Spike time tiling coefficient produces values larger than 1 #563

Celine-Serry · 2023-05-12T07:20:51Z

Hi,

Im working with the spike time tiling coefficient and have noticed that some values are larger than 1, while I thought the maximum value was 1. For example:
spiketrain E7:
array([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3,
97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1])

spiketrain E3:
array([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3,
97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1])

result in an STTC of 1.063003420467317

While spiketrain E8
array([ 20646.8, 25875.1, 26154.4, 35121. , 55909.7, 79164.8,
110849.8, 117484.1, 3731.5, 4213.9, 119995.1, 123748.1,
171016.8, 172989. , 185145.2, 12043.5, 185995.9, 186740.1,
12629.8, 23394.3, 34993.2])

spiketrain B3
array([ 10600.7, 19699.6, 22803. , 40769.3, 121385.7, 127402.9,
130829.2, 134363.8, 1193.5, 8012.7, 142037.3, 146628.2,
165925.3, 168489.3, 175194.3, 10339.8, 178676.4, 180807.2,
201431.3, 22231.1, 38113.4])

Are less synchronous, but have an STTC of 2.2794178907759863, while I thought the maximum value would be 1, suggesting perfect synchronicity?

I calculated the STTC as: sttc = spike_time_tiling_coefficient(spiketrain1, spiketrain2, dt), with dt being 0.10 * pq.s. I also I made sure I used Neo in the right way to make the spike trains.

Is it possible that there is a bug in the code for calculating the spike time tiling coefficient?

Moritz-Alexander-Kern · 2023-05-12T08:00:59Z

Hey @Celine-075 ,
thanks for reaching out to us.

I've created to the following minimal example to reproduce this, I get the results:

STTC for E7 and E3: 1.152482370309056
STTC for E8 and B3: 2.2166419047619055

import neo

import quantities as pq

from elephant.spike_train_correlation import spike_time_tiling_coefficient

spiketrain_E7 = neo.SpikeTrain([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3,
97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1], units='ms', t_stop=160000*pq.ms)

spiketrain_E3 = neo.SpikeTrain([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3,
97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1], units='ms', t_stop=160000*pq.ms)

print(spike_time_tiling_coefficient(spiketrain_E7, spiketrain_E3, dt=0.10 * pq.s))
# result: 1.152482370309056


spiketrain_E8 = neo.SpikeTrain([ 20646.8, 25875.1, 26154.4, 35121. , 55909.7, 79164.8,
110849.8, 117484.1, 3731.5, 4213.9, 119995.1, 123748.1,
171016.8, 172989. , 185145.2, 12043.5, 185995.9, 186740.1,
12629.8, 23394.3, 34993.2], units='ms', t_stop=210000*pq.ms)

spiketrain_B3 = neo.SpikeTrain([ 10600.7, 19699.6, 22803. , 40769.3, 121385.7, 127402.9,
130829.2, 134363.8, 1193.5, 8012.7, 142037.3, 146628.2,
165925.3, 168489.3, 175194.3, 10339.8, 178676.4, 180807.2,
201431.3, 22231.1, 38113.4], units='ms', t_stop=210000*pq.ms)

print(spike_time_tiling_coefficient(spiketrain_E8, spiketrain_B3, dt=0.10 * pq.s))
# result: 2.2166419047619055

Which seems very close to what you get, did I reproduce this correctly?

Celine-Serry · 2023-05-12T08:51:58Z

Hi, Thank you for answering on such short notice! Yes, its somewhat similar to what I got, although not entirely the same but that probably has to do with t_stop being 160 000 in your example, while I used 300 000, but you also produce a number > 1 (e.g., between E8 and B3, it is 2.2 in your example). Is this correct? Shouldn’t it be a number between -1 and +1? Kind Regards, Celine Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Moritz ***@***.***> Sent: Friday, May 12, 2023 10:01 AM To: ***@***.***> Cc: ***@***.***>; ***@***.***> Subject: Re: [NeuralEnsemble/elephant] Spike time tiling coefficient produces values larger than 1 (Issue #563) Hey @Celine-075<https://github.com/Celine-075> , thanks for reaching out to us. I've created to the following minimal example to reproduce this, I get the results: * STTC for E7 and E3: 1.152482370309056 * STTC for E8 and B3: 2.2166419047619055 import neo import quantities as pq from elephant.spike_train_correlation import spike_time_tiling_coefficient spiketrain_E7 = neo.SpikeTrain([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3, 97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1], units='ms', t_stop=160000*pq.ms) spiketrain_E3 = neo.SpikeTrain([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3, 97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1], units='ms', t_stop=160000*pq.ms) print(spike_time_tiling_coefficient(spiketrain_E7, spiketrain_E3, dt=0.10 * pq.s)) # result: 1.152482370309056 spiketrain_E8 = neo.SpikeTrain([ 20646.8, 25875.1, 26154.4, 35121. , 55909.7, 79164.8, 110849.8, 117484.1, 3731.5, 4213.9, 119995.1, 123748.1, 171016.8, 172989. , 185145.2, 12043.5, 185995.9, 186740.1, 12629.8, 23394.3, 34993.2], units='ms', t_stop=210000*pq.ms) spiketrain_B3 = neo.SpikeTrain([ 10600.7, 19699.6, 22803. , 40769.3, 121385.7, 127402.9, 130829.2, 134363.8, 1193.5, 8012.7, 142037.3, 146628.2, 165925.3, 168489.3, 175194.3, 10339.8, 178676.4, 180807.2, 201431.3, 22231.1, 38113.4], units='ms', t_stop=210000*pq.ms) print(spike_time_tiling_coefficient(spiketrain_E8, spiketrain_B3, dt=0.10 * pq.s)) # result: 2.2166419047619055 Which seems very close to what you get, did I reproduce this correctly? — Reply to this email directly, view it on GitHub<#563 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A6MBEVZIWBG7SFD3ZXNIYTTXFXU4NANCNFSM6AAAAAAX7EADDA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

Moritz-Alexander-Kern · 2023-05-12T14:39:49Z

Hey @Celine-075 ,

A reworked version of the code is proposed in PR #564 .

The results I get now are:

STTC for E7 and E3: 1.0
STTC for E8 and B3: -0.02

You can test this version of the code by installing:

pip install git+https://github.com/INM-6/elephant.git@fix/sttc

Then run your code and see if the results you obtain are within [-1 1].
I agree the values should be between -1 and 1.

Celine-Serry · 2023-05-12T14:40:48Z

Thank you so much! I will check this right away. Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Moritz ***@***.***> Sent: Friday, May 12, 2023 4:40 PM To: ***@***.***> Cc: ***@***.***>; ***@***.***> Subject: Re: [NeuralEnsemble/elephant] Spike time tiling coefficient produces values larger than 1 (Issue #563) Hey @Celine-075<https://github.com/Celine-075> , A reworked version of the code is proposed in PR #564<#564> . The results I get now are: * STTC for E7 and E3: 1.0 * STTC for E8 and B3: -0.02 You can test this version of the code by installing: pip install ***@***.***/sttc Then run your code and see if the results you obtain are within [-1 1]. I agree the values should be between -1 and 1. — Reply to this email directly, view it on GitHub<#563 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A6MBEV63NV7WOD56QKKJGFLXFZDUBANCNFSM6AAAAAAX7EADDA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

Celine-Serry · 2023-05-12T17:11:44Z

Between E7 and E3 I also get 1.0 now, but for E8 and B3 I get 120.83066666666667

Moritz-Alexander-Kern · 2023-05-12T21:43:15Z

Hey @Celine-075 ,
I ran the example in the following environment using those versions of neo, numpy and quantities: (Python 3.10)

Package     Version  
--------------- -------- -----------------------------
elephant        0.13.0b1 @ branch fix/sttc
Jinja2          3.1.2
neo             0.12.0
numpy           1.24.3
quantities      0.14.1

This time I adjusted t_stop to 300 000 ms.
Now I get:

STTC for E7 and E3: 1.0
STTC for E8 and B3: -0.014

Celine-Serry · 2023-05-13T06:18:23Z

Ah let me check! Thanks for the help! Verzonden vanaf mijn Galaxy

…

-------- Oorspronkelijk bericht -------- Van: Moritz Kern ***@***.***> Datum: 12-05-2023 23:43 (GMT+01:00) Aan: NeuralEnsemble/elephant ***@***.***> Cc: Celine-075 ***@***.***>, Mention ***@***.***> Onderwerp: Re: [NeuralEnsemble/elephant] Spike time tiling coefficient produces values larger than 1 (Issue #563) Hey @Celine-075<https://github.com/Celine-075> , I ran the example in the following environment using those versions of neo, numpy and quantities: Package Version --------------- -------- ----------------------------- elephant 0.13.0b1 @ branch fix/sttc Jinja2 3.1.2 neo 0.12.0 numpy 1.24.3 quantities 0.14.1 This time I adjusted t_stop to 300 000 ms. Now I get: * STTC for E7 and E3: 1.0 * STTC for E8 and B3: -0.014 — Reply to this email directly, view it on GitHub<#563 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A6MBEV4YXOECQJDNUCITHZLXF2VH5ANCNFSM6AAAAAAX7EADDA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

Celine-Serry · 2023-05-13T09:06:58Z

I've updated to the same versions as you showed, but unfortunately I still get a very large value (120.83066666666667). Ive noticed my spiketimes were not sorted so checked if that could be the problem, but that resulted in an even larger number: (645.9386666666666).

Moritz-Alexander-Kern · 2023-05-22T10:28:00Z

Hey @Celine-075 ,

I used the following example to check the difference between using sorted and unsorted spiketimes:
Using the original implementation from elephant v0.12.0, not the fixed one I provided here

import neo

import quantities as pq

from elephant.spike_train_correlation import spike_time_tiling_coefficient

spiketrain_E7 = neo.SpikeTrain([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3,
97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1], units='ms', t_stop=160000*pq.ms)

spiketrain_E3 = neo.SpikeTrain([ 1678. , 23786.3, 34641.8, 71520.7, 73606.9, 78383.3,
97387.9, 144313.4, 4607.6, 19275.1, 152894.2, 44240.1], units='ms', t_stop=160000*pq.ms)
print("results obtained with unsorted spiketrains:")
print(spike_time_tiling_coefficient(spiketrain_E7, spiketrain_E3, dt=0.10 * pq.s))

spiketrain_E8 = neo.SpikeTrain([ 20646.8, 25875.1, 26154.4, 35121. , 55909.7, 79164.8,
110849.8, 117484.1, 3731.5, 4213.9, 119995.1, 123748.1,
171016.8, 172989. , 185145.2, 12043.5, 185995.9, 186740.1,
12629.8, 23394.3, 34993.2], units='ms', t_stop=210000*pq.ms)

spiketrain_B3 = neo.SpikeTrain([ 10600.7, 19699.6, 22803. , 40769.3, 121385.7, 127402.9,
130829.2, 134363.8, 1193.5, 8012.7, 142037.3, 146628.2,
165925.3, 168489.3, 175194.3, 10339.8, 178676.4, 180807.2,
201431.3, 22231.1, 38113.4], units='ms', t_stop=210000*pq.ms)

print(spike_time_tiling_coefficient(spiketrain_E8, spiketrain_B3, dt=0.10 * pq.s))

print("results obtained with sorted spiketrains:")
spiketrain_E7.sort()
spiketrain_E3.sort()
spiketrain_E8.sort()
spiketrain_B3.sort()
print(spike_time_tiling_coefficient(spiketrain_E7, spiketrain_E3, dt=0.10 * pq.s))
print(spike_time_tiling_coefficient(spiketrain_E8, spiketrain_B3, dt=0.10 * pq.s))

The results I obtain are:

results obtained with unsorted spiketrains:
1.152482370309056
2.2166419047619055
results obtained with sorted spiketrains:
1.0
-0.019828095238095245

Maybe you get correct results if you use sorted spiketimes with the original implementation?

Celine-Serry · 2023-05-22T10:30:29Z

Yes I did sort the spiketrains, however, it produced even larger numbers after I did that. I'm also not sure why I dont obtain the same results while running the same versions of the packages ..

…alidation test (#564) * refactor run_P * refactor run_T * add checks for t_start and t_stop in run_t * add input checks and unittests * add regression test for Issue #563 * add validation tests * add check if spike times are sorted, if not sort the spikes

Moritz-Alexander-Kern mentioned this issue May 12, 2023

[Fix] spike time tiling coefficient for unsorted spiketrains, added validation test #564

Merged

2 tasks

Moritz-Alexander-Kern linked a pull request May 12, 2023 that will close this issue

[Fix] spike time tiling coefficient for unsorted spiketrains, added validation test #564

Merged

2 tasks

Moritz-Alexander-Kern added the bug Indicates an unexpected problem or unintended behavior label May 12, 2023

Moritz-Alexander-Kern pushed a commit to INM-6/elephant that referenced this issue May 23, 2023

add regression test for Issue NeuralEnsemble#563

a988016

Moritz-Alexander-Kern closed this as completed in #564 Oct 31, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Spike time tiling coefficient produces values larger than 1 #563

Spike time tiling coefficient produces values larger than 1 #563

Celine-Serry commented May 12, 2023

Moritz-Alexander-Kern commented May 12, 2023

Celine-Serry commented May 12, 2023 via email

Moritz-Alexander-Kern commented May 12, 2023

Celine-Serry commented May 12, 2023 via email

Celine-Serry commented May 12, 2023

Moritz-Alexander-Kern commented May 12, 2023 •

edited

Loading

Celine-Serry commented May 13, 2023 via email

Celine-Serry commented May 13, 2023

Moritz-Alexander-Kern commented May 22, 2023

Celine-Serry commented May 22, 2023

Spike time tiling coefficient produces values larger than 1 #563

Spike time tiling coefficient produces values larger than 1 #563

Comments

Celine-Serry commented May 12, 2023

Moritz-Alexander-Kern commented May 12, 2023

Celine-Serry commented May 12, 2023 via email

Moritz-Alexander-Kern commented May 12, 2023

Celine-Serry commented May 12, 2023 via email

Celine-Serry commented May 12, 2023

Moritz-Alexander-Kern commented May 12, 2023 • edited Loading

Celine-Serry commented May 13, 2023 via email

Celine-Serry commented May 13, 2023

Moritz-Alexander-Kern commented May 22, 2023

Celine-Serry commented May 22, 2023

Moritz-Alexander-Kern commented May 12, 2023 •

edited

Loading