Trailblazing analysis error

Dear Authors,

I am currently attempting some trailblazing calculations using YANK for a screen of some drug fragments against the SARS-CoV-2 main protease. The repo is here.

I am seeing this (as of commit c991b54) on the analysis of the x0072-explicit-trailblazing.yaml run:

yank analyze --store=./x0072-explicit-trailblazing-output/experiments/

(openmm) [mw529@login-e-6 COVID_19_Drug_Binding]$ yank analyze --store=./x0072-explicit-trailblazing-output/experiments/
/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/analyze.py:299: YAMLLoadWarning: calling yaml.load() without Loader=... is deprecated, as the default Loader is unsafe. Please read https://msg.pyyaml.org/load for full details.
  analysis = yaml.load(f)
2020-03-22 21:29:22,651: WARNING - openmmtools.multistate.multistatereporter - Warning: The openmmtools.multistate API is experimental and may change in future releases
2020-03-22 21:29:22,777: WARNING - openmmtools.multistate.multistateanalyzer - Warning: The openmmtools.multistate API is experimental and may change in future releases
2020-03-22 21:29:22,781: WARNING - openmmtools.multistate.multistatereporter - Warning: The openmmtools.multistate API is experimental and may change in future releases
2020-03-22 21:29:22,820: WARNING - openmmtools.multistate.multistateanalyzer - Warning: The openmmtools.multistate API is experimental and may change in future releases
2020-03-22 21:29:23,353: WARNING - yank.analyze - Alert: analyzed timeseries has the same or fewer number of values as discard_from_start! The whole series has been preserved to ensure there is *something* to analyze.
/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/numpy/core/_methods.py:217: RuntimeWarning: Degrees of freedom <= 0 for slice
  keepdims=keepdims)
/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/numpy/core/_methods.py:186: RuntimeWarning: invalid value encountered in true_divide
  arrmean, rcount, out=arrmean, casting='unsafe', subok=False)
/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/numpy/core/_methods.py:209: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
Traceback (most recent call last):
  File "/home/mw529/.conda/envs/openmm/bin/yank", line 11, in <module>
    load_entry_point('yank==0.25.2', 'console_scripts', 'yank')()
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/cli.py", line 73, in main
    dispatched = getattr(commands, command).dispatch(command_args)
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/commands/analyze.py", line 148, in dispatch
    single_run()
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/mpiplus/mpiplus.py", line 271, in _wrapper
    return run_single_node(rank, task, *args, **kwargs)
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/mpiplus/mpiplus.py", line 220, in run_single_node
    result = task(*args, **kwargs)
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/commands/analyze.py", line 143, in single_run
    output = analyze.analyze_directory(args['--store'], **analyzer_kwargs)
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/analyze.py", line 783, in analyze_directory
    analysis_data = auto_experiment_analyzer.auto_analyze()
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/analyze.py", line 173, in make_copy
    return copy.deepcopy(wrap(*args, **kwargs))
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/analyze.py", line 703, in auto_analyze
    _ = self.get_equilibration_data()
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/analyze.py", line 173, in make_copy
    return copy.deepcopy(wrap(*args, **kwargs))
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/yank/analyze.py", line 452, in get_equilibration_data
    i_t, g_i, n_effective_i = mmtools.multistate.get_equilibration_data_per_sample(self.u_ns[phase_name])
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/openmmtools/multistate/utils.py", line 170, in get_equilibration_data_per_sample
    g_i = np.ones([max_subset], np.float32)
  File "/home/mw529/.conda/envs/openmm/lib/python3.7/site-packages/numpy/core/numeric.py", line 207, in ones
    a = empty(shape, dtype, order)
ValueError: negative dimensions are not allowed 

I suspect the key part is:

"2020-03-22 21:36:40,185: WARNING - yank.analyze - Alert: analyzed timeseries has the same or fewer number of values as discard_from_start! The whole series has been preserved to ensure there is *something* to analyze." 

which I think is thrown from here. I am still finding my way around the code and I suspect at this point, it may be user error on my behalf.

Any suggestions or tips would be really useful.

[andrrizzi]

Another thing that might work is to increase the thermodynamic_distance parameter of thermodynamic trailblazing from 2.0 (kT) to something like 2.5 or 3.0. It should generate a smaller number of intermediate states, but the advantage of using that algorithm is that they will be nearly optimally spaced so that the statistic can tolerate a sparser protocol.

[hannahbrucemacdonald]

Pleased that rolling back pymbar helped. 400 iterations (0.4ns per iteration) is still quite short, and in the regime of equilibration really. Are you able to run for longer?

If wallclock time is the limitation at the moment, as @andrrizzi mentioned, you can always restart simulations from the checkpoint to stretch the simulation time out longer!