convert function for solvers

[geoffroyleconte]

It could be practical to have something like:

convert(::Type{KrylovSolver{T, FC}}, solver::MinresSolver) where {T, FC} = MinresSolver(...)

to convert a solver from Float32 to Float64 for example. However I'm not sure what should be the way to implement this. Should the user provide FC or could it be inferred from T ?

[amontoison]

The user should provide S (Vector{ComplexF32}, CuVector{Float64}, ...)
T and FC are determined from S.

Is it not easier to just build a new solver with the constructor XyzSolver(n, m, S) ?
I can also add n and m attributes in the solver such that

convert(::Type{MinresSolver{T, FC, S}}, solver::MinresSolver) where {T, FC, S} = MinresSolver(solver.n, solver.m, S)

but It's a more complex way to do the same thing.

[geoffroyleconte]

Yes, but that was in case we would want to keep the current solution for using warm start later for example.