[MRG] Reorganize creating coordinates. #122

jasmainak · 2020-07-24T16:19:22Z

This PR achieves few improvements:

Create all the coordinates in one function so that you modify pos_dict in only place not multiple methods
Reuse variables such as xrange
Rename xrange to xxrange since xrange is a Python builtin
Take _create_coords out of the class so it is explicit what goes in for this computation and what comes out (Zen of Python says: "Explicit is better than implicit.")
Reduce the number of attributes in Network class. No more zdiff attribute and gridpyr attribute

hnn_core/network.py

rythorpe · 2020-07-24T17:37:13Z

hnn_core/network.py

+    n_common_feeds : int
+        The number of common feeds.
+    p_unique_keys : list of str
+        The keys of the dictionary p_unique.


So far we have operated under the assumption that there is only one type of common feed: rhythmic. For the sake of consistency, we may want to treat common feeds the same way we treat unique keys, with the exception that common feeds will always be assigned to a position at the origin.

humm ... okay so how does this relate to the PR here? sorry I am a bit lost. Note unique feeds are also assigned to a position in origin.

Right now, we enter common feeds by supplying n_common_feeds. There is nothing intrinsically wrong with this, however, I think it would be clearer if instead we supplied the argument p_common_keys so that the pos_dict items associated with common feeds have keys that specify the type of common feed (e.g., tonic, rhythmic, etc.)

@jasmainak what do you think about using an input arg similar to p_unique_keys except for the common feeds?

sorry I had missed that previous comment 3 days ago. Actually I like that the common feed has a single key because the position will be the same no matter what the spike pattern of the feed is like. But I agree, there is an inconsistency between unique and common. We should try to fix this everywhere in the codebase as part of a single PR.

what's your opinion on the plot? should I remove the feeds? or how should I fix it?

I guess I don't understand why unique feeds can't then be input the same way (i.e., with just a number to specify the amount of unique feeds) since they are all assigned to the same location anyway.

what's your opinion on the plot? should I remove the feeds? or how should I fix it?

See below for my thoughts on the plot.

We can probably change it, but we need to change it deep inside the code as in here: https://github.com/jonescompneurolab/hnn-core/blob/master/hnn_core/pyramidal.py#L948. The fact that class attributes are used instead of explicitly passed arguments makes the code very tangled and inter-connected :( I opened an issue #123 to investigate this separately

jasmainak · 2020-07-24T19:09:02Z

while I was at it, I also added a tiny method to plot the cell locations. I always wanted to visualize it. Maybe HNN-GUI already has something of this sort?

rythorpe · 2020-07-24T22:21:57Z

while I was at it, I also added a tiny method to plot the cell locations. I always wanted to visualize it. Maybe HNN-GUI already has something of this sort?

I think HNN-GUI did have some sort of model visualization.

I'm not sure I understand why the position of all feeds is the same. From a big-picture model visualization perspective, I would expect the location of distal feeds to be different from the location of proximal feeds. Or is the distance from the granular (proximal) vs. superficial (distal) artificial cells defined somewhere else? If so, we may want to incorporate this into the model visualization somehow so that it doesn't confuse others.

stephanie-r-jones · 2020-07-25T13:08:54Z

Hi - I really like this plot.. and would like to ad it too our website but I wonder why all the feeds are going to the center? Best, steph ******************************************* Stephanie R. Jones, PhD Associate Professor Brown University Department of Neuroscience Stephanie_Jones@Brown.edu https://sites.brown.edu/stephanie-r-jones-lab/ Human Neocortical Neurosolver https://hnn.brown.edu <https://hnn.brown.edu/>

…

On Jul 24, 2020, at 6:22 PM, Ryan Thorpe ***@***.***> wrote: while I was at it, I also added a tiny method to plot the cell locations. I always wanted to visualize it. Maybe HNN-GUI already has something of this sort? I think HNN-GUI did have some sort of model visualization. <https://user-images.githubusercontent.com/15852194/88426659-91fdd300-cdbf-11ea-8c15-0b239972164e.png> I'm not sure I understand why the position of all feeds is the same. From a big-picture model visualization perspective, I would expect the location of distal feeds to be different from the location of proximal feeds. Or is the distance from the granular (proximal) vs. superficial (distal) artificial cells defined somewhere else? If so, we may want to incorporate this into the model visualization somehow so that it doesn't confuse others. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#122 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEX6KSPMFFZTM66G6JGN2HDR5ICRHANCNFSM4PG3V44A>.

jasmainak · 2020-07-25T16:02:37Z

From a big-picture model visualization perspective, I would expect the location of distal feeds to be different from the location of proximal feeds. Or is the distance from the granular (proximal) vs. superficial (distal) artificial cells defined somewhere else?

It's a good point, I am trying to understand this too. From what I understand the inputs can be in an arbitrary point in space. It depends where they are synaptically connected that defines whether they are distal or proximal etc. This diagram does not show synaptic connections. Do you have any suggestions to fix this? I could get rid of the inputs if that makes it clearer.

rythorpe · 2020-07-27T15:47:10Z

It's a good point, I am trying to understand this too. From what I understand the inputs can be in an arbitrary point in space. It depends where they are synaptically connected that defines whether they are distal or proximal etc. This diagram does not show synaptic connections. Do you have any suggestions to fix this? I could get rid of the inputs if that makes it clearer.

My vote would be to remove the feed locations from this plot for now. Since feeds are modeled as artificial cells (i.e., with arbitrary location), it seems fair that our plots would either omit them from a plot of modeled cell locations or visualize the feeds as we have done previously in in the tutorials. The latter option we can be added to hnn-core later if desired.

jasmainak · 2020-07-27T16:19:35Z

sounds good @rythorpe . See a462399. Removed the input feeds now from the plot

jasmainak · 2020-07-27T16:21:05Z

Feel free to merge once the CIs come green!

jasmainak force-pushed the reorganize_coords branch from d56da42 to b122ab1 Compare July 24, 2020 16:20

jasmainak requested review from blakecaldwell and rythorpe July 24, 2020 16:22

jasmainak force-pushed the reorganize_coords branch from b122ab1 to 8216fe2 Compare July 24, 2020 16:25

Reorganize creating coordinates.

be05d96

jasmainak force-pushed the reorganize_coords branch from 8216fe2 to be05d96 Compare July 24, 2020 16:46

rythorpe reviewed Jul 24, 2020

View reviewed changes

hnn_core/network.py Outdated Show resolved Hide resolved

rythorpe reviewed Jul 24, 2020

View reviewed changes

DOC: clarification

18430ce

jasmainak force-pushed the reorganize_coords branch from fa3c40f to 18430ce Compare July 24, 2020 18:26

ENH: plot cells

be9d7be

jasmainak changed the title ~~Reorganize creating coordinates.~~ [MRG] Reorganize creating coordinates. Jul 24, 2020

jasmainak mentioned this pull request Jul 27, 2020

investigate why pos_dict needs a separate key for each spike pattern #123

Closed

MAINT: remove feed locations that are arbitrary

a462399

rythorpe merged commit f31cec2 into jonescompneurolab:master Jul 27, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[MRG] Reorganize creating coordinates. #122

[MRG] Reorganize creating coordinates. #122

jasmainak commented Jul 24, 2020 •

edited

rythorpe Jul 24, 2020

jasmainak Jul 24, 2020

rythorpe Jul 24, 2020

rythorpe Jul 27, 2020

jasmainak Jul 27, 2020

rythorpe Jul 27, 2020

rythorpe Jul 27, 2020

jasmainak Jul 27, 2020

jasmainak commented Jul 24, 2020 •

edited

rythorpe commented Jul 24, 2020

stephanie-r-jones commented Jul 25, 2020 via email

jasmainak commented Jul 25, 2020

rythorpe commented Jul 27, 2020

jasmainak commented Jul 27, 2020

jasmainak commented Jul 27, 2020

[MRG] Reorganize creating coordinates. #122

[MRG] Reorganize creating coordinates. #122

Conversation

jasmainak commented Jul 24, 2020 • edited

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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jasmainak commented Jul 24, 2020 • edited

rythorpe commented Jul 24, 2020

stephanie-r-jones commented Jul 25, 2020 via email

jasmainak commented Jul 25, 2020

rythorpe commented Jul 27, 2020

jasmainak commented Jul 27, 2020

jasmainak commented Jul 27, 2020

jasmainak commented Jul 24, 2020 •

edited

jasmainak commented Jul 24, 2020 •

edited