Eric Giguère

@rochisha0 Yes.

> We allow the user to supply the `QobjEvo` class to use as an option, and allow them to specify either the class itself or a name for the class....

@hodgestar I removed the `_rhs` by changing it so the rhs is built when making the integrator instead of in `__init__`. I also simplified the way options are updated so...

It's in development. If you use qutip v5.0.0a2 on pypi or master branch here with [qutip-jax](https://github.com/qutip/qutip-jax), some solver can work on gpu. (We tested `sesolve` and `mesolve`, mcsolve should work,...

I suggest you to use the example in https://github.com/qutip/qutip-jax/issues/26. First ensure that jax is using the gpu: ``` import jax jax.devices() ``` After I would suggest to use the branch...

It should not affect the results of simulations. The diffrax package does not interact with complex number directly when used through qutip-jax. Gradient or other derivatives could be affected, this...

I don't know, since jax is developed by google I expect it to work well with gpus it provide through collab... We don't use schur decomposition for mesolve directly. I...

Only function decorated with jit are supported as coefficient in with diffrax. For spline interpolation, you could use [jax-cosmo implementation](https://jax-cosmo.readthedocs.io/en/latest/_modules/jax_cosmo/scipy/interpolate.html).

Thank you for reporting. However I am unable to reproduce it (on ubuntu, same scipy, numpy, cython, python3.11.5). So I can't tell what went wrong or how to solve it...

On x64, I had arch Linux in mind... I don't have a setup to test on aarch64, nor ever tried to emulate it. It may take some time.