Salt
@saltthewind.bsky.social
⏳'03 | 🫶🏼 any pronouns | 💬 eng/spa
your silly compsci soon-to-be graduate neighbor !
🎮 professor layton, ace attorney, ultrakill, minecraft, etc
🎥 doctor who, good omens, jjk, steins;gate, etc
🎲 compsci, math, chess, brain teasers, puzzles
your silly compsci soon-to-be graduate neighbor !
🎮 professor layton, ace attorney, ultrakill, minecraft, etc
🎥 doctor who, good omens, jjk, steins;gate, etc
🎲 compsci, math, chess, brain teasers, puzzles
You're likely right. I'd need to give it a better look, I'm honestly not giving it the proper rigor it deserves. Although, I think one can solve the closed form for the recurrence relation and get a better estimate from there.
October 3, 2025 at 4:41 PM
You're likely right. I'd need to give it a better look, I'm honestly not giving it the proper rigor it deserves. Although, I think one can solve the closed form for the recurrence relation and get a better estimate from there.
Update: With way more datapoints, a linear fit no longer is appropiate, and polynomial of the 3rd degree suits the predictions way better. Under said 3rd degree polynomial fit, we'd expect even odds at around 8 days! In streams of 8 hours at 100% efficiency, it might take 24 days!!
October 3, 2025 at 4:19 PM
Update: With way more datapoints, a linear fit no longer is appropiate, and polynomial of the 3rd degree suits the predictions way better. Under said 3rd degree polynomial fit, we'd expect even odds at around 8 days! In streams of 8 hours at 100% efficiency, it might take 24 days!!
Update: Increasing the sample size, we get a curve!! I guess a linear fit came too early to call it. With a polynomial fit of the 3rd degree, we'd expect a 100% chance of success at around 44597 attempts, which is roughly similar to the mode you found on your simulation!
October 3, 2025 at 4:17 PM
Update: Increasing the sample size, we get a curve!! I guess a linear fit came too early to call it. With a polynomial fit of the 3rd degree, we'd expect a 100% chance of success at around 44597 attempts, which is roughly similar to the mode you found on your simulation!
Thank you very much! Indeed it looks interesting. I haven't revisited college stats so I am very much lost on why it resembles an exponential distribution 🙂 but this is very cool nonetheless! Ty for lending us your processor for a while!!
October 3, 2025 at 5:18 AM
Thank you very much! Indeed it looks interesting. I haven't revisited college stats so I am very much lost on why it resembles an exponential distribution 🙂 but this is very cool nonetheless! Ty for lending us your processor for a while!!
That's pretty sick! Is it too much to ask for a distribution / histogram? I'm very curious about what shape it might take (I'm putting my bets on it resembling a surf wave 🏄♂️)
October 3, 2025 at 4:46 AM
That's pretty sick! Is it too much to ask for a distribution / histogram? I'm very curious about what shape it might take (I'm putting my bets on it resembling a surf wave 🏄♂️)
Yeah. Someone tried doing that over on Twitter: x.com/confusionm8t...
They came to the conclusion it takes around 818 on median to beat the game, assuming 1 coin toss a second. A beefier simulation could aid in verifying, though, in their own admission.
They came to the conclusion it takes around 818 on median to beat the game, assuming 1 coin toss a second. A beefier simulation could aid in verifying, though, in their own admission.
October 3, 2025 at 4:08 AM
Yeah. Someone tried doing that over on Twitter: x.com/confusionm8t...
They came to the conclusion it takes around 818 on median to beat the game, assuming 1 coin toss a second. A beefier simulation could aid in verifying, though, in their own admission.
They came to the conclusion it takes around 818 on median to beat the game, assuming 1 coin toss a second. A beefier simulation could aid in verifying, though, in their own admission.
If the linear trend is to believed, NL might reach a 50% chance of winning the game after around 1772775 attempts. At around 1 attempt a second, it could take around 20 days and a half of straight uninterrumpted attempts. In streams of 8 hours with 100% efficiency, it might take 2 months.
October 3, 2025 at 3:41 AM
If the linear trend is to believed, NL might reach a 50% chance of winning the game after around 1772775 attempts. At around 1 attempt a second, it could take around 20 days and a half of straight uninterrumpted attempts. In streams of 8 hours with 100% efficiency, it might take 2 months.
Here's a small python script to compute the chances for landing the 10 coin flips: gist.github.com/suehtemorp/e...
Here's a barplot for the chances at the first 100k attempts, in 1k groups of a 100 attempts. A linear regression fit yields a R squared statistic of around 99.999% (without much rigor)
Here's a barplot for the chances at the first 100k attempts, in 1k groups of a 100 attempts. A linear regression fit yields a R squared statistic of around 99.999% (without much rigor)
October 3, 2025 at 3:37 AM
Here's a small python script to compute the chances for landing the 10 coin flips: gist.github.com/suehtemorp/e...
Here's a barplot for the chances at the first 100k attempts, in 1k groups of a 100 attempts. A linear regression fit yields a R squared statistic of around 99.999% (without much rigor)
Here's a barplot for the chances at the first 100k attempts, in 1k groups of a 100 attempts. A linear regression fit yields a R squared statistic of around 99.999% (without much rigor)
prezooh! keep your head up 🙂↕️🪇 once you get the hang of it (you will!) it's going to be so much fun
September 25, 2025 at 7:22 AM
prezooh! keep your head up 🙂↕️🪇 once you get the hang of it (you will!) it's going to be so much fun
patched to work with CMake too!! this superbuild thing is taking off. let's make a little engine next 😝😝
August 8, 2025 at 11:27 AM
patched to work with CMake too!! this superbuild thing is taking off. let's make a little engine next 😝😝
with a borrowed toolchain, im stuck with gcc 4 and no C++. ECS in ANSI C? 🙂👈 this is the worst
August 4, 2025 at 6:04 PM
with a borrowed toolchain, im stuck with gcc 4 and no C++. ECS in ANSI C? 🙂👈 this is the worst
OneShot as well.... so tempted to go for that one as well
December 30, 2024 at 2:39 AM
OneShot as well.... so tempted to go for that one as well