Alexis Montoison

@abelsiqueira I merged your modifications in #371

Hi @VikasChidananda! `symamd` computes a fill-in reducing permutation but it's based on an heuristic. Do you always have the same permutation in Julia and in OCTAVE / MATLAB? What is...

In AMD.jl, we directly call the C routines of SuiteSparse. For MATLAB / OCTAVE they probably use the `*.m` files of the this [folder](https://github.com/DrTimothyAldenDavis/SuiteSparse/tree/dev/CCOLAMD/MATLAB). Maybe the default options are different....

Hi @VikasChidananda, can you post the number of non-zeros of the factor `Lv`? The goal of the permutation is to reduce the fill-in of the factors (sparsity of the factors)....

Ok, so the two permutations give almost the same fill-in. It's what in want in practice. If you want to compare the reorderings, `nnz` is the most relevant metric.

Can I close this PR? All the environment variables have been removed because they not needed anymore with the new artifact `SIFDecode_jll.jl`.

Added in #352 and tested in #399.

@nimgould This issue is quite old but I don't know if you are aware of this.

Done in #366

You also have `bicgstab` and `cgs`.