save sampler.results into file

[wenbinlu]

I can't find where this is discussed in the documentation. I'm doing a number of independent sampling runs of the same problem (either by different machines or by the same machine at different time). I would like to save the sampling "Results" for each independent run into a file (e.g., .h5 format). After finishing all these runs, I want to read these files so as to load the "Results" into the memory again, and then I can merge all these independent samples into a big sample using "merge_runs", which takes a list of "Results" as input. Do I need to write my own save and read script or there is already a module that handles this?

[segasai]

You need to write it yourself, but as far as I'm aware it's trivial (if you are okay with pickling)

import pickle
import numpy as np
import dynesty
import dynesty.utils as dyutil

nlive = 100

size = 10  # box size

def loglike(x):
    return -0.5 * np.sum(x**2)

def prior_transform(x):
    return (2 * x - 1) * size

def test():
    # run sampling
    ndim = 2
    rstate = np.random.default_rng(444)
    resl = []
    for i in range(3):
        sampler = dynesty.DynamicNestedSampler(loglike,
                                               prior_transform,
                                               ndim,
                                               nlive=nlive,
                                               rstate=rstate)
        sampler.run_nested()
        resl.append(sampler.results)

    # save the results
    with open('xx.pkl', 'wb') as fp:
        pickle.dump(resl, fp)

def test1():
    # read the results
    with open('xx.pkl', 'rb') as fp:
        resl = pickle.load(fp)
    # merge them
    dyutil.merge_runs(resl)

test()
test1()