API changes in dynesty>=1.1 (master branch May 7th)

I tried running the linear-regression example with dynesty + the rwalk method:

#!/usr/bin/env python
import bilby
import numpy as np
import matplotlib.pyplot as plt

label = 'linear_regression'
outdir = 'outdir'
bilby.utils.check_directory_exists_and_if_not_mkdir(outdir)

def model(time, m, c):
    return time * m + c

injection_parameters = dict(m=0.5, c=0.2)
sampling_frequency = 10
time_duration = 10
time = np.arange(0, time_duration, 1 / sampling_frequency)
N = len(time)
sigma = np.random.normal(1, 0.01, N)
data = model(time, **injection_parameters) + np.random.normal(0, sigma, N)
likelihood = bilby.likelihood.GaussianLikelihood(time, data, model, sigma)

priors = dict()
priors['m'] = bilby.core.prior.Uniform(0, 5, 'm')
priors['c'] = bilby.core.prior.Uniform(-2, 2, 'c')

result = bilby.run_sampler(
    likelihood=likelihood, priors=priors, sampler='dynesty', nlive=500,
    sample="rwalk", injection_parameters=injection_parameters, outdir=outdir, label=label, clean=True)

and got the following error after the sampling had completed and bilby was trying to save the results:

/usr/local/lib/python3.7/dist-packages/bilby/core/sampler/dynesty.py in dump_samples_to_dat(self)
    601     def dump_samples_to_dat(self):
    602         sampler = self.sampler
--> 603         ln_weights = sampler.saved_logwt - sampler.saved_logz[-1]
    604 
    605         weights = np.exp(ln_weights)

AttributeError: 'MultiEllipsoidSampler' object has no attribute 'saved_logwt'

I think dynesty has gotten rid of the saved_logwt attribute.

versions:

• Dynesty version (master branch from May 7, 2021)
• Bilby 1.1.2: (CLEAN) 15f0012b 2021-08-16 12:18:30 +1000