Pair correlations

[SvenPVoigt]

Pair Correlations can be calculated in many ways. In this pull request, I am calculating pair correlations from 2point statistics- in the new function paircorr_from_twopoint. One caveat of using two point statistics is that at large r values, probability values become very small and closely spaced. Therefore, the function provides a cutoff_r and an interpolate_n parameter for dealing with probabilities at large r values.

Additionally, I provide a dist_from_center function, which could be used in other functions as well because it is so generic.

I put both functions in pair_correlations.py, but I am open to adding dist_from_center in func.py and paircorr_from_twopoint in correlations.py. If I change the location of the functions, is there a git tool (maybe the squash and merge option) for cleaning up commits so that only the changes to final (not intermediate) files are represented?

Also, if there is any area for increasing the usage of dask, please let me know. I will try to identify some additional spots as well. Right now, I am simply validating that dask and numpy arrays can be passed as input to my functions.

[wd15]

@SvenPVoigt nice job. I'll hopefully review this in the next few days.

[wd15]

I'm a little confused. I think that paircorr_from_twopoint should return a tuple with the probabilities in the first item and the radii in the second item. I don't think it's good to store data with different units in the same array. Also, the probabilities should be shaped (n_samples, n_interpolate) to fit in with how the first axis is the sample axis as in the rest of PyMKS. What do you think?

[wd15]

New dist_from_center code?

center = lambda x: np.reshape(
    np.array(x) // 2,
    (len(x),) + (1,) * len(x)
)
    
dist_mesh = sequence(
    fmap(lambda x: np.linspace(0, x - 1, x)),
    lambda x: np.array(np.meshgrid(*x, indexing='ij')),
    lambda x: x - center(x.shape[1:]),
    lambda x: np.linalg.norm(x, axis=0)
)

[wd15]

@SvenPVoigt I've made a PR into your branch. You probably should merge upstream's master and rebase and then I'll redo the PR. Anyway I've made things more functional. Also,

• added paircorr_from_twopoint to the API docs
• make make_da keep its docstring
• Made the docstring better

You might want to check that you're happy with the API docs.

[wd15]

@auag92, @beyucel : can you both review this now. It's ready. Also, @SvenPVoigt can you also give your seal of approval in the comments. Thanks

[wd15]

@auag92 @beyucel : I added @SvenPVoigt new function to the API so please check that it reads well, https://pymks--524.org.readthedocs.build/en/524/API.html#pymks.paircorr_from_twopoint