@BrainSteel how do you get 5 for two moves? I get 7. the empty slot can reach 6 possible positions in 2 moves, and there are two ways to reach the one that's a diagonal move from the corner
oh wait
I forgot to remove the two positions you can reach after the first step
if you need a script to compute more test cases: goo.gl/m9n5MF
@MartinBüttner I want to make a feature request but I want a mod's feedback first
A comment thread started out on topic but has slowly drifted to extended discussion that belongs in a chat room. The button to migrate them hasn't appeared yet.
I think it would be cool if regular users could preempt the "move to discussion" prompt
And if all involved in the discussion accept the prompt, automatically migrate the comments to the chat
@Doorknob The point being that no moderater intervention is required
Self moderating comments. Everyone participating says "Yea, we've started an extended discussion." or "No, my comments are directly relevant to the question."
I don't have a good grasp of the actual feature I want to request yet. Maybe a prompt that appears somewhere near the comments, just like the one that appears when comment threads start to get too long? I wonder if I can propose an idea and let others on meta come up with actual implementation details.
Regarding the Strategic Voting KOTH, I think it would be interesting to convert raw scores into z-scores. Since a purely random strategy could produce a wide range of scores, it would be helpful to be able to determine when a bot's score is statically significantly better or worse than a random score.
Today's XKCD is a sports-tournament style bracket, where the contestants are well-known names, put into groups of possibly confusing names.
Give the probability that a given contestant will win the entire tournament, based on each contestant in a given round having an equal chance of winning tha...
@Geobits there are probably positions that can be solved in 79 moves but not 80 moves. That doesn't affect the trait of the sequence that Martin was referring to above
There aren't 16! valid positions, though, are there? I thought only 16!/2 were ever reachable(due to parity). So wouldn't it be 16!/4 for exactly 80 and 16!/4 for 79/81?
I mean, sure, figuring out git can take a long time for some people, but if you just use the github gui client and the github website then you don't need to learn all of that
@flawr no. you have to use a git client. but github has a gui client you can download from their site.
I don't mean for this to sound condescending, but you should probably learn some version control system. It doesn't have to be git. If you learn subversion or mercurial or monotone then there are hosting sites for those too.
I mostly get by with about 5 commands in git, it's only when people start using branches and shouting at me for creating merge commits that it falls down
git is only difficult if you do difficult things with it. if you're the only developer on your project and you don't actually care about branches and history, you can just do 'git commit -a; git push;' after you make a local change and that change will show up on github.
What's the proper algorithm for dividing 100 objects "uniformly" randomly into 3 bins?
My current method is selecting two random integers from 0 to 100 (inclusive), and the first bin gets min(a,b), the second gets max(a,b)-min(a,b) and the third gets 100-max(a,b).
Ok, so for the 15-puzzle: for odd n, f(n) is the sum of all even entries in this sequence. For even n, it's the sum of the odds. Now, how do we generate that sequence easily? ;)
I think the mistake is in the "edge" cases of when there are no objects in a bin.
In the case of two objects, in order for the first bin to be empty, either A or B must be zero. In order for the middle bin to be empty, both A and B must be the same. The first is more likely than the second.
@sparr i am trying it now with the windows gui. i created and added a repository called catch_the_cat, but it does not appear on github.com can you tell me what I have to do now?
@PhiNotPi I can write a Java class that interacts with non-Java submissions. You require all non-Java participants to supply a command that will run their program, and then you download a billion compilers when you want to test it.
I want to take the people who designed Java and Python packages, with their dependency on filesystem directory structure, and force them to use a system where they can't make subdirectories for a while.
But then a jar file packaged using that system wouldn't be compatible with one that did use the filesystem hierarchy setup because java wouldn't know where to look.
I am taking that as the actual question, and it is not correct. A JAR file is a platform-independent entity, based on the ZIP f...
Hmmm. When I got here, I felt confident posting "^ blowjob face". Now, in the year 0 PS (Post StackEgg) I am less confident. Seems like flags fly far too often these days.
So If I read a file "input.txt" within my IDE, it need to be in the main project folder. If from the command line, it needs to be in the /bin folder, or the executable needs to be moved.
@Rainbolt Make sure you post in maximum five-minute intervals the rest of the day so we know you didn't get banned.
if a language required source files to have names at least 100 characters long, that would stupidly exclude platforms with shorter filename limits. that's how I view Java and Python's dependency on folder structures.
@Sparr This might sound dumb, but I was confusing Java with the compiled code (that runs on any platform). Java is really the collection of class files written by a human. I get what you're saying now.
Is there a list, if it's not empty, of contributions to Real Math that PCCG posts have made? Has anyone here ever extended a known OEIS entry? Established a new lower or upper bound for something? These guys over here may have actually done some New Math, which inspired me to ask: reddit.com/r/math/comments/36jq0k/a_curious_property_of_82000
@Rainbolt there's a probabilistic argument in the reddit thread that there is no such number for 2 3 4 5 6, and no such number for 2 3 4 5 that's higher than 82000
Most of the content on PPCG involves clever applications of existing algorithms, or just-fast-enough-to-pass solutions to problems that supercomputers have found millions of answers/digits/etc to already. However, somewhere among all of the brainpower we spend on these challenges, surely someone ...
and I might post a challenge soon to extend an OEIS entry (it only has a handful, and I'm quite confident that a fastest code challenge would beat that limit)
Most of the content on PPCG involves clever applications of existing algorithms, or just-fast-enough-to-pass solutions to problems that supercomputers have found millions of answers/digits/etc to already. However, somewhere among all of the brainpower we spend on these challenges, surely someone ...
I love Unit Testing, and I'd be happy to spend a half day of the company's time learning proper techniques and getting all developers aligned on how we want to implement it.
We are salaried. And I stay late when we have a product to deliver
In fact, now that were are sprinting, we know that we are behind schedule before the night of a drop. So we often stay late in the middle of a sprint just to catch up.
So it's not like we are strict "5:00 time to run away" people.
But this is a process improvement thing. It needs to happen on the clock.
I bring my lunch every Thursday to attend a Lunch and Learn workshop. It's an hour long and we learn a lot. Seems like he could just do the same on Wednesday, so I suggested that.
That was last week, and it sounds like he ignored my suggestion.
I'll make a CW answer for where people can edit in all new/fixed/extended OEIS sequences. doesn't seem like it's worth to have a separate answer for each.
I've determined a rough formula for the relationship between the number of KotH submissions and the length of time I procrastinate compiling/running them, but it still needs a bit of work. Will submit for publishing soon.
I still need Rainbolt's input on that one, I need to run the numbers for GvE to see if they fit the trend.
I spent three weekends from roughly 12:00 AM to 4-6:00 AM downloading entries and compiling and running them. Actual tests ran for maybe 12 hours each, so I also had to implement logic to recover in the event that some stupid submission decided to crash when I wasn't at home.
I'm working on the scoring system for strategic voting. One thing I'm trying to do is figure out a way to determine how many trials should be run to get a meaningful result.
Each "trial" is a single election with three candidates. Each competitor would have randomly-assigned payoffs for each candidate, so they get more points if a certain candidate wins.
The strategy arises when your most-favored candidate isn't likely to win the election overall, leading to cases where it is better for you to vote for your second-most-favored candidate if that is who the other entries are likely to vote for.
The winner of the overall competition is the player with the highest total payoffs of each election. If you are good at picking who to vote for, then your score should be higher.
There is, however, a substantial amount of randomness in the number of points you get from each election.
And so I want to determine how many trials are needed to be able to determine who wins with a decent decree of accuracy.
What I've done so far is calculate the mean and standard deviation of the score of a bot that plays completely randomly. This way, the performance of other bots can be compared to see if they actually are any better.
This is what the tournament results look like for a bunch of bots that vote for their best candidate (not even using strategic voting) with a bunch of bots that vote randomly.