WIP: Add ability to record LFP's #150

jasmainak · 2020-08-26T01:52:56Z

is this the right file?

closes #68

I just copied it for now. I'll work on it to improve and understand it.

blakecaldwell · 2020-08-26T18:18:13Z

Yes, this is the right file for LFP. Calculating CSD is here too. However, plotting CSD (time vs. depth) is kind of complicated and will probably look very different. Perhaps this PR can just be for LFP.

jasmainak · 2020-09-01T03:23:41Z

okay, I improved the implementation of PSA and LSA here. Code looks much cleaner if you use numpy.

@blakecaldwell do we have some ground truth? I don't know if we're getting the right result when I run the code.

jasmainak · 2020-09-01T03:25:19Z

hnn_core/lfp.py

+        """Enables fast calculation of transmembrane current (nA) at
+           each segment."""
+        h.cvode.use_fast_imem(1)
+        self.bscallback = h.cvode.extra_scatter_gather(0, self.callback)


I had to replace the line here with this because my version of Neuron doesn't have the function. Maybe @samnemo do you know what's the right thing to do here?

@jasmainak I'm guessing you saw this post?
https://www.neuron.yale.edu/phpBB/viewtopic.php?p=15713#p15713

cvode.extra_scatter_gather seems to replace the cvode.beforestep_py functionality? Anyway, I can't find it in the neuron documentation anymore.

yeah, I figured from the API documentation that this must be the function :)

jasmainak · 2020-09-01T04:30:59Z

Here is the plot of the LFP for the two methods

they match exactly. Does it look correct to people? If so, I'll go ahead and add tests and example

Feel free to pull the branch and play with it.

blakecaldwell · 2020-09-08T13:30:51Z

they match exactly. Does it look correct to people? If so, I'll go ahead and add tests and example

The correctness of these depends on the coordinates of the electrode and cells around it (i.e. the LFP above and below the soma will be drastically different, opposite I think). Probably the best way to verify correctness is to calculate LFP at various points around a SINGLE neuron. Many Einevoll papers did this and it'd be excellent to do the same with our code. This is large enough to be a separate pull issue (perhaps a hackathon issue).

Regarding point source vs. line source: it makes sense that they would be the same with a compartmental model. Line source is said to be more appropriate for compartmental models. However, I don't know which cases the point source approximation would be different. It'd be a worthwhile learning experience to investigate this (for me or anyone).

ntolley · 2020-09-08T14:03:58Z

For validating the "biophysical intuition" a plot like this from may be useful:
source

jasmainak · 2020-09-14T21:12:23Z

@ntolley looks like you had a very productive hackathon. I just added the code as a commit here so as to not lose it. Would love to learn from you what you did. Any way we can add some print statements so I know how far the simulation went?

ntolley · 2020-09-15T03:37:07Z

I can see about implementing that. Just to make sure I understand, do you mean print statements similar to the simulate() function in parallel_backends.py?

jasmainak · 2020-09-15T03:40:27Z

yep ... something similar :) You can push straight to this branch. My patience runs out after a script runs for 30 seconds, specially if I see no output.

Also before we merge, do make sure to remind me to credit everyone involved in writing the code.

jasmainak · 2021-03-11T00:19:35Z

@ntolley let's try to revive this PR once you are done with #276

ntolley · 2021-03-11T02:18:38Z

Read my mind, I was actually thinking about this PR today as well. We can discuss later, but I think the most straightforward implementation will be passing a list of tuples containing xyz coordinates. Something like:

electrode_list = [(x1, y1, z1), (x2, y2, z2), ...]
net.record_lfp(electrode_list, method='lsa')

No idea how complicated it may get with parallelization, but we'll see!

jasmainak · 2021-03-11T03:41:17Z

That sounds good! Some of the code here which takes a long time to run can be moved to demo notebooks/gists elsewhere.

ntolley · 2021-03-27T21:35:48Z

Very rough initial pass at this. Currently electrode positions are specified as a list of tuples under net.lfp_electrode_pos. After running a simulation the output is stored under net.lfp:

While the Network class is getting pretty bloated, this seems like a natural place to store electrode information at the moment. Absolutely no idea if this is being calculated correctly, we'll need to think of some good tests to validate the code.

jasmainak · 2021-03-27T21:45:27Z

Agree about storing it in net. However, you might want to "reset" it before the next simulation. See discussion with @cjayb and @blakecaldwell about simulation times

jasmainak · 2021-03-27T21:47:53Z

examples/plot_simulate_lfp.py

+
+h.tstop = 500.0
+lfp_rec = {'lsa': ([], []), 'psa': ([], [])}
+for method in ['lsa', 'psa']:


I think this script could go into @cjayb 's repo: https://github.com/cjayb/hnn-core-examples ? Because it runs pretty slow ... could be a nice place to host some of the longer examples

jasmainak · 2021-03-27T21:50:18Z

hnn_core/lfp.py

+    from hnn_core.pyramidal import L5Pyr
+    from hnn_core.network_builder import load_custom_mechanisms
+
+    load_custom_mechanisms()
+    cell = L5Pyr()
+
+    h.load_file("stdgui.hoc")
+    h.cvode_active(1)
+
+    ns = h.NetStim()
+    ns.number = 10
+    ns.start = 100
+    ns.interval = 50.0
+    ns.noise = 0.  # deterministic
+
+    nc = h.NetCon(ns, cell.synapses['apicaltuft_ampa'])
+    nc.weight[0] = 0.001
+
+    h.tstop = 2000.0
+
+    for method in ['lsa', 'psa']:
+        elec = LFPElectrode([0, 100.0, 100.0], pc=h.ParallelContext(),
+                            method='psa')
+        elec.setup()
+        elec.LFPinit()
+        h.run()


This could be factored into a unit test. If you read the reference paper, I think you might find some corner cases in which the LSA and PSA method give the exact same result ... (I suspect)

Is this the sort of setup you're thinking should go in test_lfp,py for testing the code directly?

I think it will be good to offload the net.add_electrode() tests there as well. However, for a full simulation test I think I will add LFP recordings to test_parallel_ backends.py and compare between joblib and MPI.

yeah one test like that which checks that LSA and PSA give roughly equal results. I think they would be more equal the further you go away from the cell.

And then one where you could just look at the vres and see if it's computed correctly. I'm worried that lines like this might be incorrect (see also this figure). You can give different positions and see if the vres is computed correctly -- e.g., if it's above a segment vs if it's outside the segment etc.

jasmainak · 2021-03-27T21:55:55Z

How about renaming pos_dict to be pos_cells (slight misnomer but I can live with it ...) and then calling the LFP position pos_lfp. That way it's more discoverable. In Ipython, when you type net.pos and press tab, you'll see both options.

ntolley · 2021-03-28T02:07:18Z

Love the net.pos_lfp suggestion!

Also realized an interesting mismatch, the times recorded for the dipole is one element longer than the LFP recordings (see L55) I believe it has to do with the way the LFP is calculated. Anyone against just copying the first element's value twice?

…d only appears when they exist

blakecaldwell · 2021-04-10T09:46:55Z

hnn_core/lfp.py

+def _get_all_sections(sec_type='Pyr'):
+    ls = h.allsec()
+    ls = [s for s in ls if sec_type in s.name()]
+    return ls


Maybe an opportunity to replace these with better names?

blakecaldwell · 2021-04-10T09:48:03Z

hnn_core/lfp.py

+    use_point : bool
+        Whether to do a point source approximation
+        for extracellular currents.


Docstring needs updating

blakecaldwell · 2021-04-10T09:52:50Z

hnn_core/lfp.py

+        # verify sum i_membrane_ == stimulus
+        # s = pc.allreduce(imem_vec.sum(), 1)
+        # if rank == 0: print pc.t(0), s
+
+        # sum up the weighted i_membrane_. Result in vx
+        # rx.mulv(imem_vec, vx)


The allreduce method is typically more efficient than gathering and then summing. Maybe the commented code won't work anymore (b/c switch to imem_ptrvec), but something to consider.

blakecaldwell · 2021-04-10T09:55:42Z

hnn_core/network.py

+        assert sigma > 0.0
+        _validate_type(method, str, 'method')
+        _check_option('method', method, ['psa', 'lsa'])
+        if isinstance(electrode_pos, tuple):
+            electrode_pos = [electrode_pos]
+        for e_pos in electrode_pos:
+            assert len(e_pos) == 3


More assert statements that should be converted to Exceptions.

blakecaldwell · 2021-04-10T10:00:14Z

hnn_core/network_builder.py

+            # elec.pc.allreduce(elec.lfp_v, 1)
+            # lfp_py.append(elec_list[pos].lfp_t.to_python())


Maybe you meant to remove these? Also, realize that you may have been testing out allreduce where my comment above about efficiency. Please ignore if that's the case. Main point is just that commented lines could be removed.

jasmainak · 2021-04-24T01:09:05Z

@ntolley do you have any local changes here? Can I work on this PR if you're busy with other PRs?

ntolley · 2021-04-24T01:38:13Z

No local changes! Mostly paused this to focus on the connectivity PR's. I think the last item on the list was adding tests to ensure the LFP calculation was correct? My initial thought would be to instantiate a point neuron and/or cylinder with nrn.

cjayb · 2021-04-26T15:32:12Z

@jasmainak are you working on this? Looks great! I'd like to propose the following for moving this ahead (and creating a test)

use new connectivity API to prune all connections (non-connected cells, no spiking, no IPSPs)
send (weak) synchronous EPSPs to L5 distal apical dendrites (for test: 3 x 3 grid or even a single cell would suffice)
plot LFP as a function of depth near the cell (should go from negative to positive at soma)
calculate far-field LFP (~10x distal-to-apical distance), which should reproduce a dipole field (see attached, calculated analytically, could be used in a test)

jasmainak · 2021-04-26T15:35:03Z

I don't think I'll get to this in the coming week. If you or @ntolley want to take over, please feel free!

cjayb · 2021-04-26T18:32:50Z

Hmm, it's not clear to me how we can proceed to calculate LFP without implementing physical dimensions on x and y, in addition to the current z? Relative distances appear in the equations for "transfer resistance". Has anyone considered this? I believe NetPyNE moves the HNN network cells around to enable LFP calculations. What's the principled way of doing it, what's the "lateral extent" of the HNN network?

cjayb · 2021-04-27T18:01:36Z

Are our in-plane coordinates actually in physical (micron) units?! I just assumed they weren't I guess... There's something weird going on with the psa distance calculation, but I can't quite put my finger on it... I think I need to sit down and follow the construction of the neuron sections.

ntolley · 2021-04-28T00:21:31Z

Are our in-plane coordinates actually in physical (micron) units?! I

I see why in plane units may be arbitrary for calculating the current dipole, but HNN modeling has previously published with LFP/CSD analysis: https://www.pnas.org/content/113/33/E4885 (figure 9)

@stephanie-r-jones @samnemo do you know if there are real units for the x/y position of cells?

samnemo · 2021-04-28T01:08:38Z

As far as I know x,y position of cells in original model were in 'arbitrary' units primarily used to scale synaptic weights as a function of distance. So original x,y positions should not be considered to be in microns.

…

________________________________ From: Nick Tolley ***@***.***> Sent: Tuesday, April 27, 2021 8:21 PM To: jonescompneurolab/hnn-core ***@***.***> Cc: Neymotin, Samuel (NKI) ***@***.***>; Mention ***@***.***> Subject: Re: [jonescompneurolab/hnn-core] WIP: Add ability to record LFP's (#150) ATTENTION: This email came from an external source. Do not open attachments or click on links from unknown senders or unexpected emails. Are our in-plane coordinates actually in physical (micron) units?! I I see why in plane units may be arbitrary for calculating the current dipole, but HNN modeling has previously published with LFP/CSD analysis: https://www.pnas.org/content/113/33/E4885 @stephanie-r-jones<https://protect2.fireeye.com/v1/url?k=668f8cbf-3914b478-668d758a-000babd9f75c-b2a0885b68cfea2d&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fstephanie-r-jones> @samnemo<https://protect2.fireeye.com/v1/url?k=889af721-d701cfe6-88980e14-000babd9f75c-bf540c1064fc5ea9&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fsamnemo> do you know if there are real units for the x/y position of cells? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<https://protect2.fireeye.com/v1/url?k=e30b4769-bc907fae-e309be5c-000babd9f75c-3131f95b24be9a23&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fjonescompneurolab%2Fhnn-core%2Fpull%2F150%23issuecomment-828047661>, or unsubscribe<https://protect2.fireeye.com/v1/url?k=7e68af16-21f397d1-7e6a5623-000babd9f75c-39f543e44eb1dda9&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACS5H2ESZ6F4SH2RSOS63S3TK5IJZANCNFSM4QLI44CA>.

________________________________ IMPORTANT NOTICE: This e-mail is meant only for the use of the intended recipient. It may contain confidential information which is legally privileged or otherwise protected by law. If you received this e-mail in error or from someone who was not authorized to send it to you, you are strictly prohibited from reviewing, using, disseminating, distributing or copying the e-mail. PLEASE NOTIFY US IMMEDIATELY OF THE ERROR BY RETURN E-MAIL AND DELETE THIS MESSAGE FROM YOUR SYSTEM. Thank you for your cooperation.

samnemo · 2021-04-28T01:11:01Z

Data from that paper used laminar LFP/CSD but afaik not the model. Also, NEURON internally has coordinates for cells but that doesn't mean those positions influence the circuit dynamics.

…

________________________________ From: Neymotin, Samuel (NKI) ***@***.***> Sent: Tuesday, April 27, 2021 9:08 PM To: jonescompneurolab/hnn-core ***@***.***>; jonescompneurolab/hnn-core ***@***.***> Cc: Mention ***@***.***> Subject: Re: [jonescompneurolab/hnn-core] WIP: Add ability to record LFP's (#150) As far as I know x,y position of cells in original model were in 'arbitrary' units primarily used to scale synaptic weights as a function of distance. So original x,y positions should not be considered to be in microns.

________________________________ From: Nick Tolley ***@***.***> Sent: Tuesday, April 27, 2021 8:21 PM To: jonescompneurolab/hnn-core ***@***.***> Cc: Neymotin, Samuel (NKI) ***@***.***>; Mention ***@***.***> Subject: Re: [jonescompneurolab/hnn-core] WIP: Add ability to record LFP's (#150) ATTENTION: This email came from an external source. Do not open attachments or click on links from unknown senders or unexpected emails. Are our in-plane coordinates actually in physical (micron) units?! I I see why in plane units may be arbitrary for calculating the current dipole, but HNN modeling has previously published with LFP/CSD analysis: https://www.pnas.org/content/113/33/E4885 @stephanie-r-jones<https://protect2.fireeye.com/v1/url?k=668f8cbf-3914b478-668d758a-000babd9f75c-b2a0885b68cfea2d&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fstephanie-r-jones> @samnemo<https://protect2.fireeye.com/v1/url?k=889af721-d701cfe6-88980e14-000babd9f75c-bf540c1064fc5ea9&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fsamnemo> do you know if there are real units for the x/y position of cells? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<https://protect2.fireeye.com/v1/url?k=e30b4769-bc907fae-e309be5c-000babd9f75c-3131f95b24be9a23&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fjonescompneurolab%2Fhnn-core%2Fpull%2F150%23issuecomment-828047661>, or unsubscribe<https://protect2.fireeye.com/v1/url?k=7e68af16-21f397d1-7e6a5623-000babd9f75c-39f543e44eb1dda9&q=1&e=a8dd2442-b933-4323-949f-ae76ee01f0a2&u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACS5H2ESZ6F4SH2RSOS63S3TK5IJZANCNFSM4QLI44CA>.

________________________________ IMPORTANT NOTICE: This e-mail is meant only for the use of the intended recipient. It may contain confidential information which is legally privileged or otherwise protected by law. If you received this e-mail in error or from someone who was not authorized to send it to you, you are strictly prohibited from reviewing, using, disseminating, distributing or copying the e-mail. PLEASE NOTIFY US IMMEDIATELY OF THE ERROR BY RETURN E-MAIL AND DELETE THIS MESSAGE FROM YOUR SYSTEM. Thank you for your cooperation.

jasmainak · 2021-04-28T16:33:25Z

hnn_core/network.py

+        ----------
+        electrode_pos : tuple | list of tuple
+            Coordinates specifying the position for LFP electrodes in
+            the form of (x, y, z).


I think @cjayb is right that we should specify the units here. Should be micrometer?

Also (x, y, z) is ambiguous: NEURON objects are (x, z, y) in my mind (apical dendrite oriented along z)

jasmainak · 2021-04-28T16:34:14Z

examples/plot_simulate_lfp.py

+plt.figure()
+trial_idx = 0
+plt.plot(times, net.lfp[0]['data'][trial_idx])
+plt.plot(times, net.lfp[1]['data'][trial_idx])


add units in label ... millivolt?

jasmainak · 2021-04-28T16:39:33Z

So original x,y positions should not be considered to be in microns.

humm ... okay then I'm confused too. If the x, y positions are not in microns, doesn't that mean that the LFP calculation will be in the wrong units/scale? Maybe the influence of other cells is negligible because they are "far enough" ... ?

samnemo · 2021-04-28T17:34:04Z

1. original x,y positions were used for wiring the neurons in the circuit via synapses (distances were used to set snypatic delays and weights and nothing else in the ORIGINAL model afaik) 2. LFP was never released to public through HNN GUI - needs some work as you're all noticing - part of which involves making sure units are consistent

…

________________________________ From: Mainak Jas ***@***.***> Sent: Wednesday, April 28, 2021 12:39 PM To: jonescompneurolab/hnn-core ***@***.***> Cc: Neymotin, Samuel (NKI) ***@***.***>; Mention ***@***.***> Subject: Re: [jonescompneurolab/hnn-core] WIP: Add ability to record LFP's (#150) ATTENTION: This email came from an external source. Do not open attachments or click on links from unknown senders or unexpected emails. So original x,y positions should not be considered to be in microns. humm ... okay then I'm confused too. If the x, y positions are not in microns, doesn't that mean that the LFP calculation will be in the wrong units/scale? Maybe the influence of other cells is negligible because they are "far enough" ... ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<https://protect2.fireeye.com/v1/url?k=93df200f-cc441935-93ddd93a-ac1f6b44fec6-5d60d0d64ad4a052&q=1&e=cdd2d0bf-6765-4209-b7a9-850c60417d6f&u=https%3A%2F%2Fgithub.com%2Fjonescompneurolab%2Fhnn-core%2Fpull%2F150%23issuecomment-828604002>, or unsubscribe<https://protect2.fireeye.com/v1/url?k=282ef002-77b5c938-282c0937-ac1f6b44fec6-5e3d108464530bda&q=1&e=cdd2d0bf-6765-4209-b7a9-850c60417d6f&u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACS5H2GWMGKZXCQG57DTCGTTLA25JANCNFSM4QLI44CA>.

________________________________ IMPORTANT NOTICE: This e-mail is meant only for the use of the intended recipient. It may contain confidential information which is legally privileged or otherwise protected by law. If you received this e-mail in error or from someone who was not authorized to send it to you, you are strictly prohibited from reviewing, using, disseminating, distributing or copying the e-mail. PLEASE NOTIFY US IMMEDIATELY OF THE ERROR BY RETURN E-MAIL AND DELETE THIS MESSAGE FROM YOUR SYSTEM. Thank you for your cooperation.

cjayb · 2021-04-28T18:25:02Z

Thanks @samnemo . It's more than "just" ensuring consistency. We have to define the physical distance between cells within the "HNN column". Given that our "cells" actually represent quite large assemblies, we should probably not use biological cell-to-cell distances (on the order of microns), but at least an order of magnitude or two larger. We could base it on (Murakami & Okada, 2015)'s 1-2 nAm/mm2 for maximal neocortical tissue dipole moment. I haven't done the math yet to figure out what that implies for us, assuming the usual 10 nAm dipole moment said to be detectable using MEG.

I really like the inclusion of LFP-like signals into hnn-core, but these should not be confused with those obtainable in more realistic models. I'm struggling to get the psa-method to produce the expected dipolar field patterns, though. I need to backtrack the imem-currents coming from NEURON. It's probably "just" a matter of getting the distance calculation right, since the target equation is super-simple: $\sum_n \frac{i_{mem}}{4 \pi \sigma R}$ , where n are the sections. But we will need some physical units for all our coordinates, and to keep track of the xyz-coordinates in various places.

(Murakami & Okada, 2015): doi:10.1016/j.neuroimage.2015.02.003

jasmainak · 2021-04-29T21:54:11Z

I think we should just use physical distances based on known numbers from microscopic measurements. I'm not sure we can simply scale the LFP signal if the cell density changes ...

Note that if you get something less than 1 nAm / mm^2, it could simply mean that the cells are not firing synchronously. So, I'm not sure you can calculate backwards the cell density from the signal amplitude. The Okada constant is a maximum. Rather, given a certain signal magnitude from our simulated signals, you can say the signal must be due to x neurons (50,000 for 10 nAm?) firing synchronously. This means that a patch of the cortex that is at least (x / cell density) mm^2 is activated. Interestingly, you can come to the same conclusions from the "scaling factor" in HNN.

cjayb · 2021-05-09T19:15:14Z

Quick shoutout here: I think I've found a bug in the current implementation. I need to investigate further, but will return with more soonest. I've already taken this branch and rebased against master, hope to have an analysis (and PR) ready by Tuesday.

jasmainak · 2021-06-17T19:38:07Z

closed by #329

jasmainak changed the title ~~Start LFP~~ WIP: Start LFP Aug 26, 2020

jasmainak force-pushed the lfp branch from 69ecfc0 to 076d943 Compare August 27, 2020 20:22

jasmainak mentioned this pull request Aug 27, 2020

[MRG] MAINT: move neuron -> neuron_network #158

Merged

jasmainak force-pushed the lfp branch from 827911d to 69abb8b Compare August 31, 2020 22:04

jasmainak commented Sep 1, 2020

View reviewed changes

jasmainak force-pushed the lfp branch from fd22ee6 to 55e1cc5 Compare September 1, 2020 03:37

blakecaldwell mentioned this pull request Sep 8, 2020

move analysis code from HNN GUI to start hnn_core.analysis submodule #139

Closed

6 tasks

jasmainak mentioned this pull request Sep 8, 2020

add CSD plot for comparing against animal studies #68

Closed

blakecaldwell mentioned this pull request Sep 19, 2020

[WIP] Align laminar LFP to L5 features jonescompneurolab/hnn#57

Closed

jasmainak added this to the 0.2 milestone Jan 18, 2021

ntolley force-pushed the lfp branch from 5829b76 to 8855eba Compare March 27, 2021 02:52

ntolley changed the title ~~WIP: Start LFP~~ WIP: Add ability to record LFP's Mar 27, 2021

jasmainak commented Mar 27, 2021

View reviewed changes

ntolley added 3 commits April 5, 2021 19:54

Change lfp dict key to 'data', modify plot_cells() so electrode legen…

93240bc

…d only appears when they exist

Add tests for net.add_electrode

cf2e5d4

Move LFP tests to separate file

aab9d28

ntolley force-pushed the lfp branch from 055962e to aab9d28 Compare April 5, 2021 23:58

ntolley added 4 commits April 5, 2021 20:35

Clean up lfp tests

f800779

Add tests to compare lfp across parallel backends

bece3a5

Replace lfp backends comparison with assert_allclose()

bec6018

move test from parallel_backends

881b860

blakecaldwell reviewed Apr 10, 2021

View reviewed changes

jasmainak commented Apr 28, 2021

View reviewed changes

cjayb mentioned this pull request May 2, 2021

explore response properties of individual HNN network cells #323

Open

4 tasks

ntolley mentioned this pull request May 4, 2021

[MRG] Improve cell classes #322

Merged

6 tasks

cjayb mentioned this pull request May 10, 2021

[MRG] Improved LFP calculation #329

Merged

12 tasks

ntolley closed this Jun 17, 2021

		# elec.pc.allreduce(elec.lfp_v, 1)
		# lfp_py.append(elec_list[pos].lfp_t.to_python())

WIP: Add ability to record LFP's #150

WIP: Add ability to record LFP's #150

Conversation

jasmainak commented Aug 26, 2020 • edited Loading

blakecaldwell commented Aug 26, 2020

jasmainak commented Sep 1, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jasmainak commented Sep 1, 2020

blakecaldwell commented Sep 8, 2020 • edited Loading

ntolley commented Sep 8, 2020 • edited Loading

jasmainak commented Sep 14, 2020

ntolley commented Sep 15, 2020

jasmainak commented Sep 15, 2020

jasmainak commented Mar 11, 2021

ntolley commented Mar 11, 2021 • edited Loading

jasmainak commented Mar 11, 2021

ntolley commented Mar 27, 2021

jasmainak commented Mar 27, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ntolley Apr 5, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jasmainak commented Mar 27, 2021

ntolley commented Mar 28, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jasmainak commented Apr 24, 2021

ntolley commented Apr 24, 2021

cjayb commented Apr 26, 2021

jasmainak commented Apr 26, 2021

cjayb commented Apr 26, 2021

cjayb commented Apr 27, 2021

ntolley commented Apr 28, 2021 • edited Loading

samnemo commented Apr 28, 2021 via email

samnemo commented Apr 28, 2021 via email

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jasmainak commented Apr 28, 2021

samnemo commented Apr 28, 2021 via email

cjayb commented Apr 28, 2021

jasmainak commented Apr 29, 2021 • edited Loading

cjayb commented May 9, 2021

jasmainak commented Jun 17, 2021

jasmainak commented Aug 26, 2020 •

edited

Loading

blakecaldwell commented Sep 8, 2020 •

edited

Loading

ntolley commented Sep 8, 2020 •

edited

Loading

ntolley commented Mar 11, 2021 •

edited

Loading

ntolley Apr 5, 2021 •

edited

Loading

ntolley commented Mar 28, 2021 •

edited

Loading

ntolley commented Apr 28, 2021 •

edited

Loading

jasmainak commented Apr 29, 2021 •

edited

Loading