Devblog: Exoplanets: The Next Phase Of Project Discovery


(Edlorna Tinebe) #41

I just learned that if you click on the yellow diamond below your missed transits after grading, it will fold the graph for you. The rest of the controls are still locked, though, and I’d love to be able to tinker with the view after being graded.

(Cliff Beauland) #42

I simply cant understand where there is a patterin in this. It is no different from the surrounding readings - similar results for all time lengths. This is antithetical to the entire tutorial experience.

And theres clearly something up if empty data like that is counted as apart of the pattern.

I wanna be doing this - but I just don’t feel like wasting my time on failure because something is up with TQ.

(Sumo Podenco) #43

Same issue here, I was seriously trying to see something, but then I got this:

I have no clue why, so this ends here for me.

(Valarian Dumonte) #44

Adding my feedback as well. I was doing really good at the exoplanets for the first few after tutorial, but now i get every single one incorrect. I’m done unless this is a bug and needs to be fixed. If it’s fixed i’ll try again, otherwise in its current state I wont participate anymore

(Ideki) #45

I would be nice if we could also zoom in on the fold because it can be hard to get a good alignment otherwise

(Ceja Bayle) #46

The problem with the hidden transitions in background noise is the self-scaling Y-axis (upper left). You can not find any irregularity in an amplitude (and this is what you have to do) when the Y-axis changes in every image section.

(Alan Mathison) #47


Looks like, at this level, we’re getting most of the same patterns. I recognize this light curve as one that I got as well. The blank 4 day area is a data dropout. This can happen for a number of reasons, including technical problems, a change in the satellite orbit, or the star in question going behind the sun or (more rarely) a planet in our solar system.
Alan Mathison, C1
Signal Cartel

(Markavian) #48

Screenshots of my best analysis, that I still failed:

Without averaging the samples to see dips, the data is too noisy to reliably spot the periods. It would be useful after identifying two spots, to be able to divide/multiply the transit.

E.g. if I pick 2 points 6 days apart, it would be useful to split the fold to either 3 days or 12 days to see if the transit matches with existing data.

The other tool I want is the ability to change the sample resolution to smooth over the bumps by producing an average. This might help clean up the view when folded up over dozens of transits.

(Kiauze) #49

That sure doesn’t help, but that alone doesn’t explain what we see.

From the tutorial itself:

And, judging by the examples people have posted, this is not what we’re seeing.

(Edlorna Tinebe) #50

I’ve found myself wanting a button to automatically divide the period by some number.

I’m still not entirely clear on what the detrend button does. Or, rather, I’m not sure what the detrend SLIDER does.

(Cliff Beauland) #51

Folding feels uselessly minutiae due to the lack of clearer resolution and zoom features while folding.

So much noise in the interface that fluctuations are nearly invisible.

(Ideki) #52

Detrend average the curve. So that instead of having an average that goes up and down other large period of time, they all appear to be in the same range.
It is not always useful mind you.

So far I find that I use detrend on curves that are not too noisy.
Otherwise I just try to see a pattern and/or some dips that I can match with some others.

(Veritas Totient) #53

I quit.

(Alan Mathison) #54


I’d like to mention that I’ve done this type of work professionally for about ten years (although with variable stars, not specifically exo-planets (although since exo-planets are what is sexy now, and transiting planets operate exactly like transiting binary stars, although on a smaller level, the variable star community has gotten big into exo-planet detection).

I have to say that I didn’t play around with the program on SISI, so I can’t comment on the changes that were made with what we’re seeing on TQ. Based on the comments, there does seem to have been a change in datasets, at least.

I’d also like to mention that we’re dealing with satellite data and the changes that we’re looking for vary by only a couple of percentage points.

This is real data, folks. It doesn’t surprise me that its messy. Having said that, I suspect that the data we are all getting now have been specifically selected both to test us, and test the system. Look for changes of some sort over the next week.
Alan Mathison, C1
Signal Cartel

(Cliff Beauland) #55

It feels like whatever is analyzing transits - is identifying solar activity as transits - mostly in the eruptive / pulsating variable - rather than true transits.

(Alan Mathison) #56

@Cliff_Beauland, probably not eruptive (cataclysmic) variables. By definition those are too chaotic. But pulsating variables? Yea, that I can see.

I would like to figure out exactly how the “computer” is coming up with the “correct analysis,” I admit. I knew this wasn’t going to be easy, but what we’re getting doesn’t seem statistically significantly above the noise level in some cases.
Alan Mathison, C1
Signal Cartel

(Cliff Beauland) #57

From what I can tell, @Alan_Mathison - this is a fold of the full 30d.

From your experience, is there a pattern in this?

(Armark Bether) #58



I can see a rise, but afaik a rise cannot be a planet. Please fix ? And fix like half of the samples ?

(Alan Mathison) #59

@Cliff_Beauland, I’m more experienced with visual and CCD data that has a greater Y-axis deflection, I grant you. Having said that, no, I can’t see ANY patter in that light curve, especially, specifically, what the computer is showing there.

From what I’ve been able to see, these type of examples are not very rare. Based on the fact that we’re dealing with deflections of a couple of percentage points, especially given error bars, It’s not easy, no, but I agree with you here: I see no pattern there.

In these cases I’d like to get some indication of where this result comes from. Is this an independent computer analysis of the data from some time ago? Is this a human analysis of the data, and if so, how?

I have to believe - and I grant you I AM assuming this, I don’ KNOW - that the images we’re being given now - today, and for the next few days - are calibration images for both us and the system. If that assumption is correct, I would like to know how this particular image is a dip, yes.

It’s not you. :slight_smile:
Alan Mathison, C1
Signal Cartel

(Aineko Macx) #60

[quote=“Joia_Crenca, post:27, topic:8026”]
I’m at 2/10 “Editing the orbital period” and can’t seem to find the action needed to satisfy this section of the tutorial.
[/quote]Yup, same here.