip.h` line 96 requires it. Branch rational_vector_cra_types...

bug

there is an `RNS` class in `algorithms/rns.h`, which seems to be mostly a wrapper around givaro's Chinese Remaindering machinery. This seems to be a duplicate of LinBox CRA. It should...

Create a set of routines generating random matrices (dense, sparse, blackboxes), possibly with prescribed properties: rank, determinant, rank profile, rank profile matrix, signature, etc Should move, replace and factorize pieces...

enhancement

In many places, field element assignment are still done with operator=, which may work with basic types, but is likely to fail with mutliprecision/ polynomial types. We should get rid...