search

j-faria / kima

Exoplanet detection in RVs with DNest4 and GPs
http://www.kima.science
MIT License
15 stars 10 forks source link

Aliases #49

Open astudillo-defru opened 6 years ago

astudillo-defru commented 6 years ago

Hi, kima favors a 2 planets model but the posterior distribution of the orbital periods shows that one is the 1-day alias of the other. How can I prevent this? Thank you!

eggplantbren commented 6 years ago

Do you know whether the ambiguity is genuine? If it is, there's nothing you can do (but it should favour a one planet model, sometimes with one period and sometimes with the other period). If it is really just an alias (and this is clear given the data), then this shouldn't happen and I can't figure out why it is happening without more information.

astudillo-defru commented 6 years ago

Thanks for your reply. I am (analytically) sure it is the genuine 1 day alias. In addition, the periodogram of RV residues when subtracting one of the periodicity do not show power excess at the alias periodicity. The posterior distribution of the orbital periods from kima shows both (alias) periodicities at the same time. What additional information do you need?

j-faria commented 6 years ago

I'm also surprised the two planet model is favoured because that means solutions with keplerians at the two periods together fit the data better. In cases with an alias I would expect two nearly equal peaks in the period posterior but the one planet model being preferred.

There is no 3rd signal in the data?

As a 'solution' (a workaround, really) try restricting the prior to periods above 1 day.

eggplantbren commented 6 years ago

In addition, the periodogram of RV residues when subtracting one of the periodicity do not show power excess at the alias periodicity.

In this situation, kima should not be inferring both frequencies. Is it possible to provide the data?

astudillo-defru commented 6 years ago

Thank you for the support, but sorry, for the moment I can not provide the data

eggplantbren commented 6 years ago

Can you make a fake dataset that causes the same problem?

On Wed, 15 Aug 2018, 2:21 AM astudillo-defru, notifications@github.com wrote:

Thank you for the support, but sorry, for the moment I can not provide the data

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/j-faria/kima/issues/49#issuecomment-412888974, or mute the thread https://github.com/notifications/unsubscribe-auth/ABgVOmbeCM0OdOo-awk_WboJbgIVa_lGks5uQtzhgaJpZM4VyiLK .