Want to help search for Exo planets?

[Daniel_DeGagne]

A few of you might know I play a game called EVE. Few more of you might have heard of it. Normally I wouldn’t do an actual post about it but they’re doing something really really cool atm.

As the title says EVE atm is helping out a group of Scientists searching for planets around other stars.

You can more about it here.: Exoplanets: The Next Phase Of Project Discovery | EVE Online

Just skip the end part where they talk about rewards. Those are only for ingame so unless you want to actually play the game not going to help much.

Now actually doing the search isn’t all that easy. They wouldn’t need help if it was. But there’s lot of samples they train you on then test you on as you go through the pile of data. From what I can see there’s at least a 100,000 stars to go through. And because they are having more then one person doing each star to make sure they get the best chances this could take awhile.

So far I’ve possibly found 1 new star. (as in I was the first person too look at a dataset and found a cycle of dimming which will take further reviews and closer examination to confirm or deny. But hey, its a more note worthy use of my time then actually playing the game.)

Things to Note:

-The game is FREE to play and to do the exo-planet search.

-If you log in you can find me in game under the name Mining Forman. (Yes Forman is misspelled, No it wasn’t by me)
I can help with trying to get started and help make it easier to spot the planet signs.

-You can use this link to sign up. It won’t help you (or me) if you just search for planets. But if you want to try the game out it gives you some extra skill points.

.http://secure.eveonline.com/signup/?invc=929b351a-4606-4ccd-85d2-29d0f5427512&action=buddy

[Daniel_DeGagne]

Thought I’d post some screen shots so you can see what a planet from another solar system actually looks like. Whether its flat or not like ours isn’t determined

These graphs are the actual data from the telescope and show how much light was received from the star. I believe every hour (Not sure) These are 30 days worth of light data.

!
There is obviously something causing this star to dim on a very regular basis. 5.191 days to be exact. Astronomers are of course convinced by now this dimming can only be caused by a planet. An important note though is the scale of the dipping. If you look at the left you’ll see that its around 3-4%. Its huge. Either a Super Jupiter or even a small star.

This is another planet. If you look at the scale you’ll see the drop is less then 1%. Maybe 0.5% But that’s still a lot.
This also gives you a better idea just how much a star light fluxes.
It could also be argued that there might be a smaller planet much closer orbiting say around once a day. But could such a planet be possible. Could just be a very regular star fluctuations during this point in time.

Now this one might at first look like it has a planet as well. But think about it, if there’s a large planet circling around it do you think this is how it would look? Nope its just a pulsating star. Not a pulsar btw that’s something different.

Can you spot the planet?

Here.

Now this is an example of a star they havn’t officially classified with a planet so its up to the players to help pick one out. You’ll notice there’s a light blue line on the left that several players have suggested is a planet. My hunch is that it isn’t though. Because you can’t see it there but if you zoom in you could see that large dip is in fact a single data point. Knowing what I know atm about planets (yes very little but still more then before) this doesn’t really make sense. Suggests to me noise. What you do see though is some small yellow lines on the chart. Several people have marked out something small that might be orbiting every 12-24 hrs or so. Just for all the data you see there isn’t really any good tools in this graph to help isolate those time periods.

Now if your thinking this is easy here’s one that you really have to have a good eye for. Either someone or an AI at the astronomy department has decided there is a planet in all that noise. And no you don’t get those nice yellow lines at the start to help find the planet. More then 50% of the stars look like this or even worse. I failed with this one.

Not sure how many I’ve done so far. Maybe a 1,000. Lot of the easy ones can be done in like 10secs. But the hard ones that may have a planet. Takes like 1-2 mins.

This data is from a satellite called COROT that was launched in 2006 and collected data for 6 years. We’re not even onto the Keplar data yet which has much higher resolution.

Anyways there’s way more to the actual physics behind the whole thing. Put more then 1 planet close to each other and suddenly the regularity pattern gets way way more complex. So far I’ve not seen any obvious stars with more then one planet but they are out there and possibly quite common. Keplar has spotted one star that so far has 7 identified planets all about the size of Earth. With maybe 1-2 in the Goldilocks zone.

[3bien]

This is really cool Dan, thanks for the run-down. Do most people in Eve do
it by eye or do data analysis looking for cyclic patterns?

[Daniel_DeGagne]

You don’t get access to the actual values so all you get to work with is the graph.

And there’s only one option which is to calculate a mean average over a 2, 3, 5 or 24hr time period and remove that so you just see the short fluctuations of data. Detrend as I think mathlabs also called it.

You basically have to look for those valley drops like in the first picture and then try to see if they repeat at regular intervals and mark them on the graph. It does make that process easier because you pick the first point then as you move the mouse it will show 2 or more lines to indicate the orbital period on top of the graph data. So you can see if it aligns with more then 1 drop.

I know that any of the data that shows a short spike in light can only be caused by the sun fluctuations but there is no way to filter that noise out. Would make this way way more easy…

I’ve developed an eye for the more hidden ones but my success rate on the tests is still only about 80%. They throw in periodic test graphs (Like the last picture) to see if your improving, and to weed out those who just click nothing when they don’t see the obvious planet

[mike]

Would you be interested in doing a group night where we sit for an hour or two listening to “The Planets” The Planets - Wikipedia

…while we learn how to find some…

[Daniel_DeGagne]

Yeah that’s an idea. Though to do this you have to install the full game so everyone needs a windows 7 (or higher) machine that can handle DX9 with reasonable speeds. Or Max 10.9 +. And about 10 gigs free. Though the game installs only 2 gigs to start and caches the rest only as you need.

There’s about a 5-10min process to setup the game and player but its all free.

A good mouse and the higher the screen resolution the better… My pictures are at 4K because thats what I find best for the detail. There is a zoom in on the data so if works fine even below 1080. Its just the higher the resolution the less zooming you need to do.