Fabian Joswig

Hi Daniel, I can compile the current `develop` branch on my laptop using gcc 9.4.0 and `../configure --enable-comms=none --enable-simd=AVX2 --prefix $prefix CXX=g++ MPICXX=g++ LDFLAGS=-L$prefix/lib/ CXXFLAGS="-I$prefix/include/`. I don't know about the...

Hey @PiaLJP & @s-kuberski, any updates on this PR? Do you still want to proceed of should we close it for now?

[NumPy 2.0](https://github.com/numpy/numpy/releases/tag/v2.0.0) has now been released. With [autograd not being maintained anymore](https://github.com/HIPS/autograd) we might have to think about alternatives if my proposed changes are not included in a new release...

Autograd is mainly needed for automatic differentiation & error propagation in fits, roots, integration and linear algebra functions. It conveniently provides the relevant wrappers for most numpy and scipy functions....

We managed to create a new autograd release that supports numpy 2 (https://github.com/HIPS/autograd/releases/tag/v1.7.0), I'm preparing a new pyerrors version that also supports numpy 2 and drops python 3.8 support (https://github.com/fjosw/pyerrors/pull/239)

I also remember discussing this with @s-kuberski some time ago. So far we have been very strict about not introducing breaking changes in minor releases. Should we consider making an...

Hi @jkuhl-uni & @s-kuberski , > What was the bug in the existing implementation? The missing 'float()'? Yes, we currently compute the gradients for `__pow__` via autograd which cannot cope...

> As soon as someone runs a script twice and saves the output somewhere to process it afterwards, I would think that a dedicated naming scheme is very important to...

Not quite done but already in decent shape for a review @s-kuberski @jkuhl-uni: - Most of the changes are just type annotations - I changed a few lines to help...

Feel free to get started with fixing stuff, you should have permissions to push to the branch, I think I won't have time to work on this in the comming...