decode module

Module for decoding spatial position from neuronal activity

(c) 2016 C. Schmidt-Hieber GPLv3

decode.decodeMLNonparam(activity_map, activity_time, nentries=4)[source]

Decode spatial position from neuronal activity. Compute maximum likelihood non-parametrically.

Parameters:
  • activity_map (3D list) – \(r\), Spatial activity map of shape (nrois, ncrossings, x)
  • activity_time (numpy.ndarray) – \(c\), Activity time series, shape (ntimepoints, nrois)
  • nentries (int, optional) – Mean number of entries per bin in the histogram
Returns:

L – Decoded spatial maximum likelihood map for each time bin, shape (x, y, ntimepoints).
Return type:

numpy.ndarray
decode.decodeMLPoisson(ratemap, counts_time)[source]

Decode spatial position from neuronal activity. Compute maximum likelihood assuming spikes are a Poisson process. Follows Dan Manson’s code published here: https://d1manson.wordpress.com/2015/11/19/non-trivial-vectorizations/

Parameters:
  • ratemap (numpy.ndarray) – \(r\), Spatial firing rate map of shape (x, nrois) or (x, y, nrois)
  • counts (numpy.ndarray) – \(c\), Spike counts per time bin, shape (ntimepoints, nrois)
Returns:

L – Decoded spatial maximum likelihood map for each time bin, shape (x, y, ntimepoints). \(L = \prod_{i=0}^{nrois}{\frac{r_i(x,y)^{c_i(t)}e^{-r_i(x,y)}}{c_i(t)!}}\)
Return type:

numpy.ndarray