How to obtain passband luminosity ratio？

[alan0986]

Hello,
I need to obtain the luminosity ratio of the two components when I run mcmc. However, in the 'component-coupled' mode, the secondary component luminosity is not visible. I try to get the luminosity ratio with the following custom constraints, and transfer the luminosity uncertainties of the primary component to the luminosity ratio after the mcmc runs.

lumidata = b.compute_pblums()
pblumratio_def = phoebe.parameters.FloatParameter(qualifier='pblumratio',
default_unit=u.dimensionless_unscaled,
value=1,
description='Passband Luminosity ratio', overwrite_all=True)
pblumratio_param, created = b.get_or_create('pblumratio',pblumratio_def, context='dataset', dataset='lcT')
lsh = pblumratio_param
rsh = np.float64(lumidata['pblum@secondary@lcT'])/b.get_parameter(qualifier='pblum', component='primary', dataset='lcT')
b.add_constraint(lsh,rsh)

However, when I run
b.uncertainties_from_distribution_collection(solution='emcee_sol', parameters=['pblumratio'], tex=True)

NotImplementedError: cannot convert object with physical_ type=power/radiant flux to solar units

Or is there any other way to solve this problem?
Thanks

[kecnry]

I think your units error is because you are casting the pblum@secondary to a float, which then results in setting the expression with non-unitless units. In theory you would divide two parameters, but as you've mentioned, the secondary luminosity isn't a parameter when the mode is set to component-coupled. Even if this didn't raise a units error, this would be using a fixed value for the secondary luminosity while computing the luminosity ratio - is that what you are intending to do?

Since you're using component-coupled, both luminosities will likely be changing, and so you would need to call b.compute_pblums for each value of the primary's luminosity. Unfortunately, PHOEBE doesn't yet support a nice way of doing this, but depending on your use case, you might find the discussions in #571 (essentially the same use-case but just to access the secondary luminosity instead of the ratio) or possibly #546 and #443 useful. I'll also just link to this discussion topic for possibly adding native support within phoebe if you have any thoughts you want to add there.

[alan0986]

Ok，OK, I can get the distribution of 'pblum@primary stars' . There may be obvious mistakes in my code. I hope you can give me some suggestions. Thanks.

def compute_pblum_secondary(b, pblum_primary):
    b.set_value('pblum@primary@lcT@lc@dataset', value=pblum_primary)
    # b.set_value('teff@secondary@star@component', value=teff_secondary)  
    return b.compute_pblums()['pblum@secondary@lcT@lc@dataset']

lp1=b.get_parameter('pblum@primary@lcT@lc@dataset',  ).get_distribution('emcee_sol')
lp1
#lp2=b.get_parameter('teff@secondary@star@component', ).get_distribution('emcee_sol')
lp3 = distl.function(compute_pblum_secondary, args=(b, lp1), vectorized=False)

[kecnry]

I will need to try to reproduce this or write up a full working example, but that will take a little time. If you manage to find the solution in the meantime, please let us know. Just to clarify - what version of PHOEBE are you using?

[alan0986]

Ok, I am using PHOEBE 2.4. Thanks again for your help.

[kecnry]

The issue was that the function does not accept taking the bundle itself as an argument, so you just need to remove that - sorry for the confusion (I just wrote the original pseudocode from memory). I now created a full example for propagating distributions through luminosity coupling that includes this syntax fix and hopefully explains in more detail. You'll still need to modify this logic slightly if you want the ratio instead of the secondary luminosity (in addition to the specific distributions that you have sampled from MCMC). Let me know if you have any suggestions on how to improve the example or can't figure out how to modify it for your situation.

[alan0986]

Thank you very much for your patience and help.