No Description

Paul Spitzner 7282219b79 Update 'LICENSE' 3 years ago
dat 1493caa466 first commit 3 years ago
fig 1493caa466 first commit 3 years ago
plt 1493caa466 first commit 3 years ago
run 1493caa466 first commit 3 years ago
.gitignore 1493caa466 first commit 3 years ago
LICENSE 7282219b79 Update 'LICENSE' 3 years ago
README.md ab058dbb7e updated readme 3 years ago
datacite.yml 6ee63f523d Update 'datacite.yml' 3 years ago

README.md

Mr. Estimator Triallength Dependence

This is a collection of scripts that produce Fig.~4 of the paper on Mr. Estimator.

Dependencies

conda install h5py numpy matplotlib tqdm
pip3 install -U 'mrestimator[full]'

Outline

  1. create a realization of a branching process of certain length, timescale and number of trials. This is done in run/triallength.py
  2. calculate the correlation coefficients. This is done in run/triallength.py.
    • Because the first two steps take significant time, we attach a copy of the data in ./dat/ that contains precalculated correlation coefficients.
    • The run/triallength.py script takes an integer argument (think thread or job id) because we ran this on the cluster. Every id creates it's own hdf5 output file.
    • Merge the individual files with run/merge_hdf5.py to obtain data files similar to the ones provided here.
  3. fit the exponentials. This is done with plt/plot_merged.py.

Note that the realizations for long trials (large targettau and targetlength in run/triallength.py) will take on the order of days computation and multiple gigabytes of ram. Adjust as needed.

datacite.yml
Title MR Estimator Triallength Dependence
Authors Spitzner,F. P.;Max Planck Institute for Dynamics and Self-Organization;ORCID:0000-0001-9774-4572
Description Resources needed to produce Fig. 4 of the article.
License CC-BY (http://creativecommons.org/licenses/by/4.0/)
References MR. Estimator, a toolbox to determine intrinsic timescales from subsampled spiking activity [arXiv:2007.03367] (IsReferencedBy)
Funding
Keywords neuroscience
intrinsic timescale
branching parameter
autocorrelation time
subsampling
Resource Type Dataset