search

mrtommyb / GP_model_Kepler_data

modeling a Kepler light curve using a Gaussian Process
MIT License
0 stars 0 forks source link

What happens if we leave out the RV data? #3

Open dfm opened 10 years ago

dfm commented 10 years ago

If it's not too hard, it would be great to do an MCMC run that didn't include the RV data. We claim (in the abstract) that one possible reason for the discrepancies is the different noise model and it would be really great to demonstrate that.

mrtommyb commented 10 years ago

Good idea. I'll set this up and run.

mrtommyb commented 10 years ago

I still haven't run this. I'll try to get it done before I submit the paper but I'm not going to consider it a blocker.

mrtommyb commented 10 years ago

I've run this simulation, we get worse constraints on things like eccentricity but the results are totally consistent. So, I think we can conclude that you get the wrong answer by not modeling the correlated noise. This is a big thumbs up for @dfm . Great stuff.

dfm commented 10 years ago

Awesome! We probably don't need to include any figures in the paper but a few words would be good. I'm really stoked (and officially not surprised :wink:) that it worked out that way!

mrtommyb commented 10 years ago

I'm going to have a go at rewriting the discussion section to make it clear that we think the other groups came to the wrong conclusion because of the correlated noise in the data. This allows us to make a bigger deal of the GP model and how awesome it is.