[ENH] Neaspec multichannel reader #739
Conversation
|
Thanks!
This is very nice, thank you. From the FFT side, this PR seems good. Perhaps @markotoplak can put a second set of eyes on the reader side. |
markotoplak
changed the title
There was a problem hiding this comment.
I only have complaints about the test file size and this is the only thing I will enforce.
I do not deal with such files ofter therefore I am not the one to judge this from usability point of view. Also, someone else has to decide what to do with many Nea readers we know have - are there any to remove or merge together?
...ontrib/spectroscopy/datasets/NeaReaderMultichannel_test/Test_Au_Fourier_Scan_Synchrotron.txt
Outdated
Show resolved
Hide resolved
|
@borondics, @stuart-cls, @brnovasco, please tell me when this is ready - I am not a good judge of file formats. If I understood the above comments, row and column should be changed to map_x and map_y, and then this is it? Please do not merge yourselves because the test file should be made smaller and the big one rebased out. |
|
My only wish here is to have "map_x" and "map_y" as meta names. Maybe we could make a definition file that all orange-spectroscopy can import and define these there? In some other readers the final step is a helper function that constructs the data table from the different parts and that has "map_x" and "map_y". So, we could unify things that way too. Also, in this particular case it should be row -> map_x and column -> map_y because the measurement method is row-by-row, which introduces special need for the data analysis. As for test files, it should be really easy to just delete all the data except for a few lines, then it is a really dramatic space saving. |
|
Here's my understanding:
|
…le channels Reader needs to resample the data to the same domain to compensate M position deviations and comply with the orange-spectroscopy data model.
…he polar form (separated amplitude and phase channels in alternating rows).
….gsf format Do so by making the reader calculate its datapoint spacing and save it as metadata that can be consumed in the fft widget. This allow the owfft widget to treat the data from the NeaReaderGSF and the NeaReaderMultiChannel the same way, as it should since they both use the same complex fft and provide data in similar form.
… scaling the domain to [µm] and renaming column and row to standard map_x and map_y vars.
brnovasco
force-pushed
the
neaspec-multichannel-reader
branch
from
cb6b3a2 to
781b1d2
Compare
|
Hi, @markotoplak and @stuart-cls. I think its ready with the requested changes. Can you check? |
This PR is related to the discussion in #717.
Following up on the development of a reader for the Neaspec data containing multiple channels with separated data for phase and amplitude in multiple runs, rows and columns.
We are creating a reader that performs a interpolation of the data to be able to have all interferograms in the same domain related to the raw position sensor reading instead of its array index.
This imply changes in the OWFFT widget so we avoid duplicate code for the complex fft calculation. We then standardize the way the widget performs the fft for the .gsf and .txt data from Neaspec simplifying the logic on the FFT widget and improving the NeaReaderGSF class.