Closed hcferguson closed 3 years ago
@hcferguson It might be good to try it out in a standalone specviz, then if it is a specviz issue open a ticket there and link back to here. I suspect it is a data / specviz issue not specifically cubeviz.
Actually more likely the problem is in specutils (the fitting itself). That should be easier to try standalone anyway.
This dataset seems to be broken. It's full of nans and is effecting other things in specutils/specviz like the statistics calculations. I'm not surprised that the model fitting fails (although, I believe I've fixed the issue of it crashing).
That said, the statistics and model fitting work when selecting regions that have actual data. I'm going to assume this is sufficient?
@hcferguson please try with the current master in my fork (pip install git+https://github.com/nmearl/specviz --upgrade
) and let me know if you experience the same issue.
Mixed success.
I've tried it about a dozen times. It mostly worked. but it crashed twice with a segfault and a complaint abut QTextCursor. That was after a fair amount of adjustment of the plot scaling. I think the ROI did not include the NAN's (the stats looked normal).
Stack trace:
(Make sure 'QTextCursor' is registered using qRegisterMetaType().)
WARNING: Model is linear in parameters; consider using linear fitting methods. [astropy.modeling.fitting]
WARNING:astropy:Model is linear in parameters; consider using linear fitting methods.
[1] 79475 segmentation fault cubeviz
@hcferguson Can you update with the latest changes in my fork and give it another shot?
@nmearl Tried it about a dozen times with no crash.
Ran into a separate (new) issue when I was a bold and tried fit spaxels
and then aborted. I'll file that separately.
Tried to fit bjames SINFONI data with a linear model (no ROI)
Stack trace