Can only run encore convergence on DCDReader/MemoryReader trajectories

[lilyminium]

Expected behavior

I can run encore.[dr/c]es_convergence on any MDAnalysis universe trajectory.

Actual behavior

It fails because prepare_ensembles_for_convergence_increasing_window fails to transfer trajectories into memory before calling timeseries on them.

Code to reproduce the behavior

Show us how to reproduce the failiure. If you can, use trajectory files from the test data.

>>> import MDAnalysis as mda
>>> from MDAnalysis.tests.datafiles import PSF, DCD,  GRO, PDB, TPR, XTC, TRR,  PRMncdf, NCDF
>>> from MDAnalysis.analysis import encore

>>> u = mda.Universe(TPR, XTC)
>>> encore.dres_convergence(u, 10)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-23-24fe8767487d> in <module>
----> 1 encore.dres_convergence(u, 10)

~/pydev/mdanalysis/package/MDAnalysis/analysis/encore/similarity.py in dres_convergence(original_ensemble, window_size, select, dimensionality_reduction_method, nsamples, ncores)
   1672 
   1673     ensembles = prepare_ensembles_for_convergence_increasing_window(
-> 1674         original_ensemble, window_size, select=select)
   1675 
   1676     coordinates, dimred_details = \

~/pydev/mdanalysis/package/MDAnalysis/analysis/encore/similarity.py in prepare_ensembles_for_convergence_increasing_window(ensemble, window_size, select)
    693     """
    694 
--> 695     ens_size = ensemble.trajectory.timeseries(ensemble.select_atoms(select),
    696                                               order='fac').shape[0]
    697 

AttributeError: 'XTCReader' object has no attribute 'timeseries'

Currently version of MDAnalysis

• Which version are you using? (run python -c "import MDAnalysis as mda; print(mda.__version__)") 0.20.2-dev
• Which version of Python (python -V)?
• Which operating system?

[richardjgowers]

Related work:

#1400 (comment)

#2271

I think maybe the cleanest fix would be to scrap the concept of .timeseries() (which dumps coordinates into memory) and instead use .transfer_to_memory() (which also dumps coordinates into memory, but in a more useful form). Ie a MemoryReader is much better than a numpy array

[orbeckst]

There are legitimate reasons to have a whole trajectory as one big array; however, one can use MemoryReader.get_array() (or just MemoryReader.coordinate_array (although this is not explicitly documented). The MemoryReader.timeseries() method is useful, though, because it can reshape the array – how often that is used is another question.

[hmacdope]

linked to #3761