Last interaction info missing

[pfreddy]

Hi,
a new error appeared with the latest development version: the GUI starts well, but the widget with the last interaction information doesn't open. Error message is attached below.
There was a similar bug earlier, which is connected to the change from model refrence from Simulation model reference (Issue 690 & 691 & 720), but this error message seems somehow different.

...
File "/home/barna/anaconda3/envs/tardis2/lib/python2.7/site-packages/tardis/analysis.py", line 104, in update_last_interaction_filter
'level_number_upper']].ix[last_line_in_count.index]
...
KeyError: "['atomic_number' 'ion_number' 'level_number_lower' 'level_number_upper'] not in index"

[yeganer]

@pfreddy I'll have a look next week.

[unoebauer]

Hi @pfreddy. I am currently looking at this problem since also @ssim noticed that the gui is again broken.

The reason seems to be a change in the data frame structure of the line data. While in the old days, the line list had one index, the line_id it now has a multi index, consisting of atomic_number, ion_number, level_number_lower, level_number_upper. This also means, that these values do not appear as data columns in the data frame. However, the gui script expects this.

Note below the change is structure:

Old version, only one index

In [5]: mdl.plasma.atomic_data.lines.iloc[0]
Out[5]: 
wavelength            6.677200e+01
atomic_number         1.400000e+01
ion_number            5.000000e+00
f_ul                  2.703017e-02
f_lu                  4.054525e-02
level_number_lower    0.000000e+00
level_number_upper    3.600000e+01
nu                    4.489793e+16
B_lu                  4.545377e+07
B_ul                  3.030251e+07
A_ul                  4.043917e+10
wavelength_cm         6.677200e-07
Name: 8, dtype: float64

Current version, multi-index

In [53]: mdl.plasma.atomic_data.lines.iloc[0]
Out[53]: 
line_id          6.781300e+04
wavelength       6.677200e+01
f_ul             2.703017e-02
f_lu             4.054525e-02
nu               4.489793e+16
B_lu             4.545377e+07
B_ul             3.030251e+07
A_ul             4.043917e+10
wavelength_cm    6.677200e-07
Name: (14, 5, 0, 36), dtype: float64

[unoebauer]

Update: @wkerzendorf and I will try to fix this tomorrow during our Tardis sprint.

[yeganer]

@wkerzendorf @unoebauer Sorry for the delay. You are right, the problem arises because of the change in the indexing of atom_data.lines which I did as part of merging the carsus atomic database. The goal was to remove constant re-indexing and to provide a 'standard' format for indexes within tardis.

As I am the one who broke it I'm happy to help you fix it or fix it myself. I think I have most issues fixed in a local branch but I need haven't done any tests because I have to figure out how to start the GUI first :)