Azercoco
Azercoco
Also in PALC, the solver is expected to have a field `solver`, which can be equal to `nothing` but has to be present nevertheless or the compilation will fail.
Yes, that was the issue, the work around I found was just to add an solver field with a nothing value. No the cleanest but it did work.
This may be the same issue as #191.
Actually, it seems that the implemention of `exp` is not well defined. Calling it on a `DiagonalOperator` convert it to a Matrix type instead of preserving the type
Thanks for the example. I did not saw it. The interest of RKIP methods and Split-Step are mostly their performance when implemented with FFTs. RKIP can moreover give adaptive time...
The adaptative methods for split problem does not seems to take profit of semilinear structure of the problem. Currently, for a semilinear problem, we have the choice between: - Adaptive...
@jtravs I did experiment a little with exponential integrator but only for fixed size step. This tends to made them slower than adaptative method if one does not take great...
Hi, I have started to write a PR and I saw than the Newton code is duplicated between the PALC algorithm and from the Newton solver (with the latter not...