compute_dc2_()

Compute the cut-off distance squared (dc2_) for a site K using the distribution of the pairwise feature distances. dc is chosen at the bottom P.dc_percent distance percentile. Temporal neighbors are used for distance pairs according to P.nTime_clu parameter. If P.nTime_clu==1, all neighbors are used, and half of the neighbors are used according to the temporal order if nTime_clu==2.

Syntax

• dc2_ = compute_dc2_(mrFet12, viiSpk12_ord, n1_, n2_, P)

Input

• mrFet12: Feature matrix for n1_+n2_ spikes (nFeatures_spk x (n1_+n2_): single)
• viiSpk12_ord: Temporal ordering of the spikes
• n1_: Number of spiking events whose peak occurs at the site K
• n2_: Number of spiking events whose second peak occurs at the site K
• P: Parameters struct (P) containing the percentile threshold value (P.dc_percent).

Output

• dc2_: squared cut-off distance

Operations

• Up to 1000 spikes are subsampled from n1_ spikes
• Up to 4000 spikes are subsampled from n1_ spikes
• Distance matrix is computed for 1000 x 4000 spikes if the spike pairs are temporally adjacent (P.nTime_clu)
• Distance percentile is computed (P.dc_percent, 2% by default)