Fix events sorting bug and clarify some documentation #56
Conversation
| @@ -43,15 +43,14 @@ | |||
| src_outputs = hcp.anatomy.compute_forward_stack( | |||
| subject=subject, subjects_dir=subjects_dir, | |||
| hcp_path=hcp_path, recordings_path=recordings_path, | |||
| # speed up computations here. Setting `add_dist` to True may improve the | |||
| # accuracy. | |||
| # Speed up computations here by setting `add_dist` to False. | |||
There was a problem hiding this comment.
@Eric89GXL, IIRC, you added distances for some specialized src spaces analysis, right? It doesn't actually improve accuracy, correct?
There was a problem hiding this comment.
The add_dists also adds patch info which is used in forward computation to make things slightly better
There was a problem hiding this comment.
Okay, good to know. I'll add that comment back in
|
@dengemann, do I need to rerun all the example/tutorial analyses on my end? |
|
No. I need to run it :) we should actually find a way to use circle ci. Btw
can you run the reproduce erf example and tell me if outputs improved?
…On Fri, 21 Apr 2017 at 15:21, Mark Wronkiewicz ***@***.***> wrote:
@dengemann <https://github.com/dengemann>, do I need to rerun all the
example/tutorial analyses on my end?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#56 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AB0fio0DXMxh1x84oWHQgl5UHHqkCLbvks5ryQHJgaJpZM4NEr1B>
.
|
|
I'm hadn't been analyzing the WM data, so I'll download it and get back to you |
|
Hmm, sorting events when processing the data from scratch made the correlation with the fully-processed HCP data go down (~0.7 to ~0.4). This is definitely not what I was expecting. Ideas?
|
|
Hi @dengemann :) |
|
Bonjour @dengemann :) |
|
Yes, good timing. We need to look at this. I'm receptive these days,
…On Sun, Oct 29, 2017 at 6:11 PM Mark Wronkiewicz ***@***.***> wrote:
Bonjour @dengemann <https://github.com/dengemann> :)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#56 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AB0fiqp3jzaTLkDEmqwVyDyR-GWdPYG1ks5sxLHRgaJpZM4NEr1B>
.
|
|
I don't think I'll have much time to work in this (starting a new job :D), but I would suggest merging this asap. I don't know why the examples get slightly worse, but the bugfix for |
|
I can find some time in the following days.
…On Sun, Oct 29, 2017 at 6:28 PM Mark Wronkiewicz ***@***.***> wrote:
I don't think I'll have much time to work in this (starting a new job :D),
but I would suggest merging this asap. I don't know why the examples get
slightly worse, but the bugfix for plot_reproduce_erf.py is important.
Anyone who uses that code will have their analysis pipeline destroyed
because the columns in the events array are currently all being sorted
independently. End result is that event timings will no longer line up with
the time point of the raw file.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#56 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AB0fisOGiHeqLjTC1srtPIbVFUO4nJC3ks5sxLWzgaJpZM4NEr1B>
.
|
No description provided.