Planet Hunters Talk

Too Many Simulations

  • dantoto by dantoto

    Can you please not have so many simulated transiting stars. It is a bit of a waste of time when there are more simulations than genuine data not to mention it is incredibly disheartening to find something that looks promising only to find that it is a simulated transit.

    Posted

  • mrniaboc by mrniaboc admin, scientist

    Apologies. This it being taken care of. The rate was set higher than we intended.

    Cheers

    Grant

    Posted

  • dfischer123 by dfischer123 scientist

    Dear Dantoto,

    You might read the blog post by Joey Schmitt about our need for synthetic classifications. We have dialed back the number of simulations, but you're actually doing as much for the science when you find (or don't find) the simulated data as when you find the real data. If users identify half of the simulated transits for R= 2 Rearth, then we know that the number of detected 2REarth planets must be doubled. This is something that the Kepler team is struggling to do - you all will discover the frequency of planets of various sizes by telling us what can and can't be detected.

    I know it's not as exciting, but it really is science!

    Thanks so much for your help!
    Debra Fischer

    Posted

  • mschwamb by mschwamb scientist

    Hi,

    You can find the blog post Debra is referring to here.

    Cheers

    ~Meg

    Posted

  • dantoto by dantoto

    Ah I see, sorry that makes much more sense. I guess it also helps weed out the people who just randomly click through and mess up your data. I know it seems rather cynical but humans eh. I have found and tagged a few very strong candidates so far so it would be awesome to see if anything comes of that data.

    Posted

  • SteveKennett by SteveKennett

    As a scientist I would like to see some of the collected statistics on the simulated data and the methods used in the selection of star types into which the simulations are inserted. This is quite a multidimensional problem as the 'detector' is somewhat random. The recognition of patterns with a low signal to noise ratio will depend on quite a number of user factors; training, visual acuity, age of the user, motivation and education all of which are not well known. The star types will have impact on the probability of planet development and stability of orbits and again this will have implications on the observed numbers and the effectiveness of the simulated data.

    I would be interested in helping out if possible.

    regards,

    Stephen

    Posted

  • dfischer123 by dfischer123 scientist

    Hi Steve,

    Thanks for the feedback and interest in the statistical analysis! I wanted to let you know that we are looking into the transit depths for the sims. Something seems amiss to me (but it might be fine). When we generated these, we verified that the simulations were not "impossible" to detect (just more difficult as the planets approach 1 R_earth) - at least for the Mdwarfs, which is what we are showing now.

    The sims will help us capture the empirical range of factors that you describe. But, they've got to be right and I'm double checking everything with the Yale folks this week.

    Posted