Jack S. Baker
Jack S. Baker
Also, this bug (or related bug) seems to pre-exist in another tutorial. Please see https://covalent.readthedocs.io/en/latest/tutorials/machine_learning/mnist_classifier.html . Upon comparing code cells 11 and 14, you will see that cell 11 executes...
So @cjao does this work if you choose a derivative free optimizer? (Try Powell or something)
bump on this @tovrstra would love to know if there exists a text/texts one can use side-by-side to this code implementation!
I think I will add some basic unit testing for Molecule and Functional classes. I think Integration testing is as good as it needs to be though!
I contacted the developers to see the best way to expose the energy densities. Let's wait and see...
Ok so this should in principle we quite straight-forward. Great find @PabloAMC. I'll make an attempt at this shortly.
Related to #83
Having playing around with the [multiple hosts parallelism](https://jax.readthedocs.io/en/latest/multi_process.html) in JAX, I came across many issues on Perlmutter with the detection of GPUs. I'm giving [mpi4jax](https://github.com/mpi4jax/mpi4jax) a go for this task...
Changing to `safe_eigh` did not fix things! I'm seeing this with diatomic molecules using `6-31G` bases. I.e, ```python Na2_bl = 3.0789 Na2_mol = pyscf.M( atom = 'Na %.5f 0.0 0.0;...
although I have now noticed that it also occurs in `gd.mse_energy_loss`