jmackereth / asteroestimate

python-based detection probabilities and scaling relations for asteroseismology (Work In Progress!)

Hi there 👋

asteroestimate's Issues

Include scaling relations as a separate entity

It would be neat to have an in-built set of scaling relations for quickly propagating uncertainties/MCMC sampling etc.

A quick test has been done with Yu's Kepler sample

Tanda: a quick test with Yu's Kepler sample is done. As shown in the plot below, there is a rough correlation between the luminosity and nu_max, however, spreads are obvious. Seismic luminosities are calculated with observed Teff and seismic Radius in Yu2016; The Gaia Luminosities are from Berger2018, who did some corrections on the Gaia luminosities; the nu_max are from Yu2016, derived with the Sydney pipeline.

A few thoughts:
Using this as a test for Ted's method might be useful;
It make me a bit concerned that the mass distribution is not consistent as luminosity increases;
If we trust the relations here and assume it works for TESS targets, we may use this observed relation as prior instead of the scaling relation. (It can be fitted with a linear function with spreads.)
Any thoughts or comments, just add here.

Implement the scatter in the intrinsic mapping function as part of the Monte Carlo process over the input values.

The function that maps the observables to Numax is uncertain. We don't know how much by. There is an issue to decide what the magnitude of this uncertainty is. Here we need to decide how to include the uncertainty into the Monte Carlo process.

Write a wrapper function to Monte Carlo over input uncertainties.

So this will be the main way that people can interact with the software. It will be a function (inside a class?) with inputs and out the prior as the summary statistics of a Gaussian distribution and the samples from the Monte Carlo.

https://www.dropbox.com/sh/g9duucyyxgy9e3u/AABdwdX0YsSzrcy_lvZu-M25a?dl=0

Get the catalogue and make some sensible cuts. This will be our testing data.

query by Gaia source ID

would be really handy to be able to supply Gaia source_id only and return the detection probability

this could use astroquery or similar to get the necessary info. May also need to use Tess-point to get the number of sectors/CCD positioning?

Build in priors on unknown parameters

need to implement some way of including 'ignorance' on some inputs... The best example of this is the mass prior which can be passed to probability.from_phot. However, a general use stellar population prior might be handy!

Use astropy units

Would be great to implement astropy unit use as an option everywhere. This might make things more compatible with other things!

Test the scalling relation function with Yu's Kepler sample, which has mass, radius, nu_mux, delta_nu, Teff, etc. The aim is to find a better mass assumption.

The Yu sample is an excellent test sample and we don't anticipate any (major/significant) differences in numax from one mission to the other. We could do plenty of tests using the Yu sample.

Note that there are differences Kepler and TESS, including: The giants will be further away in Kepler; The patch of sky is different; Metallicity and mass distributions may be different; ...

Decide on the sensible value for the intrinsic spread accounting for the uncertainty in the relation that maps from Gaia/2MASS stuff to Numax.

Here we need a plan and some effort to estimate the spread in the prior that comes from our uncertainty in the map from observables to numax. We'll need to test the relationship against values we have a degree of trust in. We'll need to account for the uncertainty in the observable values when we calibrate this too. There are probably endless ways and endless fancy methods of doing this. How can we do this in the simplest possible but still acceptable way?