true luminosity

[Guoyani519]

Hi, Is the luminosity value obtained from this code the true luminosity value?

[kecnry]

compute_pblums exposes how phoebe will interpret the input parameters related to passband-luminosity to pass to the backend at time t0@system, ultimately scaling the light curve fluxes. Since you're using pblum_mode == 'absolute', there won't be any scaling of the fluxes, and so this is effectively the intrinsic (before any spots, irradiation, etc) luminosity, in the light curve's passband, at time t0. If you want the observed luminosity after accounting for spots, irradiation, etc, or at a specific time/phase then you can compute a mesh dataset and calculate the luminosity from the per-element values. See https://phoebe-project.org/docs/latest/tutorials/pblum and https://phoebe-project.org/docs/latest/tutorials/MESH for details.

[Sam4892]

Hi Kyle,
Can I ask a REALLY stupid question here... I'm an MSc student who is very poor at coding, for reference!
I have loaded my observations in as relative fluxes to a comparison star, so the fluxes show at values around 0.08. When I build a model like the tutorials, the fluxes are between 1-2 W m^2 (of course) - so if I plot both together, my data just looks like a flat line at the bottom.
If I do some of the steps reducing the pblum as in the tutorial you link here, can I bring down the model so I can see both light curves together, or is that going to make the model unphysical? I was debating changing all my observational fluxes to magnitude, but I'm not sure how I would run that in PHOEBE.
I know this is almost certainly basic stuff, but my tutor can't help me further.
PHOEBE is a bit too clever for me, I think!
Thanks, Sam

[kecnry]

Hi Sam

The model (created with run_compute is phoebe synthesizing what observations would look like given all the input parameters as they are). You can then either manually change the input parameters to see how they affect the output forward model (perhaps relative to your data), or you can use any of the solvers implemented in phoebe which each take different approaches to proposing changes to your input parameters so that the model will be in better agreement with your observational data. If this is what you want to do, I'd suggest reading the phoebe paper on the inverse problem and the relevant tutorials (see section 8 on the inverse problem and all the advanced tutorials nested under that). There are also some recordings and additional notebooks from part 2 of our 2022 workshop that might be helpful (although note that these might be slightly out of date in a few cases).

If you only care about the vertical scaling, you can tell phoebe to handle that automatically when creating a forward model by setting pblum_mode to dataset-scaled, but please note that you will want a more robust handling of passband-luminosity for any finalized model - this is just convenient sometimes to get started. See more details in the pblum tutorial.

[Sam4892]

Thank you Kyle (working on a Sunday?!)

Appreciate you writing that for a novice (it took me 3 weeks to even be able to load a dataset!) Sadly I can't use the inverse solvers as the system is a contact binary. I have used the pblum_mode method you suggested which has meant I can fit the model curve to the observations with the info I had managed to calculate from B-V index etc. Because it's all now showing fluxes of 0.08 W m^2 it does have the hilarious situation of component stars at 5460 and 5000 K but only 0.08 solar masses, but at least I can try and model the impact a spot would have!

If I can ask one more idiot question... from looking at my light curves intially, I think I have a ratio of the radii of around 0.8. As requivratio@binary is contrained by requiv of the two components, and these are themselves constrained by semi-major axis, q and potentials, how do I need to flip these to either put in requivratio of 0.8, or set values eg requiv@primary = 1.25, requiv@secondary = 1?

Really appreciate your help here, thank you!
Sam

[kecnry]

Sadly I can't use the inverse solvers as the system is a contact binary.

There aren't any estimators currently written specifically for contact binaries, but optimizers and samplers will work!

it's all now showing fluxes of 0.08 W m^2 it does have the hilarious situation of component stars at 5460 and 5000 K but only 0.08 solar masses, but at least I can try and model the impact a spot would have!

This might be ok to play with spots, but is an unphysical system (as you noticed). Something in Kepler's third law is likely incorrect (I'm guessing the semi-major axis of the orbit or potentially the period if you're working in phase-space) that is driving the masses so low.

As requivratio@binary is contrained by requiv of the two components, and these are themselves constrained by semi-major axis, q and potentials, how do I need to flip these to either put in requivratio of 0.8, or set values eg requiv@primary = 1.25, requiv@secondary = 1?

By default requiv@primary and q are the free parameters that you can set. Unfortunately PHOEBE doesn't support providing both requivs and having q/potential determined (not all solutions will exist) for contact systems. You can choose whether you want to provide requiv@primary, pot@contact_envelope, or fillout_factor@contact_envelope to define the "size" of the envelope alongside q and sma.

[Sam4892]

Thanks Kyle! I will have another look at the optimizers, see if they can find something now I've got a little further with it.

I'm pretty confident in the period, the source was part of SuperWASP originally so there is a long baseline of observations. I only have photometric data now, no spectroscopy, so I have little info on the masses or the semi-major axis. I changed the sma a few times and the masses also changed in the model without altering the curves at all, so I guess this might be things I may never find out.

My system originally exhibited the O'Connell Effect in SuperWASP, and appears to do so still, so modelling for spots / hot spots is what I really need to do, and I seem to be able to do that regardless of the parameters mentioned, which is good - I'd just like to get those too :)

It was actually using mino-xx0207's flipped parameters from thread #1016 that finally let me get the model to work - unfortunately, this means I now can't change requiv@primary or q as they are constrained again, and any other flips I've tried haven't worked, so I've... run away!!!

Really appreciate your time - I hope I can get some good arguments about spots as possible causes of the O'Connell Effect from this!
Sam

[Guoyani519]

Hello, following your suggestion, I calculated the luminosity for different phases, but why are the luminosity values the same for different phases?

[Raindogjones]

Presumably because your system is circular, so neither star changes during the orbit. If you make the system eccentric, such that the tidal distortion of one or both stars varies appreciably, then these values won't be constant.