Awesome

@sabaki/deadstones

Simple Monte Carlo algorithm to determine dead stones on a Go board.

Installation

Use npm to install:

$ npm install @sabaki/deadstones

To use this module, require it as follows:

const deadstones = require('@sabaki/deadstones')

This module supports fetching the WASM file via fetch on the web if no node environment is found. Use a bundler like webpack and call the following method right after import or require:

deadstones.useFetch('./path/to/deadstones_bg.wasm')

Building

Make sure you have the Rust toolchain installed via rustup. This project uses the native WASM target which you can acquire with:

$ rustup target add wasm32-unknown-unknown

Make sure you have wasm-pack, Node.js 8 or higher, and npm installed. Clone this repository and install its dependencies with npm:

$ git clone https://github.com/SabakiHQ/deadstones
$ cd deadstones
$ npm install

To build WASM binaries and to start tests, use the following commands:

$ npm run build
$ npm test

API

Board Data

The board arrangement is represented by an array of arrays. Each of those subarrays represent one row, all containing the same number of integers. -1 denotes a white stone, 1 a black stone, and 0 represents an empty vertex

Example

[[ 0,  0,  1,  0, -1, -1,  1,  0, 0],
 [ 1,  0,  1, -1, -1,  1,  1,  1, 0],
 [ 0,  0,  1, -1,  0,  1, -1, -1, 0],
 [ 1,  1,  1, -1, -1, -1,  1, -1, 0],
 [ 1, -1,  1,  1, -1,  1,  1,  1, 0],
 [-1, -1, -1, -1, -1,  1,  0,  0, 0],
 [ 0, -1, -1,  0, -1,  1,  1,  1, 1],
 [ 0,  0,  0,  0,  0, -1, -1, -1, 1],
 [ 0,  0,  0,  0,  0,  0,  0, -1, 0]]

Vertex

Board positions are represented by an array of the form [x, y] where x and y are non-negative integers, zero-based coordinates of the vertex. [0, 0] denotes the top left position of the board.

This module exposes four functions:

async deadstones.guess(data[, options])

• data - Board data
• options <Object> (optional)

  • finished <boolean> (optional) - Default: false

    If set true, deadstones will assume that player areas have been completely surrounded, yielding better results.

  • iterations <integer> (optional) - Default: 100

    The number of random playthroughs to make.

Returns an array of vertices that Sabaki thinks are dead.

async deadstones.getProbabilityMap(data, iterations)

• data - Board data
• iterations <integer> - The number of random playthroughs to make.

Returns an array of arrays of the same size as data. Each entry is a number between -1 and 1 and corresponds to a vertex. A number closer to -1 is more likely controlled by white and a number closer to 1 is more likely controlled by black.

async deadstones.playTillEnd(data, sign)

• data - Board data
• sign -1 | 1 - White player corresponds to -1, black player is represented by 1.

Makes random alternating moves, starting with the player determined by sign, until only eye filling moves can be made. Then all eyes that are left will be filled with the corresponding color. This final board arrangement data will be returned.

async deadstones.getFloatingStones(data)

• data - Board data

A fast function that returns an array of vertices of stones that are inside enemy territory and do not surround more than one point of territory themselves.