Add a "filter" column

[haifagmati]

After several tests for the HG model, I suggest defining a unique structure for all filters and adding a 'filter' column instead of modifying the designations for H, G, Err_H and Err_G.

[JulienPeloton]

Hello @haifagmati -- thanks for the issue! Could you be more explicit on the foreseen changes? Are you referring as to the input or output of the parameter estimation? And how the modification would look like ?

[haifagmati]

For the HG model, I need to retrieve the results, for different objects and different filters, without doing tests for each returned designation and having a unique structure.

[JulienPeloton]

I'm not sure I get what you mean by "unique structure". If I take an object and estimate parameters, I obtain for example:

{'chi2red': 3.6057319598180935,
 'status': 2,
 'fit': 0,
 'rms': 0.09166807890172138,
 'rms_1': 0.09296629310298726,
 'rms_2': 0.09049974636640824,
 'n_obs': 219,
 'n_obs_1': 103,
 'n_obs_2': 116,
 'min_phase': 19.8815493924,
 'min_phase_1': 19.8815493924,
 'min_phase_2': 19.8822928532,
 'max_phase': 38.5103181958,
 'max_phase_1': 38.5100206594,
 'max_phase_2': 38.5103181958,
 'H_1': 15.399593468419473,
 'err_H_1': 0.08203959335281041,
 'G_1': 0.1187684948217654,
 'err_G_1': 0.046319412723385134,
 'H_2': 15.063650641519713,
 'err_H_2': 0.08229461202334086,
 'G_2': 0.16174326385412047,
 'err_G_2': 0.04948557444073601}

In this case, there are 2 filters available (1 & 2), but there could be one only. What other structure would you propose?

[haifagmati]

My proposition is, for a single filter with the 'HG" model, I need to obtain the following structure :

['SSO', 'chi2red', 'status', 'fit',‘filter’, 'rms', 'n_obs', 'min_phase', 'max_phase', 'H', 'err_H’, 'G', 'err_G']

for example, for filter 1 :

{'SSO':X
'chi2red': 3.6057319598180935,
'status': 2,
'fit': 0,
‘filter’:1
'rms': 0.09296629310298726,
'n_obs': 103,
'min_phase': 19.8815493924,
'max_phase': 38.5100206594,
'H': 15.399593468419473,
'err_H’: 0.08203959335281041,
'G': 0.1187684948217654,
'err_G': 0.046319412723385134
}

I need this structure to run the query for different filters in parallel and reconstitute the results into a single object without testing the returned designations

[haifagmati]

this is an example of a test that, I am currently doing after each execution of a single filter:

                    if key.startswith('H_'):
                        data_element['H'] = value
                    elif key.startswith('err_H_'):
                        data_element['err_H'] = value
                    elif key.startswith('G_'):
                        data_element['G'] = value
                    elif key.startswith('err_G_'):
                        data_element['err_G'] = value

[JulienPeloton]

Hi @haifagmati -- Unfortunately I want to keep the service as generic as possible. I do not know a priori how many filters will be given by the users, hence I need to a way to tag without ambiguity the output, and let the users extracting relevant information.

Though, if you know you are always using one filter at a time, you can use this:

# get SSO data with one filter
pdf = ....

# extract the filter name
filt = np.unique(pdf['i:fid'].values)[0]

# estimate parameters
hg = estimate_sso_params(
  pdf['i:magpsf_red'].values,
  pdf['i:sigmapsf'].values,
  np.deg2rad(pdf['Phase'].values),
  pdf['i:fid'].values,
  p0=[15.0, 0.15],
  bounds=([0, 0], [30, 1]),
  model='HG',
  normalise_to_V=False
)

# At this point, the structure contains `_{}` names 
# so just overwrite them
hg_ = {k.split('_{}'.format(filt))[0]: v for k, v in hg.items()}
hg_['filter'] = filt

print(hg_)
{'chi2red': 2.988535572880441,
 'status': 2,
 'fit': 0,
 'rms': 0.09857987298612855,
 'n_obs': 103,
 'min_phase': 19.8815493924,
 'max_phase': 38.5100206594,
 'H': 15.39959347298408,
 'err_H': 0.08827723185449289,
 'G': 0.11876849739578137,
 'err_G': 0.04984117236905842,
 'filter': 1}

Would that work for you?