[MRG] DOC: use morlet instead of multitaper #226

jasmainak · 2020-12-20T05:42:52Z

closes #223

I was messing around with the gamma examples and realized that HNN-GUI uses a morlet wavelet for TFR. This is an attempt to make them more similar and add a colorbar.

While I was at it, I also updated the tonic input example and made a couple of fixes there.

jasmainak · 2020-12-20T05:46:01Z

examples/plot_simulate_gamma.py


 fig, axes = plt.subplots(2, 1, sharex=True, figsize=(6, 6))
-dpls[0].plot(ax=axes[0], layer='agg', show=False)
+axes[0].plot(times, data)


I was in two minds whether to add tstart and tstop to the plotting routine or just directly use matplotlib to plot. I settled with the later ... the utility of a plotting routine for Dipole object seems a bit limited to me unless we come up with more use cases

Are you thinking of just removing the plot() methods from Dipole? I agree it's fairly unnecessary considering most people are capable of making a line plot.

Now there's certainly a case to be made for more visualization options (multi-trial with mean dipole overlaid for example), but that would be better suited for viz.py.

right, so perhaps we need a plot_dipole function? Still it's rather straightforward to do with 2 lines of Python code, but maybe we can add functionality later.

jasmainak · 2020-12-20T05:49:17Z

examples/plot_simulate_gamma.py

+                         n_cycles=n_cycles, output='power')
+
+im = axes[1].pcolormesh(times, freqs, power[0, 0, ...], cmap='RdBu_r',
+                        shading='auto')


shading='auto' was needed to prevent a deprecation warning in matplotlib.

codecov-io · 2020-12-20T05:52:49Z

Codecov Report

Merging #226 (e49b95b) into master (31b2a93) will increase coverage by 0.04%.
The diff coverage is 87.50%.

@@            Coverage Diff             @@
##           master     #226      +/-   ##
==========================================
+ Coverage   66.84%   66.88%   +0.04%     
==========================================
  Files          21       21              
  Lines        2277     2283       +6     
==========================================
+ Hits         1522     1527       +5     
- Misses        755      756       +1

Impacted Files	Coverage Δ
hnn_core/network.py	`50.36% <83.33%> (+0.36%)`	⬆️
hnn_core/tests/test_feed.py	`100.00% <100.00%> (ø)`

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 31b2a93...602a157. Read the comment docs.

jasmainak · 2020-12-20T16:03:47Z

hnn_core/network.py

+            feed_times = self.feed_times
+        else:
+            feed_times = dict()
+        feed_times.update({src_type: list() for src_type in src_types})


this is a bit of a spaghetti anti-pattern here. We should try to improve this. I think _instantiate_feeds should be called just once with n_trials rather than being overwritten by another call. It's surprising.

@cjayb hope this does not cause too much of a rebase issue for you?

@cjayb hope this does not cause too much of a rebase issue for you?

don't worry about it, I'm still knee-deep in figuring out parconnect. I'll rebase when the time comes, still working on and need to simplify #221 (once I get it working, that is)

ntolley · 2020-12-22T15:16:08Z

examples/plot_simulate_gamma.py

 freqs = np.arange(20., 100., 1.)
-n_cycles = freqs / 4.
+n_cycles = freqs / 8.


Is this to keep a consistent time resolution between morlet and multitaper? I'm struggling to remember exact impact for the two methods.

no, not really. MNE complains about wavelet cycles being too long if n_cycles = freqs / 4. Not sure why it doesn't do the same with multitaper. Anyhow I think HNN-GUI uses n_cycles = freqs / 7 but MNE doesn't allow that, so I picked the closest possible :)

As for the difference between the two, this page gives a nice summary. Essentially, morlet is like doing multitaper with a single Gaussian taper

Not sure why it doesn't do the same with multitaper.

maybe an opportunity to make a small PR to MNE if you are willing to dig into it? :)

jasmainak · 2020-12-23T22:39:07Z

@ntolley incorporated the raster plot in this example now. Take a look at the example in the CircleCI artifact tab once it is built.

It's trying to reproduce this: https://jonescompneurolab.github.io/hnn-tutorials/gamma/gamma.

The merge button is yours when you're happy with the changes/plots.

jasmainak · 2020-12-23T23:35:37Z

new plots: https://755-168215891-gh.circle-artifacts.com/0/html/auto_examples/plot_simulate_gamma.html#sphx-glr-auto-examples-plot-simulate-gamma-py

jasmainak · 2021-01-03T04:27:45Z

ping ping, any blockers with the reviews here?

cjayb

The new plots are beautiful! Few minor comments, otherwise do merge asap. I'll have to rebase #221 anyway, let's get this in there before.

cjayb · 2021-01-03T14:10:20Z

examples/plot_simulate_gamma.py

+###############################################################################
+# The network requires some time to reach steady state. Hence, we omit the
+# first 50 ms in our time-frequency analysis.
+


Suggested change

trial_index = 0 # pick first trial

cjayb · 2021-01-03T14:11:45Z

examples/plot_simulate_gamma.py

+mask = dpls[0].times > tstart
+times = dpls[0].times[mask]
+data = dpls[0].data['agg'][mask]


Suggested change

mask = dpls[0].times > tstart

times = dpls[0].times[mask]

data = dpls[0].data['agg'][mask]

include_samples = dpls[trial_index].times > tstart

...

Just to be explicit about what you're doing. Users may not feel comfortable with 'magic' constants like 0 before they get the hang of the objects.

okay with adding trial_index. I actually prefer mask as opposed to include_samples though. It's easier to read the code because I know that it's a masked array as opposed to being indices or slices.

cjayb · 2021-01-03T14:16:30Z

examples/plot_simulate_gamma.py

+# Notice that the Layer 5 pyramidal neurons are now firing nearly
+# synchronously. They in turn synchronously activate the inhibitory basket
+# neurons, which then inhibit the pyramidal neurons for ~20 ms, when the
+# tonic drive outweighs the inhibition and the pyramidal neurons firing again
+# creating a ~50 Hz PING rhythm. This type of synchronous rhythm is sometimes
+# referred to as “strong” PING.


Suggested change

# Notice that the Layer 5 pyramidal neurons are now firing nearly

# synchronously. They in turn synchronously activate the inhibitory basket

# neurons, which then inhibit the pyramidal neurons for ~20 ms, when the

# tonic drive outweighs the inhibition and the pyramidal neurons firing again

# creating a ~50 Hz PING rhythm. This type of synchronous rhythm is sometimes

# referred to as “strong” PING.

...

# neurons, which then inhibit the pyramidal neurons for ~20 ms. Once the tonic

# drive again outweighs the inhibition, the pyramidal neurons fire again,

# creating a ~50 Hz PING rhythm. This type of synchronous rhythm is sometimes

# referred to as “strong” PING.

Nitpick

jasmainak · 2021-01-04T06:35:47Z

Gonna commit a small crime and self-merge to keep things moving with #221

Thanks for the reviews @ntolley and @cjayb

jasmainak · 2021-01-04T06:36:29Z

if you disagree with something, feel free to follow up with PRs :)

jasmainak requested a review from rythorpe December 20, 2020 05:43

jasmainak commented Dec 20, 2020

View reviewed changes

jasmainak changed the title ~~DOC: use morlet instead of multitaper~~ [MRG] DOC: use morlet instead of multitaper Dec 20, 2020

jasmainak commented Dec 20, 2020

View reviewed changes

jasmainak force-pushed the gamma_tfr branch 2 times, most recently from 2500a63 to e34d37b Compare December 20, 2020 16:07

ntolley reviewed Dec 22, 2020

View reviewed changes

jasmainak added 4 commits December 23, 2020 17:05

DOC: use morlet instead of multitaper

e885478

DOC: add tonic input to the example

8c9d65d

TST to check that tonic inputs are not overwritten by dipole.py call

15b2976

DOC: add raster plot

e49b95b

jasmainak force-pushed the gamma_tfr branch from e34d37b to e49b95b Compare December 23, 2020 22:35

jasmainak mentioned this pull request Dec 28, 2020

[MRG] Refactor API for creating external network drives #221

Merged

9 tasks

cjayb approved these changes Jan 3, 2021

View reviewed changes

ENH: address comments by chris

602a157

jasmainak merged commit 93c79d9 into jonescompneurolab:master Jan 4, 2021

jasmainak deleted the gamma_tfr branch January 4, 2021 06:35

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[MRG] DOC: use morlet instead of multitaper #226

[MRG] DOC: use morlet instead of multitaper #226

jasmainak commented Dec 20, 2020 •

edited

jasmainak Dec 20, 2020

ntolley Dec 22, 2020

jasmainak Dec 22, 2020

jasmainak Dec 20, 2020

codecov-io commented Dec 20, 2020 •

edited

jasmainak Dec 20, 2020

jasmainak Dec 20, 2020

cjayb Dec 20, 2020

ntolley Dec 22, 2020

jasmainak Dec 22, 2020

jasmainak Dec 22, 2020

jasmainak commented Dec 23, 2020

jasmainak commented Dec 23, 2020

jasmainak commented Jan 3, 2021

cjayb left a comment

cjayb Jan 3, 2021

cjayb Jan 3, 2021

cjayb Jan 3, 2021

jasmainak Jan 4, 2021

cjayb Jan 3, 2021

jasmainak commented Jan 4, 2021

jasmainak commented Jan 4, 2021

[MRG] DOC: use morlet instead of multitaper #226

[MRG] DOC: use morlet instead of multitaper #226

Conversation

jasmainak commented Dec 20, 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

codecov-io commented Dec 20, 2020 • edited

Codecov Report

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 Dec 23, 2020

jasmainak commented Dec 23, 2020

jasmainak commented Jan 3, 2021

cjayb left a 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 Jan 4, 2021

jasmainak commented Jan 4, 2021

jasmainak commented Dec 20, 2020 •

edited

codecov-io commented Dec 20, 2020 •

edited