5 ms·

I was very surprised that the article made no mention of such a bias. It was my first thought, too. I'm not terribly up-to-date on how Kepler has been detecti

by 3JPLW 13y ago

I was very surprised that the article made no mention of such a bias. It was my first thought, too. I'm not terribly up-to-date on how Kepler has been detecting planets, but isn't it due to the 'wobble' of the stars? Surely they'll find planets that orbit faster (and wobble their stars faster) first. It'll take many many years of very consistent observations (or one very lucky day) to see a wobble (or occlusion) from a planet like our Jupiter.

schiffern 13y ago
Kepler uses occlusion, not wobbling.
- Udo 13y ago
  Still, a very rare event requiring a precise alignment of the star system that gets more improbable and more rare the further out a planet is. And I imagine we need 2-3 data points of transitions at least to exclude false positives - that simply doesn't happen within a reasonable time (e.g. Neptune has an orbital period of 165 years).
- teamonkey 13y ago
  Yes, and it's designed to detect Earth-sized planets as well as larger gas giants.
  - Tuna-Fish 13y ago
    But detection probability is directly correlated to planet size and shortness of the local year. Eventually, Kepler and Kepler-like probes will be able to locate all earth-like planets with orbits that cross the path between the star and us. However, this technique will find the larger planets with shorter orbits first.
    - teamonkey 13y ago
      Yes indeed, but the point is that according to current theories Kepler should already be finding more Earth-sized planets than it has and few inner gas giants. The observable evidence does not match the predicted model.
  - Retric 13y ago
    Can detect is one thing, can detect at the same distance is another. Assume you can detect an earth like planet at say 100 light years and a Jupiter sized planet at 1,000 light years and you find 1000x as many Jupiter sized planets that does not mean there are 1,000x as many just that you can detect them in 1,000x as much space. Granted I don't know there search pattern but we are talking about less than 1,000 systems at this point. And I think we have found 2 reasonably earth like planets already which suggests there are at a minimum millions of them in this galaxy.
- lutorm 13y ago
  True, but occultations are also very biased towards planets with small semimajor axis, because the solid angle that produces an occultation goes roughly as 1/a.
  - schiffern 13y ago
    You use "but" as though you're disputing some statement I made, when I intentionally avoided making any statement outside my knowledge. I said Kepler does use occlusion, and does not use wobbling. I'm not trying to single you out here. I'm just curious what one has to say on HN so as not to be corrected. Is there some "don't assume I've implicitly made a bunch of unsupported remarks" initialism that I can tack on to the end of all my posts? /DAIIMABOUR
    - dandelany 13y ago
      No need to get defensive, lutorm was just pointing out that the assumption made by the OP is still supported. His comment was simply a productive, interesting addition to yours, unlike this complaint.
      - fusiongyro 13y ago
        I'm glad he brought it up. I have noticed it is getting awfully hard to say anything precisely enough to get it past the cadre of pedants.
        cmccabe 13y ago
        Actually, the correct expression is "cohort of pedants."
        fusiongyro 13y ago
        Are these tears of laughter or rage?
sbierwagen 13y ago
Most exoplanets are discovered by the radial velocity method, (tracking the doppler shifts in the spectrographic lines in the light emitted by the star) but the best data is produced by Kepler, which uses occlusion and thus can tell you how big the planet is, by how much the star dims. Both methods are biased towards finding big planets (which produce clear, easy to detect spikes) that are close to their star. (so you get enough data points) Kepler tends to be a little more biased, simply because it hasn't been running long enough: http://en.wikipedia.org/wiki/File:Exoplanet_Period-Mass_Scatter_Discovery_Method_TR.png http://en.wikipedia.org/wiki/File:Exoplanet_Period-Mass_Scat... Kepler was launched in 2009. Realistically, you'd have to run it for another three or four centuries if you wanted to detect exoplanets that orbit as far out as Neptune. (orbital period: 165 years) No wonder we're seeing a lot of planets orbiting close in!
- S4M 13y ago
  That's a very interesting point. I knew about this method but didn't connect it to the article. it would be interesting to have some stats on the number of stars that were probed and that were found to have exoplanets. I have no idea of what the actual numbers are, but if we need to probe 100000 stars to find one that has an exo planet, it means that the radial velocity method might not be that efficient.

by 3JPLW 13y ago

schiffern 13y ago

Kepler uses occlusion, not wobbling.

Udo 13y ago
Still, a very rare event requiring a precise alignment of the star system that gets more improbable and more rare the further out a planet is. And I imagine we need 2-3 data points of transitions at least to exclude false positives - that simply doesn't happen within a reasonable time (e.g. Neptune has an orbital period of 165 years).
teamonkey 13y ago
Yes, and it's designed to detect Earth-sized planets as well as larger gas giants.
- Tuna-Fish 13y ago
  But detection probability is directly correlated to planet size and shortness of the local year. Eventually, Kepler and Kepler-like probes will be able to locate all earth-like planets with orbits that cross the path between the star and us. However, this technique will find the larger planets with shorter orbits first.
  - teamonkey 13y ago
    Yes indeed, but the point is that according to current theories Kepler should already be finding more Earth-sized planets than it has and few inner gas giants. The observable evidence does not match the predicted model.
- Retric 13y ago
  Can detect is one thing, can detect at the same distance is another. Assume you can detect an earth like planet at say 100 light years and a Jupiter sized planet at 1,000 light years and you find 1000x as many Jupiter sized planets that does not mean there are 1,000x as many just that you can detect them in 1,000x as much space. Granted I don't know there search pattern but we are talking about less than 1,000 systems at this point. And I think we have found 2 reasonably earth like planets already which suggests there are at a minimum millions of them in this galaxy.
lutorm 13y ago
True, but occultations are also very biased towards planets with small semimajor axis, because the solid angle that produces an occultation goes roughly as 1/a.
- schiffern 13y ago
  You use "but" as though you're disputing some statement I made, when I intentionally avoided making any statement outside my knowledge. I said Kepler does use occlusion, and does not use wobbling. I'm not trying to single you out here. I'm just curious what one has to say on HN so as not to be corrected. Is there some "don't assume I've implicitly made a bunch of unsupported remarks" initialism that I can tack on to the end of all my posts? /DAIIMABOUR
  - dandelany 13y ago
    No need to get defensive, lutorm was just pointing out that the assumption made by the OP is still supported. His comment was simply a productive, interesting addition to yours, unlike this complaint.
    - fusiongyro 13y ago
      I'm glad he brought it up. I have noticed it is getting awfully hard to say anything precisely enough to get it past the cadre of pedants.
      - cmccabe 13y ago
        Actually, the correct expression is "cohort of pedants."
        fusiongyro 13y ago
        Are these tears of laughter or rage?

sbierwagen 13y ago

Most exoplanets are discovered by the radial velocity method, (tracking the doppler shifts in the spectrographic lines in the light emitted by the star) but the best data is produced by Kepler, which uses occlusion and thus can tell you how big the planet is, by how much the star dims. Both methods are biased towards finding big planets (which produce clear, easy to detect spikes) that are close to their star. (so you get enough data points) Kepler tends to be a little more biased, simply because it hasn't been running long enough: http://en.wikipedia.org/wiki/File:Exoplanet_Period-Mass_Scatter_Discovery_Method_TR.png http://en.wikipedia.org/wiki/File:Exoplanet_Period-Mass_Scat... Kepler was launched in 2009. Realistically, you'd have to run it for another three or four centuries if you wanted to detect exoplanets that orbit as far out as Neptune. (orbital period: 165 years) No wonder we're seeing a lot of planets orbiting close in!

S4M 13y ago
That's a very interesting point. I knew about this method but didn't connect it to the article. it would be interesting to have some stats on the number of stars that were probed and that were found to have exoplanets. I have no idea of what the actual numbers are, but if we need to probe 100000 stars to find one that has an exo planet, it means that the radial velocity method might not be that efficient.