KS3 having peoblem with getMeWtW while KS2 works!

[alirezasaeedi1988]

Hello everyone,

I am using a Neuronexus 32 channel electrode and facing the following error in Kilosort3 while kilosort2 works fine on my data!
Index in position 1 is invalid. Array indices must be
positive integers or logical values.

Error in getMeWtW (line 30)
iList = int32(gpuArray(isort(iNear, :))); % return the
list of pairs for each template

Error in trackAndSort (line 100)
[WtW, iList] = getMeWtW(single(W), single(U), Nnearest);

Somehow rez.W and rez.U are empty arrays. Do you have any suggestions?

Thank you in advance.

[shirquinn]

i'm having the same issue
kilosort
Time 0s. Computing whitening matrix..
Getting channel whitening matrix...
Channel-whitening matrix computed.
Time 1s. Loading raw data and applying filters...
Time 25s. Finished preprocessing 596 batches.
vertical pitch size is 25
horizontal pitch size is 0
0

157

0.09 sec, 1 batches, 628 spikes
4.88 sec, 101 batches, 67439 spikes
8.95 sec, 201 batches, 136865 spikes
12.87 sec, 301 batches, 208768 spikes
16.54 sec, 401 batches, 284892 spikes
20.05 sec, 501 batches, 364819 spikes
23.49 sec, 596 batches, 443026 spikes
time 39.20, Shifted up/down 596 batches.
0.03 sec, 1 batches, 600 spikes
5.07 sec, 101 batches, 64531 spikes
9.06 sec, 201 batches, 131621 spikes
12.76 sec, 301 batches, 201529 spikes
16.15 sec, 401 batches, 275777 spikes
19.53 sec, 501 batches, 353484 spikes
22.64 sec, 596 batches, 429278 spikes
Elapsed time is 0.000074 seconds.
Index in position 1 is invalid. Array indices must be positive integers or logical values.

Error in getMeWtW (line 30)
iList = int32(gpuArray(isort(iNear, :))); % return the list of pairs for each template

Error in trackAndSort (line 100)
[WtW, iList] = getMeWtW(single(W), single(U), Nnearest);

Error in ksGUI/runSpikesort (line 792)
[obj.rez, st3, tF] = trackAndSort(obj.rez);

Error in ksGUI/runAll (line 628)
obj.runSpikesort;

Error in ksGUI>@(,)obj.runAll() (line 319)
'Callback', @(,)obj.runAll());

Error while evaluating UIControl Callback.

[alejandrotorrado]

I believe this is related to the use of linear probes, see issue #338.
I was running into the same issue with 16-channel linear Neuronexus probes.

[shirquinn]

I tried the fix offered in issue #338 and the sorting worked fine, thanks

[marius10p]

That fix should be fine. I pushed an official fix now setting the minimum horizontal pitch to 5um.