Giuseppe Carleo

Diffrax

1

https://github.com/patrick-kidger/diffrax has been realeased, including support for SDEs. I think it might make sense to eventually drop our internal ODE solver in favor of this. What do you think @femtobit...

Internal qgt solve method does not accept dense ravellings

1

```python import netket as nk import jax ma = nk.models.RBM(alpha=1, dtype=float) sa = nk.sampler.MetropolisLocal(nk.hilbert.Spin(0.5, 16), n_chains=16) vs = nk.vqs.MCState(sa, ma) qgt = vs.quantum_geometric_tensor() n_pars=vs.n_parameters v=jax.numpy.ones((n_pars,)) # This works x,info=jax.scipy.sparse.linalg.gmres(qgt,v) #...

WIP: Add a test for qgt

3

This adds a simple test for the quantum geometric tensor (it should use the other testing facility for variational things, but it currently is...) The tests pass for real parameters,...

Add LogVal to MCState

8

In order to have a complete interface in ```MCState``` (and also to easily implement some missing features in 3.0, such as #556 ) it is important to add a (differentiable)...

Invert ordering of spin states to match standard S_z operator

2

This addresses what raised in #607, namely that historically the basis states of ```Spin``` did not match the eigenstates of S_z. I have also added a clarification in the docs...

Write supervised learning as a python driver

4

@ShHsLin if you had time to look into this at some point it would be great

Quantum State Reconstruction (```netket.Qsr``` in the last 2.1 version) is not currently available in the development branch (and soon to be released) v3.0 Most likely we won't be able to...

Improve statistics with Geyer style autocorrelation functions

1

Our estimates of variance and correlations times are based on re-blocking single chains and analyzing multiple independent chains [code [here](https://github.com/netket/netket/blob/v3.0/netket/stats/mc_stats.py)]. We should implement however also robust estimators as in equations...

Tutorial on MPI usage

2

It would be helpful to add a tutorial explaining how to run things in parallel (mpirun etc). There are some users who are not familiar with mpi (see #173 )...

Generalized Hartree fock

4

For spinful fermions, `Slater2nd` implements a product of determinants for each spin component. For example for spin 1/2 it gives $$ \psi(n)=\mathrm{Det}(\phi^{d}_i(r^{d}_j) ) \times \mathrm{Det} (\phi^{u}_i(r^{u}_j) ) $$ where $r^{u}_j$...