Alexis Montoison

@Wimmerer I created a branch for you https://github.com/JuliaSmoothOptimizers/Krylov.jl/tree/spmv_graphblas. You just need to add your methods [here](https://github.com/JuliaSmoothOptimizers/Krylov.jl/blob/spmv_graphblas/src/krylov_utils.jl#L222). :smiley:

If you open a PR, I can easily use our `JSOBot` to run the benchmarks and see the improvments (https://github.com/JuliaSmoothOptimizers/Krylov.jl/pull/456).

@vpuri3 Is it related to this [issue](https://github.com/JuliaSmoothOptimizers/Krylov.jl/issues/476) ?

`CHOLMOD` doesn't have in-place solves. So if you use sparse Cholesky or LDL factorizations you should use `\` for the backward and forward sweeps. On Slack, TIm Davis explained that...

Yes I know, but LDLFactorizations is not used in `LinearOperators.jl`. The problem of Tim is that they didn't implemented `permute!` and `invpermute!` routines in CHOLMOD. He also said that it...

Nice examples !

For beginning, transfer based on residual norms could be interesting. We just need to compute exactly the residual norms of QMR for that. #156

LLᵀ and LDLᵀ could be used to solve systems with `M` and `N` within saddle-point and SQD systems. Is it still relevant to call it preconditioners when a process is...

Note that Armv8-A and Armv8-R Architectures support: > Configurable rounding modes > Configurable Default NaN behavior > Configurable Flush-to-zero behavior Julia is maybe compiled with a different rounding mode for...

The test with CR just failed with a GitHub Actions build on Ubuntu :thinking: