Rakul Posted May 1, 2019 Report Share Posted May 1, 2019 The former. Should be a quicker resolution this month. Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 1, 2019 Report Share Posted May 1, 2019 Interesting. How exactly do we calculate the hotspot probabilities, anyway? Quote Link to comment Share on other sites More sharing options...
Rakul Posted May 2, 2019 Report Share Posted May 2, 2019 Trial and error. Unless the Mars coordinates are discovered. Quote Link to comment Share on other sites More sharing options...
Rakul Posted May 2, 2019 Report Share Posted May 2, 2019 May 2019 Per the Mars hotspot: Moon Lat Moon Lon Probability -21 172 90.51890% Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 2, 2019 Report Share Posted May 2, 2019 Interesting. And what's the correlation between Mars and Moon coordinates, then? Quote Link to comment Share on other sites More sharing options...
Rakul Posted May 2, 2019 Report Share Posted May 2, 2019 There's a formula to it. Someone created a spreadsheet, I believe Chintan, to enter tested coordinates and odds from both the Moon or Mars that kicks back odds on various new coordinates to try. Quite handy! Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 2, 2019 Report Share Posted May 2, 2019 Very cool. If you could find a link to that spreadsheet, I'd be happy to help contribute to the process. In the meantime, I'll test the coordinates you provided. Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 2, 2019 Report Share Posted May 2, 2019 &lat=-21&lon=172 only produces 99% effectiveness. Close, but not quite there. Quote Link to comment Share on other sites More sharing options...
Rakul Posted May 2, 2019 Report Share Posted May 2, 2019 Well damn. 90.5% was pretty good odds. Here's the chart: Moon Lat Moon Lon Probability -21 172 90.51890% -21 173 9.48110% Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 2, 2019 Report Share Posted May 2, 2019 I'll test with my other moon facility. Results to follow. Quote Link to comment Share on other sites More sharing options...
Rakul Posted May 2, 2019 Report Share Posted May 2, 2019 Oh and the spreadsheet... https://forums.cybernations.net/topic/71214-mars-hotspot/page/26/?tab=comments#comment-3497725 Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 2, 2019 Report Share Posted May 2, 2019 I can confirm &lat=-21&lon=173 produces 100% effectiveness. Thanks for the spreadsheet and other data. Quote Link to comment Share on other sites More sharing options...
Rakul Posted May 2, 2019 Report Share Posted May 2, 2019 Whew. Glad to hear it! I have two 99% in my testing and only one 100%. At least I got it. heh I'm not unhappy with 99% personally. Quote Link to comment Share on other sites More sharing options...
Luna Posted May 7, 2019 Report Share Posted May 7, 2019 Hello everyone, it's Luna (formerly known as Chintan). It's been just over a year and a half since I updated the Moon & Mars Hotspot Finder, and while the previous version seems to be working great, I think it's time for an update since we've had 19 new data points since then. So here's the new version (it's the exact same link as before): Moon & Mars Hotspot Finder To use it make a copy ( File -> Make a copy ...), then enter your test spots and give it a few seconds to calculate. Please do this even if you have an older version, because your old copy won't automatically update. Let me know if you find any problems with the spreadsheet, or have any questions or comments. The moon hotspot affects the mars hotspot a LOT, and vice versa. In fact, if one hotspot is found, the number of possibilities for the other one is reduced to 4 maximum (often fewer), and because of the skewed probability distribution, you can usually find the hotspot in 1 try. So definitely check out the other thread. Most people don't realize that the other celestial body affects them. Updates: List of Previously Observed hotspots updated - This is perhaps the biggest change. If you've used this tool before you may have noticed that the hotspots observed in previous months have a much higher probability. Hotspots tend to repeat, so if something was seen in a previous month, the hotspot is more likely to appear there again. There have been 12 newly observed locations in the 19 months since the last update, and all of these have been added to the spreadsheet. This means if one of those 12 locations turns out to be the hotspot again in the future (which will almost certainly happen at some point) you can nail it in fewer tries. Elimination of 30 possible locations, reducing the list from 1282 to 1252 possible locations - This isn't as big a deal as you might think, because those 30 locations all had low probabilities anyway. But it's still nice to see the list get shorter and shorter. And it's a one way change. Once possible locations are gone, they are gone forever and the list can only shrink, never grow. Maybe one day we can get down to the minimum of 1056 possible locations. May 2019 might have been an annoying month because the location with 90.5% odds only gave 99%, and the actual hotspot was the one with 9.5% odds. But the good thing is, those low odds locations are the ones that eliminate possible hotspots. Since they are on the fringes of the probability distribution, they provider stronger data than the boring, routine ones in the middle of the distribution. The last time locations were eliminated was a whopping 26 months ago, back in March 2017, so it is a rare occurrence. Update of probability distribution based on newest data - I calculate the probabilities using a Bayesian model, and adding new data updates the distribution. Having more data narrows the distribution, but you're unlikely to notice the difference. We have 117 months worth of data now, and I updated the probability distribution based on all that data. Quote Link to comment Share on other sites More sharing options...
starschwar Posted May 7, 2019 Report Share Posted May 7, 2019 Thank you for your efforts, Luna. Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 1, 2019 Report Share Posted June 1, 2019 Based on last month's locations (I never moved the 99% facility), here's some results from Luna's table: Moon Lat Moon Lon Mars Lat Mars Lon Probability Previously Observed In These Months -55 101 -84.00000002 -150 89.90760% 12/2012 -55 101 -84.00000002 -149 8.47062% -54 101 -84.00000002 -149 1.62115% -55 101 84.00000002 -150 0.00062% Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 1, 2019 Report Share Posted June 1, 2019 -55, 101 results in 99% effectiveness. Re-evaluating... Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 1, 2019 Report Share Posted June 1, 2019 Moon Lat Moon Lon Mars Lat Mars Lon Probability Previously Observed In These Months -54 101 -84.00000002 -149 100.00000% Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 1, 2019 Report Share Posted June 1, 2019 -54, 101 is still 99%. Curious. Quote Link to comment Share on other sites More sharing options...
Wilcroft Posted June 1, 2019 Report Share Posted June 1, 2019 Here's what I get based on the Mars location: Moon Lat Moon Lon Probability -55 101 98.45671% -55 100 1.54329% Just to note @starschwar: Because floating point numbers (i.e. non-integers) are weird, you get some odd values - that's why the spreadsheet has you enter -50.9999999 rather than -51, for example, even though the number displayed on the map is -51. If you don't have this part correct, the spreadsheet might not work correctly. Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 1, 2019 Report Share Posted June 1, 2019 Thank you, that's very useful information. Using non-decimal values in cybernations, -55, 100 also results in 99% efficiency. Close enough for me. At least we'll have three different data points for me to contribute to next month's search. Quote Link to comment Share on other sites More sharing options...
Colin Myrhh Posted June 8, 2019 Report Share Posted June 8, 2019 Moon base for June @100%: &lat=-54.99999999&lon=101 Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 9, 2019 Report Share Posted June 9, 2019 18 hours ago, Colin Myrhh said: Moon base for June @100%: &lat=-54.99999999&lon=101 Confirmed. Well done. Quote Link to comment Share on other sites More sharing options...
Luna Posted June 10, 2019 Report Share Posted June 10, 2019 (edited) Ok, that's weird. (-55,101) worked last time the hotspot was there back in Dec 2012 (link). Maybe the server side code is running on a different CPU now. Different CPUs have different floating point rounding errors. That's bad news for the spreadsheet because it means the floating point rounding errors for the latitudes may be wrong, and worse, they could be different every month even if the hotspot is a repeat from a previous monnth. For now the spreadsheet values seem to be very good guesses, but we'll see how well that works going forward. At least people know about the rounding error and can adjust if what the spreadsheet suggests doesn't work. Longitudes shouldn't need a floating point correction due to how the great circle formula works. For latitudes start with the closest integer and try +1e-8, -1e-8, +2e-8, -2e-8, etc. (e.g. for -55, try -55, -54.99999999, -55.00000001, -54.99999998, -55.00000002). If +/-2e-8 doesn't get it, you can go on to +/-3e-8 and so on, although if you get that far there's probably some other problem happening. Edited June 10, 2019 by Luna Quote Link to comment Share on other sites More sharing options...
starschwar Posted June 10, 2019 Report Share Posted June 10, 2019 Your insight is appreciated. We'll see how things go next month. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.