search

California-Planet-Search / rvsearch

RV Planet Search Pipeline Based on RadVel
http://california-planet-search.github.io/rvsearch/
MIT License
10 stars 5 forks source link

rvsearch not picking up significant signals #185

Closed zhexingli closed 3 years ago

zhexingli commented 3 years ago

Hello,

I'm not quite sure if this is an issue. But feel free to share your thoughts on this. So I created a synthetic rv dataset with three planets: Planet b: P = 254 days, Planet c: P = 4862 days, Planet d: P = 1810 days, with the first two planets have relatively large K and the third planet with extremely small K (below 0.3 m/s). See the rv below: rv

I was expecting the first two planets to be picked up by rvsearch. The first planet did, but the second didn't, see the results plot: PeriodSearch

Instead of getting the 4862 day orbital period, it picked up a random 33-day signal (possibly alias), which really bummed me out coz I didn't add anything close to 33 days into the data. The dataset was created with almost even sampling with some rms scatter to the rv.

Any idea what's causing this and how I could avoid this in the future when doing blind search on other targets? Thanks.

awhoward commented 3 years ago

It looks like there's a very regular set of peaks in the periodograms stemming from regular sampling and aliases. Try choosing random times of observation.

zhexingli commented 3 years ago

It looks like there's a very regular set of peaks in the periodograms stemming from regular sampling and aliases. Try choosing random times of observation.

Thanks Andrew. Yeah those forest of peaks are all within 100-day period and I agree must be coming from near regular sampling and aliases. But does that mean the most ideal regular rv observations might not be as good as random irregular observations?