@@bagelnine9 tags I think there called are users put on games so people count it as psychological horror also blender is counted as psychological horror
i've actually used the hypercube --> table of squares idea in reverse for an implementation of sudoku. In essence, sudoku is a 3x3 grid of 3x3 grids, which is what makes this possible. This is a little differrent from the video, as the 4D grid was only an internal part to hold the state of the sudoku board (I still displayed it in 2D normally). I did this because it makes checking the boxes as easy as it is to check the row and columns.
My favorite setting is using a 5x5x5x5 grid, with wrap-around on in all 4 dimensions. Then I have it place 55 to 65 mines. I don’t play every game, I use the auto-start that looks for a single cell with no neighboring mines. Most of the time there isn’t one. But if there is one, I play that grid. It’s… quite challenging. Most of it can be solved using single-dimensional deduction, especially with only 55 mines. Bump it up to 60 or 65 and the amount of opportunities where two-dimensional deduction becomes necessary is greatly increased. I’d considered coding up the game and coding up a machine to play it using one-dimensional deduction. Coding up two-dimensional deduction would be hard. Okay, I just finished watching the video… yeah, I might look at your bot just to use it to run the auto-start since that is such a pain in the ass, having to generate dozens of games before I can get one that has a single zero-space to start on. And once I filter them that way, my win rate on full wrap-around is much higher than your bot’s. I also track the number of guesses I have to make, where I can’t absolutely know the square I’m clicking is safe. It’s usually like 4 or 5. I’ve been wanting to maybe do something like this for a while, and I know some Python, I may see if I can get this bot up and solving these like it means to win.
Wow, 5^4 with 60 mines and wraparound! That is mighty impressive. As a human I was totally lost at 30 mines, 20 was doable but very challenging. Also I heavily rely on corners and edges, so with wraparounds I'd be extra lost. Let me know if you have any questions about the code. Although from what you are describing you don't really need a full-scale solver. Just click the cell and see if there are any white on the screen. Click new game if there isn't.
@@GamesComputersPlay 5^4 with 20 mines was a good place to start as a human. At that density, it was similar to regular 2d minesweeper. I figured out some techniques from there as I bumped up the mine count. I think I want to code them up. Do you have a recommendation for an IDE? When I’ve used Python before, it was always with the Python editor acting essentially as a front-end to control elements of a program that couldn’t be controlled by a GUI. So I’ve never run standalone Python code.
@@ahgflyguy I started with IDLE, the one that comes with the Python by default. By now I outgrew it and use Visual Studio Code - I think it's one of the most popular code editors overall, if not the most popular. It may be a bit overwhelming at first (it was for me), but I can see why it is so popular - it is quite good once you have learned the ropes. And free too. I don't have much experience with coding GUIs with Python - I saw it in the course I took, but never tried it myself.
You can probably attach debugger to the process, read the grid from the memory dump and trigger routines in the code. That would be much faster to run than interacting with GUI in python, which is already slow as F.
I've just downloaded and played my first game of the 4D Minesweeper and it took me 37 minutes to beat 10 mines on the 4x4x4x4 hypercube. I tried 15 mines on the same hypercube and was able to beat it in 23 minutes. Then I tried 30 mines and after 57 minutes in, when I had 157 revealed cells, I made a mistake and lost that was painful because I started without using good start and it had a potential.
Update: I have just beaten 30 mines on 4x4x4x4 on my 3rd try (no "nice start"). In-game time is 1 hour 35 minutes and 12 seconds but I paused a few of times to solve a couple systems of equations so I'd say overall time is about 2 hours. Looking forward into 5x5x5x5 and more mines :)
Yes, you are right. I remember Inotices wraparounds and played with them a few games. Somehow by thetime the bot was ready I forgot about them. I should add this waparound logicto the simulator too.
I agree, it can be annoying when AI is everywhere, where it is not too applicable. There are still problems where good old algorithms are still the best choice.
If you want your brain to explode you should check out the 4d Rubik’s cube someone made. I think there’s a 5d one as well but the representation kinda breaks down at that point
Also on Steam is a 4D version of Snake that uses this same table-of-squares representation strategy! It's not free, but it is only seven dollars and has an insane number of different gamemodes that I think are well worth the price, so I can recommend it regardless.
Who knows, in a few years maybe mentioning if it is a human / robot will be normal for sports / entertainment etc... We had the first AI rapper a few months ago, I think AI writers / painters / entertainers are not that far away.
Yaaay new video! I was wondering, how would you go about solving a puzzle on a hexagonal grid? I’d like to try making a program for the mini puzzles in Opus Magnum, but I can’t figure out how to represent it.
I haven't played that one , so can't say about this particular game (although it was on my wishlist, so I am planning to get to it someday...) I know how I would approach a game like HexCells (which is minesweeper on hexagonal board) - the same logic as 4D: ignore the physical dimension, just look at which cells are the neighbors. From the screenshots of Magnum Opus, I'd say something similar can work for this game: ignore the grid, don;t try to put them in an array. Instead, treat cells as nodes in the graph. Might be a little cumbersome to build such a graph, but once you do that, things should get much easier.
For something even more funky, consider pentagons with similar shapes to the faces of the dual of the snub dodecahedron, except use one 60 deg. angle and four 120 deg. ones in each so they tile the plane. You could also mess with 'left' or 'right' handedness.
Thanks! I know it is virtually impossible to get rid of the accent completely, especially when you learned English in a regular soviet school and had your first conversation with a real foreigner in your twenties :) But I strive for something like "you don't sound like a native speaker, but I'm not sure where to place your accent".
It is interesting that the 5^4 grid is so much harder than a 4^4 one. The 4^4 reaches 40% win rate at ~12% mine density, while the 5^4 one does the same at ~4.5% mine density already! How does this trend continue for even larger grids?
Yes, I noticed that too! At first I thought it was weird, but this is an explanation I have: The reason I think are corners and edges - they are a great help when playing. Size 4 cube has higher "corners / all cells" ratio than size 5 - so, higher success rate. So I guess bigger cubes (tesseracts too) with the same density would become a little harder, but by ever smaller margin. It's the same with 8x8 beginner and 16x16 Intermediate. While the density is exactly the same, a beginner has a slightly higher success rate. I think I should add a "wrap around" option to the simulator, which has exactly zero corners. Guessing that success rate would plumeth for all dimensions.
@@GamesComputersPlay That is true. The size 4 tesseract has 6.25% interior cells, while the size 5 one has 12.96% interior cells, vs 56.25% and 76.5625% interior squares for 8x8 and 16x16 grids respectively. If that hypothesis is correct, then the win rate vs mine density graph should become independent of grid size with wrap around enabled.
Actually, I may be dead wrong. I have implemented a wrap-around version and first thing I noticed is 2D expert win rate is 50% higher on wrap-around. Need more research to figure out what's going on. Maybe it has something to do with the number of neighbors - so it has different effect on different dimensions. Or mayeb mine density and corners together have some unusual non-linear effect on win rate
@@GamesComputersPlay Depending on how you start your 2D boards, I can see why there is an increase. The only benefit of edges in 2D minesweeper (in particular, corners) is the greatly increased odds of getting an opening - something vital in the standard first click safe (but not an opening) variation. After that initial opening, however, edges are worse - as they decrease the amount of information you can get resulting in far more likely forced guess (50/50) scenarios.
What if there was a 3d minesweeper cube that had a coat of 2d minesweeper you had to clear to play inside. Or open world minesweeper. Or minesweeper blocks in minecraft. Or minesweeper rpg
I think there is a minesweeper RPG. I haven’t played it but I wouldn't expect much. It's like a black jack first person shooter... or point-and-click chess... If you can do it does not mean you should do it.
50/50 - yes, but not always. Any non-certain move can be the last one. Even if it finds a cell that is, say 80% safe, it is still 20% a mine and could end the game.
@@GamesComputersPlay ah alright, and why would there be an 80% safe? I think because there isn't enough info yet? Or only 50/50s left and that's the best option?
I really enjoyed watching that. That said, I watched a guy play with way more mines and win on TH-cam recently. I'm fairly sure that he's better than your bot.
I didn't look into that much, but I think you are right. I saw in the program there are some presents of saved games (from the author, I assume) with different mine counts and solution times. Number count goes from 20 to 40. I simulated those settings. With 4^4 40 mines, the bot's win rate is around 2-3%. I doubt humans would enjoy playing with such a low win rate. I have 2 possible explanations: 1. The "Smiley face" button. Apparently the game offers the "Start with the opening" button, which I completely missed. Maybe that tips the scale enough for those 40-mines games to be winnable and enjoyable. 2. There are, indeed, strategies specific to 4-dimensional minesweeper, that I don't know about and don't use. I kinda think it's probably the first one, but I can't be sure.
Yes, I found that video, very interesting! Actually, seeing how confidently he solves it, makes me want to check if there is something missing in the bot. Maybe there is something in those 4 dimensions I didn't take into account.
Well, there is no readily available function for that, but there is a way to make it work. Create a new minesweeper game with custom field (it can read it from a string) Try playing the game (the way a simulator does it). The solver can provide information which method was used for the latest move. If it is a "Probability" method - it is a guessing game. I also notice that usually "no guessing" games indicate where the first move has to be made. In case of this solver the first move is always at (0,0) - so player should start the game in the same way.
Funny enough, I have an idea for a video, where 1D minesweeper would appear. We'll see, too early to say anything. 1D minesweeper is actually way less player-friendly than 2D. Lots of guessing.
I have thought about it actually. There are a few ways of approaching it - each with its own drawbacks. One would probably be a machine learning. Which I am still not too confident in. Also, it would take an enormous amount of trial and error - and this is why, for example Code Bullet recreates games first, to be able to run multiple trials at the same time, not being bound by the actual game. I’d rather find a way to solve it, while only using the actual game. Maybe someday…
I haven't - I only tried hexagonal (the game was called HexCells and I can't recommend it enough). I'm still not sure if all those non-square variant can even work in 3D to be honest.
It was written with 3.10, but I don't see why it shouldn't work on any slightly older version. By solve boards you mean launch the bot or solve any position you give it? Bot should be easy - just launch minesweeper_bot file, make sure you have the minesweeper game on the screen and click F10. If it can do it - it will do it. If it can't - it is probably not the default skin or size. If you want to just give any position to it and see which cell it recommends to click - there is no convenient way of doing it, but if you are ready to get your hands dirty - it wouldn't be too hard to make it work like that. There is a MinesweeperSolver object that needs to be initiated with the field size and mines number. After that there is a "solve" function within that object that would accept an NP array with the field data (1-8 for numbers, various values for closed cell, mine, etc). That function will return lists of safe cells and cells with mines it could find (or a cell that would be best to pay if it had to guess). There are maybe a few bits that I am missing - but that should be the general idea. I tried to write it as clearly as possible with plenty of comments, so with a little effort, it should be possible to figure stuff out.
@@paarsenathan6087 In the "main" method you'll see there is a parameter passed to the program. That's the number of games it would play one after another - you can change it by changing that number. Also, there is a "fail-safe" built-in in a clicking module - if you quickly move mouse to the very top left of the screen, if would force program to stop.
Exponential growth is a dangerous thing. Simulating a minesweeper game on a 8-dimensional cube with a size of 4 almost crashed my laptop. (To bee fair, current solver is optimized for highest win rate, not for robustness, but still). And that was only 8 dimensions...
Что интересно буквально вчера кто-то выложил перевод моего предыдущего видео про сапера: th-cam.com/video/uHAREopJlW4/w-d-xo.html А так вообще нет, только на английском.
I've not even seen 3d minesweeper before, this is insane
The thing that made me laugh the hardest is that 4D Minesweeper is part of the *Psychological Horror genre.*
If that's not AI-generated, that's kinda hilarious.
@@bagelnine9 tags I think there called are users put on games so people count it as psychological horror also blender is counted as psychological horror
I mean, I looked at it and got scared.....
Can't wait till this guy discovers 5D Chess with Multiverse Time Travel, also available on Steam.
i've actually used the hypercube --> table of squares idea in reverse for an implementation of sudoku.
In essence, sudoku is a 3x3 grid of 3x3 grids, which is what makes this possible.
This is a little differrent from the video, as the 4D grid was only an internal part to hold the state of the sudoku board (I still displayed it in 2D normally).
I did this because it makes checking the boxes as easy as it is to check the row and columns.
My favorite setting is using a 5x5x5x5 grid, with wrap-around on in all 4 dimensions. Then I have it place 55 to 65 mines. I don’t play every game, I use the auto-start that looks for a single cell with no neighboring mines. Most of the time there isn’t one. But if there is one, I play that grid. It’s… quite challenging. Most of it can be solved using single-dimensional deduction, especially with only 55 mines. Bump it up to 60 or 65 and the amount of opportunities where two-dimensional deduction becomes necessary is greatly increased. I’d considered coding up the game and coding up a machine to play it using one-dimensional deduction. Coding up two-dimensional deduction would be hard. Okay, I just finished watching the video… yeah, I might look at your bot just to use it to run the auto-start since that is such a pain in the ass, having to generate dozens of games before I can get one that has a single zero-space to start on. And once I filter them that way, my win rate on full wrap-around is much higher than your bot’s. I also track the number of guesses I have to make, where I can’t absolutely know the square I’m clicking is safe. It’s usually like 4 or 5. I’ve been wanting to maybe do something like this for a while, and I know some Python, I may see if I can get this bot up and solving these like it means to win.
Wow, 5^4 with 60 mines and wraparound! That is mighty impressive.
As a human I was totally lost at 30 mines, 20 was doable but very challenging. Also I heavily rely on corners and edges, so with wraparounds I'd be extra lost.
Let me know if you have any questions about the code. Although from what you are describing you don't really need a full-scale solver. Just click the cell and see if there are any white on the screen. Click new game if there isn't.
@@GamesComputersPlay 5^4 with 20 mines was a good place to start as a human. At that density, it was similar to regular 2d minesweeper. I figured out some techniques from there as I bumped up the mine count. I think I want to code them up. Do you have a recommendation for an IDE? When I’ve used Python before, it was always with the Python editor acting essentially as a front-end to control elements of a program that couldn’t be controlled by a GUI. So I’ve never run standalone Python code.
@@ahgflyguy I started with IDLE, the one that comes with the Python by default. By now I outgrew it and use Visual Studio Code - I think it's one of the most popular code editors overall, if not the most popular. It may be a bit overwhelming at first (it was for me), but I can see why it is so popular - it is quite good once you have learned the ropes. And free too.
I don't have much experience with coding GUIs with Python - I saw it in the course I took, but never tried it myself.
You can probably attach debugger to the process, read the grid from the memory dump and trigger routines in the code. That would be much faster to run than interacting with GUI in python, which is already slow as F.
WOW i just watched your other minesweeper videos for fun, didn't expect this video to be out right NOW
yes
cant wait for 5d minesweeper
I've just downloaded and played my first game of the 4D Minesweeper and it took me 37 minutes to beat 10 mines on the 4x4x4x4 hypercube. I tried 15 mines on the same hypercube and was able to beat it in 23 minutes. Then I tried 30 mines and after 57 minutes in, when I had 157 revealed cells, I made a mistake and lost that was painful because I started without using good start and it had a potential.
Update: I have just beaten 30 mines on 4x4x4x4 on my 3rd try (no "nice start"). In-game time is 1 hour 35 minutes and 12 seconds but I paused a few of times to solve a couple systems of equations so I'd say overall time is about 2 hours. Looking forward into 5x5x5x5 and more mines :)
Real men play 5D minesweeper
Not to mention there are also wraparounds which remove the edges and corners completely
The wraparounds are absolutely key. The way I play, they are necessary.
Yes, you are right. I remember Inotices wraparounds and played with them a few games. Somehow by thetime the bot was ready I forgot about them.
I should add this waparound logicto the simulator too.
Thanks for not calling this AI and for not trying to solve the problem using a neural network.
I agree, it can be annoying when AI is everywhere, where it is not too applicable. There are still problems where good old algorithms are still the best choice.
If you want to know the formula for the maximum number of mines in one space could be, it's really simple.
The formula is-
(3^d)-1 (d=dimention)
do 5d minesweeper with multiverse time travel next
If you want your brain to explode you should check out the 4d Rubik’s cube someone made. I think there’s a 5d one as well but the representation kinda breaks down at that point
Also on Steam is a 4D version of Snake that uses this same table-of-squares representation strategy! It's not free, but it is only seven dollars and has an insane number of different gamemodes that I think are well worth the price, so I can recommend it regardless.
Thanks for the tip! Looks like it has a free demo too.
I don’t even saw a 3D minesweeper before and now you told me that there is a 4d version and even there exists a bot that can solve it 😮
My favourite part is the "Psychological horror" tag on steam
Very interesting concept.
Nice work! I found this game as I was about to download the 4D miner demo from steam, which is more or less 4D minecraft. Also $0.
It always gives me a chuckle when you remark in the captions that you are a human.
Who knows, in a few years maybe mentioning if it is a human / robot will be normal for sports / entertainment etc... We had the first AI rapper a few months ago, I think AI writers / painters / entertainers are not that far away.
@@GamesComputersPlay True
Yaaay new video! I was wondering, how would you go about solving a puzzle on a hexagonal grid? I’d like to try making a program for the mini puzzles in Opus Magnum, but I can’t figure out how to represent it.
I haven't played that one , so can't say about this particular game (although it was on my wishlist, so I am planning to get to it someday...)
I know how I would approach a game like HexCells (which is minesweeper on hexagonal board) - the same logic as 4D: ignore the physical dimension, just look at which cells are the neighbors.
From the screenshots of Magnum Opus, I'd say something similar can work for this game: ignore the grid, don;t try to put them in an array. Instead, treat cells as nodes in the graph. Might be a little cumbersome to build such a graph, but once you do that, things should get much easier.
For something even more funky, consider pentagons with similar shapes to the faces of the dual of the snub dodecahedron, except use one 60 deg. angle and four 120 deg. ones in each so they tile the plane. You could also mess with 'left' or 'right' handedness.
I usually cringe when i hear my compatriots accent and I am kind of surprised by the lack of thereof in your speech, very good! :O
Thanks!
I know it is virtually impossible to get rid of the accent completely, especially when you learned English in a regular soviet school and had your first conversation with a real foreigner in your twenties :)
But I strive for something like "you don't sound like a native speaker, but I'm not sure where to place your accent".
It is interesting that the 5^4 grid is so much harder than a 4^4 one. The 4^4 reaches 40% win rate at ~12% mine density, while the 5^4 one does the same at ~4.5% mine density already! How does this trend continue for even larger grids?
Yes, I noticed that too! At first I thought it was weird, but this is an explanation I have:
The reason I think are corners and edges - they are a great help when playing. Size 4 cube has higher "corners / all cells" ratio than size 5 - so, higher success rate. So I guess bigger cubes (tesseracts too) with the same density would become a little harder, but by ever smaller margin.
It's the same with 8x8 beginner and 16x16 Intermediate. While the density is exactly the same, a beginner has a slightly higher success rate.
I think I should add a "wrap around" option to the simulator, which has exactly zero corners. Guessing that success rate would plumeth for all dimensions.
@@GamesComputersPlay That is true. The size 4 tesseract has 6.25% interior cells, while the size 5 one has 12.96% interior cells, vs 56.25% and 76.5625% interior squares for 8x8 and 16x16 grids respectively.
If that hypothesis is correct, then the win rate vs mine density graph should become independent of grid size with wrap around enabled.
Actually, I may be dead wrong. I have implemented a wrap-around version and first thing I noticed is 2D expert win rate is 50% higher on wrap-around.
Need more research to figure out what's going on. Maybe it has something to do with the number of neighbors - so it has different effect on different dimensions. Or mayeb mine density and corners together have some unusual non-linear effect on win rate
@@GamesComputersPlay Depending on how you start your 2D boards, I can see why there is an increase. The only benefit of edges in 2D minesweeper (in particular, corners) is the greatly increased odds of getting an opening - something vital in the standard first click safe (but not an opening) variation. After that initial opening, however, edges are worse - as they decrease the amount of information you can get resulting in far more likely forced guess (50/50) scenarios.
3:55 just so you know if you click the :-) button it will start you on a 0
Oh, I didn't know that. I actually thought it was a new game button (like in windows minesweeper), that for some reason didn't work...
Wow great video!
We're going to need 5D minesweeper for a 81
What if there was a 3d minesweeper cube that had a coat of 2d minesweeper you had to clear to play inside. Or open world minesweeper. Or minesweeper blocks in minecraft. Or minesweeper rpg
I think there is a minesweeper RPG. I haven’t played it but I wouldn't expect much. It's like a black jack first person shooter... or point-and-click chess...
If you can do it does not mean you should do it.
awsome!!! :) now 5d minesweeper
nice vid
Cool! I came from ur comment btw
Hey! Love the video! I have a question: why would it fail? Because of 50/50s?
50/50 - yes, but not always. Any non-certain move can be the last one. Even if it finds a cell that is, say 80% safe, it is still 20% a mine and could end the game.
@@GamesComputersPlay ah alright, and why would there be an 80% safe? I think because there isn't enough info yet? Or only 50/50s left and that's the best option?
@@kr1v Yes, it will turn to guessing only if there is no sure safe cells left
i love minesweeper
Underexposed channel, hope you grow bigger
I really enjoyed watching that.
That said, I watched a guy play with way more mines and win on TH-cam recently.
I'm fairly sure that he's better than your bot.
I didn't look into that much, but I think you are right.
I saw in the program there are some presents of saved games (from the author, I assume) with different mine counts and solution times. Number count goes from 20 to 40. I simulated those settings. With 4^4 40 mines, the bot's win rate is around 2-3%. I doubt humans would enjoy playing with such a low win rate.
I have 2 possible explanations:
1. The "Smiley face" button. Apparently the game offers the "Start with the opening" button, which I completely missed. Maybe that tips the scale enough for those 40-mines games to be winnable and enjoyable.
2. There are, indeed, strategies specific to 4-dimensional minesweeper, that I don't know about and don't use.
I kinda think it's probably the first one, but I can't be sure.
Yooooo Aliensrock just recently played this
Yes, I found that video, very interesting! Actually, seeing how confidently he solves it, makes me want to check if there is something missing in the bot. Maybe there is something in those 4 dimensions I didn't take into account.
In 4d there are 8p sourunding squares 3x3x3x3 - 1 is 80
I can't play 2d minesweeper, but damn, I really want to to try and play it in 4d, my brain is gonna melt if I try, but I really wanna try
So 80 is the most you can get in 4d minesweeper.
I watched the last video just yesterday
Hey, I have that game!
lets computer plays original flippy bird
imagine 5d minesweeper
with interdimensional time travel
I’m trying but my tiny ape brain refuses to comprehend it for me.
@@Ataraxia0 it's "multiverse time travel"
@@Ataraxia0 and an extra half-dimension
If you are active now, try botting skyplunger game, it will be harder one you ever did.
"hey guys, tyler here, welcome to minesweeper 4d..."
I know what you talking about, I just watched that video. :)
You should make a program that can make minesweeper for any desired dimension
I kinda did. The problem with dimensions higher that 4 is to how to display them in a not too overwhelming way...
@@GamesComputersPlay yeah, probably
Can the algorithm determine if a given minesweeper board is solvable without guessing?
Well, there is no readily available function for that, but there is a way to make it work.
Create a new minesweeper game with custom field (it can read it from a string)
Try playing the game (the way a simulator does it).
The solver can provide information which method was used for the latest move. If it is a "Probability" method - it is a guessing game.
I also notice that usually "no guessing" games indicate where the first move has to be made. In case of this solver the first move is always at (0,0) - so player should start the game in the same way.
Will you return? :)
Yes, but what about 81?
the fourth dimension getting out of hand
What about an 81-
what about at 729?
i know you do simpler games but would it be possible to make python play Terraria?
ok, time for 27
When 1D
Funny enough, I have an idea for a video, where 1D minesweeper would appear. We'll see, too early to say anything.
1D minesweeper is actually way less player-friendly than 2D. Lots of guessing.
What about an 81?
Isnt it like a 3d representation of a 4d thing in a 2d screen in a 3d world
Well it uses squares not cubes so idk why it would be a 3d representation of anything…
Make computer plays "The World's Hardest Game" :D
I have thought about it actually. There are a few ways of approaching it - each with its own drawbacks.
One would probably be a machine learning. Which I am still not too confident in. Also, it would take an enormous amount of trial and error - and this is why, for example Code Bullet recreates games first, to be able to run multiple trials at the same time, not being bound by the actual game.
I’d rather find a way to solve it, while only using the actual game. Maybe someday…
Have you ever tried triangle minesweeper? You get 12 connections on 2d, I'm not even sure how many you would get in 3d.
I haven't - I only tried hexagonal (the game was called HexCells and I can't recommend it enough).
I'm still not sure if all those non-square variant can even work in 3D to be honest.
Add a fifth dimension.
What python version is the solver for
and how do I make it solve boards
It was written with 3.10, but I don't see why it shouldn't work on any slightly older version.
By solve boards you mean launch the bot or solve any position you give it?
Bot should be easy - just launch minesweeper_bot file, make sure you have the minesweeper game on the screen and click F10. If it can do it - it will do it. If it can't - it is probably not the default skin or size.
If you want to just give any position to it and see which cell it recommends to click - there is no convenient way of doing it, but if you are ready to get your hands dirty - it wouldn't be too hard to make it work like that.
There is a MinesweeperSolver object that needs to be initiated with the field size and mines number. After that there is a "solve" function within that object that would accept an NP array with the field data (1-8 for numbers, various values for closed cell, mine, etc). That function will return lists of safe cells and cells with mines it could find (or a cell that would be best to pay if it had to guess).
There are maybe a few bits that I am missing - but that should be the general idea. I tried to write it as clearly as possible with plenty of comments, so with a little effort, it should be possible to figure stuff out.
@@GamesComputersPlay pressing f10 made it work thx is there any way to stop it or does it stop on its own
@@GamesComputersPlay is there a way to make it work on custom boards or does it already do that
@@paarsenathan6087 In the "main" method you'll see there is a parameter passed to the program. That's the number of games it would play one after another - you can change it by changing that number.
Also, there is a "fail-safe" built-in in a clicking module - if you quickly move mouse to the very top left of the screen, if would force program to stop.
k, well what about 5D minesweeper?
Where are you?
In China
holy f
play 3d minesweeper in minecraft
Only 4D? Why bother if you aren't going for a full 10?
Exponential growth is a dangerous thing.
Simulating a minesweeper game on a 8-dimensional cube with a size of 4 almost crashed my laptop. (To bee fair, current solver is optimized for highest win rate, not for robustness, but still). And that was only 8 dimensions...
@@GamesComputersPlay truly our computers have hit their limit- *big minesweeper*
What about 81? 😂
Max number in 2d is a 8
In 3d it's a 26
In you're playing 4d, try to get 80, LOL)
i didnt even understand what is the difference between 2d and 4d here :/
А на русском видео есть? Или это только для англичан канал? В описании написано что вы из России)
Что интересно буквально вчера кто-то выложил перевод моего предыдущего видео про сапера: th-cam.com/video/uHAREopJlW4/w-d-xo.html
А так вообще нет, только на английском.
@@GamesComputersPlay ясно, удачи в ведении канала
9:29 WHERE 4D?
Underwhelminly mundane, isn't it? And that was exactly the point.
isnt 4d minesweeper still in 3 dimensions just using dimensions 1,2 and 4 in a sense?
pshh 4D is yesterdays discovery, wheres 5D minesweeper 🥱