update default dipole data #344
Conversation
|
Is this the right idea @jasmainak? |
|
yeah, this looks fine. Do the tests on the other branch pass with this? |
| @@ -23,7 +23,7 @@ | |||
| 0.550 -0.6139 -0.0003 -0.6136 | |||
| 0.575 -0.6146 -0.0003 -0.6143 | |||
| 0.600 -0.6153 -0.0003 -0.6150 | |||
| 0.625 -0.6159 -0.0003 -0.6156 | |||
| 0.625 -0.6160 -0.0003 -0.6156 | |||
There was a problem hiding this comment.
why are these numbers only up to 4th decimal place? Isn't Python numerical accuracy greater?
There was a problem hiding this comment.
I think this is just the rounding convention we've always used when writing to dipole .txt files.
There was a problem hiding this comment.
right, but our precision for comparison is apparently higher than this. Something to think about in the future. Also, yet another reason to move away from text files. The more precision you add, the more bulky it becomes. Because each character is stored as an ASCII rather than the whole number as a floating point.
There was a problem hiding this comment.
True. What would be a better format?
There was a problem hiding this comment.
npy / hdf5 / mat ...
see this for example: bids-standard/bep021#1 (comment)
Yep! |
|
Thanks @rythorpe ! |
Updates our ground-truth dipole data generated by
default.jsonto address removal of rounding error in cell positioning in #314.