Skip to content

Latest commit

 

History

History
142 lines (117 loc) · 5.81 KB

README.md

File metadata and controls

142 lines (117 loc) · 5.81 KB
Raw

How To Play

  • Click-Here to play now on web or mobile. You can also build the game yourself.
    • Start clicking on the gameboard! }
  • To modify game settings just click Settings
    • Skill is ranged between 1 and 5
      • 1 is random placement
      • 2-4 allows for deviations from the best position by a specific percentage, when applicable
        • 2 is 75% deviation
        • 3 is 50% deviation
        • 4 is 25% deviation
      • 5 is best placement or 0% deviation

Evaluation Colors

  • Evaluations colors can be enabled (default) or disabled via the Settings screen
  • The visible the color the more valuable the position
    • Red indicates is player O
    • Green indicates is player X
    • Various shades of color mixing between Red and Green reprents the shared value of that square between players

Cool Stats

To provide the statistics above, it took ~18 minutes for 8 threads to individually generate 125,000 games each at ~8.3 ms/game. See the How To Generate AI/ML Datasets For Training section to learn how to generate your own datesets for statistical analysis and AI/ML training!

1000000 games, at expert, on a 10x10 board

  • Games Drawn: 89.32%
  • O Wins: 2.02%
  • X Wins: 97.98%
  • Compilation: 8 threads at ~8.3 ms/game with total time of ~18 minutes

1000000 games, at expert, on a 9x9 board

  • Games Drawn: 21.25%
  • O Wins: 36.76%
  • X Wins: 63.24%
  • Compilation: 8 threads at ~1.7 ms/game with total time of ~7 minutes

How To Generate AI/ML Datasets For Training

  • Open the connect.html file in your web browser from a website or a local web server
  • Click on the DB link at the top of the page
  • Change configure the settings for your needs
  • Click the Apply button
  • Wait until threads complete and the data has been processed
  • Click the Download button to download the newly generated DB, in your specified format, to your computer

MultiThreaded DB Generation

  • In the Gameplay DB Generator page, scroll down until you see Threads. Increase the number of threads to scale the multithreading capabilities of the generator.
  • MultiThreading is accomplished via JavaScript's WebWorkers

How To Parse

Clipboard Copies

  • Example: X0303035O;512,257,513,514,258,0
    • Part1: X0303035O
      • *0303035O represents the player played as X or O
      • X**03035O represents the gameboard size in the A direction
      • X03**035O represents the gameboard size in the B direction
      • X0303**5O represents the connect size
      • X030303*O represents the O opponent's skill where 0 is an AI/ML engine and 1-5 is noob-to-expert
      • X0303035* represents the winner as X, O, or D for draw
    • Part2: 512,257,513,514,258,0
      • This is an array of postionHashes (see DB Parsing for how to parse). The first value is always X.
    • Description: This was a human vs computer game on a 3x3 gameboard with a connect size of 3 (tic-tac-toe). The computer was playing on expert difficulty, and won after a total of 6 moves.

DB Parsing

  • Common to all formats
    • X always goes first
    • Positions are hashed into hexadecimal format
      • EG: Placement (A2,B3) is encoded as 0x0203
      • // Decode
(positionHash: number) => { A: ((positionHash >> 8) & 0xff), B: (positionHash & 0xff)}

// Encode
(A: number, B: number) => ((A & 0xff) << 8) | (B & 0xff)
  • CSV
    • This format is file size optimized
    • Game data example OX:1,2,3;OX:2,3,4!OX:4,5,6,7,8,9; ...
      • OX:1,2,3; the length of the array is 3. 3 is odd, so X won. Unless a ! appears instead of a :. 
        • let winnerX: boolean = Boolean(array.length % 2)
      • OX:...;
        • O represents the skill of O in a single digit. A value of 0 indicates an AI/ML engine was used.
        • X represents the skill of X in a single digit. A value of 0 indicates an AI/ML engine was used.
      • OX!...;
        • The ! instead of : indicates that the game was a draw
  • JSON
    • This format is compute optimized
    • Games are stored in games: HistoryReportInstance[]
      • HistoryReportInstance is defined as 
        • {
	h: number[], // history of placements
	o: number, // skill of player o
	x: number, // skill of player x
	w: boolean, // true is o winner, false is x winner, and null is drawn game
}

Build

Node.js is required to build this app nodejs.org

Output files from the build processes are stored in the dist directory

All

  • npm install to download the dependencies to your node_modules directory

Dev

  • This is for active coding/development
  • npm run dev to watch for code changes and live-reload browser if changed
  • npm run test-dev to watch for code changes and re-run unit tests (Jest) if changed
    • Leverage the printGameboard(displayOPieces?: boolean, note?: string) function to visualize the evaluations in the command line 
      • *title-here*
   - *note-here*
(x)  A0   A1   A2   A3   A4   A5   A6   A7   A8   A9
B0 [   ][   ][   ][   ][   ][   ][   ][   ][   ][   ]
B1 [   ][   ][   ][   ][   ][   ][   ][   ][   ][   ]
B2 [   ][   ][   ][   ][ 25][   ][   ][   ][   ][   ]
B3 [   ][   ][   ][   ][ X ][   ][ 25][   ][   ][   ]
B4 [   ][   ][ 25][ X ][ X ][ X ][ 25][   ][   ][   ]
B5 [   ][   ][   ][   ][ X ][   ][   ][   ][   ][   ]
B6 [   ][   ][   ][ X ][ 25][   ][   ][   ][   ][   ]
B7 [   ][   ][ 25][   ][   ][   ][   ][   ][   ][   ]
B8 [   ][   ][   ][   ][   ][   ][   ][   ][   ][   ]
B9 [   ][   ][   ][   ][   ][   ][   ][   ][   ][   ]

Prod

  • This builds the final production grade version of the app
  • npm run prod to generate the optimized apps
    • Use npm run serve to start a web environment to play the game with all the features