langou

I agree with @thijssteel that I am not convinced about the argument "reduce considerably the number of routines". one example of use of this PR: getrs.hpp solves Ax = b...

> do you know why LAPACK only uses a single directory for all methods? Maybe there is a good reason. The "why" is not clear to me. I like the...

Hi, I feel that you are reporting an issue about OpenBLAS. The GitHub of OpenBLAS is at https://github.com/xianyi/OpenBLAS I think you want to report over there and not here. Julien.

We want to solve Ax = b where A symmetric positive definite and b is a right-hand side. When equilibration is enabled, DPOSVX computes diagonal matrix S such that S(i,i)...

> my question was about why the value of 0.1 and why the min(A(i,i))/max(A(i,i)) criterion. The goal of equilibration is to have min(A(i,i))/max(A(i,i)) = 1. Equilibration "equilibrates" all diagonal entries...

(*) A. van der Sluis (1969) “Condition numbers and equilibration of matrices” in Numer. Math., 14 (1):14–23, showed that the algorithm to compute D leads to nearly optimal equilibration in...

Hi, Good observation. I cannot remember the reason for us dropping the LAPACKE files from doxygen. Maybe there is no reason and this just happened during an update. Maybe we...

I agree with @mgates3. I agree that it would be better and easier too and probably faster to break down this PR in several smaller PRs that we would slowly...

With respect to the initial post of this discussion thread ( https://github.com/Reference-LAPACK/lapack/pull/1043#issue-2473469880 ) and the lines of codes at: https://github.com/Reference-LAPACK/lapack/blob/9c0ef6653d291f27dc8d3c7aa7f3adc27cced5f0/SRC/slaqr3.f#L659-L667 Indeed when we do not want the Schur factor T,...

⏸️ I would like to press the pause button on this. ⏸️ I am not saying "no", but I would like that we give time for folks to express their...