greataptic/dist/vector.js

A simplistic neural network library.

function _logit(x) {
  return Math.log(x / (1 - x));
}

let Vector = module.exports.Vector = class Vector {
  constructor(data) {
    if (data.data != null) data = data.data;
    this.data = Array.from(data);
    this.dims = this.data.length;
  }

  co(coord) {
    return this.data[coord] != null ? this.data[coord] : null;
  }

  size() {
    return Math.sqrt(this.data.reduce((a, b) => a + Math.pow(b, 2), 0));
  }

  sum() {
    return this.data.reduce((a, b) => a + b, 0);
  }

  add(_b) {
    let b = _b.dims != null ? _b : new Vector(_b);
    let res = [];

    for (let i = 0; i < Math.min(this.dims, b.dims); i++) res[i] = this.data[i] + b.data[i];

    return new Vector(res);
  }

  pow(pow) {
    return new Vector(this.data.map(x => Math.pow(x, pow)));
  }

  sub(_b) {
    let b = _b.dims != null ? _b : new Vector(_b);
    let res = [];

    for (let i = 0; i < Math.min(this.dims, b.dims); i++) res[i] = this.data[i] - b.data[i];

    return new Vector(res);
  }

  multiplyVec(_b) {
    let b = _b.dims != null ? _b : new Vector(_b);
    let res = [];

    for (let i = 0; i < Math.min(this.dims, b.dims); i++) res[i] = this.data[i] * b.data[i];

    return new Vector(res);
  }

  divideVec(_b) {
    let b = _b.dims != null ? _b : new Vector(_b);
    let res = [];

    for (let i = 0; i < Math.min(this.dims, b.dims); i++) res[i] = this.data[i] / b.data[i];

    return new Vector(res);
  }

  multiplyFac(b) {
    return new Vector(this.data.map(a => a * b));
  }

  divideFac(b) {
    return new Vector(this.data.map(a => a / b));
  }

  combiner(combinerFunc) {
    return function (b) {
      if (!b.dims) b = new Vector(b);
      return new Vector(this.data.map((x, i) => combinerFunc(x, b.data[i])));
    };
  }

  map(func) {
    return new Vector(this.data.map(func));
  }

  dot(b) {
    if (this.data.length === 0 || b.dims === 0) return 0;
    let res = 0;

    for (let i = 0; i < Math.min(this.dims, b.dims); i++) res += this.data[i] * b.data[i];

    return res;
  }

  static is(vec) {
    return !!(vec && vec.data && vec.dims && vec.size && vec.combiner && vec.map);
  }

  static fill(opts) {
    if (opts.map != null) return new Vector(new Array(opts.length).fill(0).map((_, i, a) => opts.map(i, a)));else return new Vector(new Array(opts.length).fill(opts.value != null ? opts.value : 0));
  }

  static random(dims) {
    return new Vector(new Array(dims).fill(0).map(() => _logit(0.25 + 0.5 * Math.random())));
  }

  static randomOne() {
    return _logit(0.25 + 0.5 * Math.random());
  }

  static randomSize(dims, size) {
    return new Vector(new Array(dims).fill(0).map(() => (Math.random() * 2 - 1) * size));
  }

  static zero(dims) {
    return new Vector(new Array(dims).fill(0));
  }

};
//# sourceMappingURL=vector.js.map