Clemens

> > Instead of ctx.re(x), ctx.im(x), and ctx.conj(x), use x.real, x.imag and x.conjugate(). These are [standardized Python accessors for complex numbers](https://docs.python.org/3/library/numbers.html#numbers.Complex) which both mpmath and gmpy2 already support. > >...

This looks good so far. For the precond_parameter class, most of the `Integer` fields there are actually just de-facto enums and could simply be `int`. For the remaining ones, it...

> So, changing all the `Integer` fields in the parameters to be hard-coded to either `int` or `int64` and then importing one of the two versions in the python code...

> ```c++ > template matrix_sparse matrix_sparse::transpose_in_place() { > orientation = other_orientation(orientation); > std::swap(number_rows, number_columns); > return *this; > } > ``` > > Why transpose in place just swap number...

> And in practice, CCS format calculate of ILU0 is wrong, hoping your reply. Can you give a reproducible test case where it's wrong?

Hello Marat, It's a bit of a gap that this isn't implemented by default. Part of the problem is that conversion to this fraction of polynomials representation isn't well-conditioned in...

Hello, Are you talking about a complex-valued function defined over a real interval? I believe this should be possible, though I'm not sure if all the details are implemented. Best...

If there is a complex argument, it's not possible to use BRASIL since it only works over intervals. You could however use the AAA algorithm, which is also contained in...