mortonne / cymr

Implementation of the CMR model using Cython.

I am a Professor at the University of Wisconsin-Milwaukee and Director of the Human and Machine Learning (HaML) Lab. I use computational modeling and analysis of neuroimaging data to understand human memory.

I've developed a number of software packages for research:

• WikiVector: Tools for quantifying conceptual knowledge based on Wikipedia entries.
• Psifr: Analysis and visualization of free recall data.
• PsifrR: R package for analysis and visualization of free recall data.
• CyMR: Computational modeling of free recall data.
• PsiReact: Hierarchical Bayesian modeling of response time data.
• Mindstorm: Advanced neuroimaging analysis.
• Wiki2Vec: Modeling semantic similarity of well-known stimuli.
• fPrep: Scripts for preprocessing of fMRI data using FSL, FreeSurfer, and ANTS.
• Aperture: Matlab toolbox for univariate and multivariate analysis of EEG data.

cymr's Issues

Package fails to install from PyPI

Installation from source code works correctly, but fails when installing from PyPI. This seems to be caused by the fact that the build system only adds .c files to the source distribution and not .pyx files, but installation expects there to be .pyx files. I'm not sure what the usual pattern is here when distributing source code that includes .pyx files. It looks like part of the reason this happens is that an external library, math.c, is used.

parameter_sweep crashes when sweeping over only one parameter

I was trying to just sweep over B_rec, but it kept crashing. Then I added B_enc to the param_names and it worked. It may be easiest to replicate by taking the plot_sweep_B code and removing B_enc from the param_names. I don't want to upload my scratch script into the cymr project, and it won't let me attach a py file here.

Here's the error output when there's just one param being swept over, "name must be a hashable type"

File "/Users/polyn/PycharmProjects/cymr_dev/tutorial.py", line 79, in
results = model.parameter_sweep(study, param, param_name, param_sweep,
File "/Users/polyn/opt/anaconda3/envs/cymr_dev/lib/python3.8/site-packages/cymr/fit.py", line 650, in parameter_sweep
results.index.rename(param_names, inplace=True)
File "/Users/polyn/opt/anaconda3/envs/cymr_dev/lib/python3.8/site-packages/pandas/core/indexes/base.py", line 1378, in rename
return self.set_names([name], inplace=inplace)
File "/Users/polyn/opt/anaconda3/envs/cymr_dev/lib/python3.8/site-packages/pandas/core/indexes/base.py", line 1325, in set_names
idx._set_names(names, level=level)
File "/Users/polyn/opt/anaconda3/envs/cymr_dev/lib/python3.8/site-packages/pandas/core/indexes/base.py", line 1245, in _set_names
raise TypeError(f"{type(self).name}.name must be a hashable type")
TypeError: Float64Index.name must be a hashable type

Error when running a simulation

In [12]: from cymr import cmr

In [13]: model = cmr.CMRDistributed()

In [14]: sim = model.generate(data, param, param_def=param_def, patterns=patterns, n_rep=5)

Error Message:
KeyError Traceback (most recent call last)
in ()
----> 1 sim = model.generate(data, param, param_def=param_def, patterns=patterns, n_rep=5)

2 frames
/usr/local/lib/python3.6/dist-packages/cymr/cmr.py in prepare_list_param(n_item, n_sub, param, param_def)
158 )
159 else:
--> 160 Lcf_trial = primacy(n_item, param['Lcf'], param['P1'], param['P2'])
161 Lcf = np.tile(Lcf_trial[:, None], (1, n_sub))
162

KeyError: 'P1'