webppl/src/dists/logNormal.js

Probabilistic programming for the web

'use strict';
var ad = require('../ad');
var base = require('./base');
var types = require('../types');
var util = require('../util');
var numeric = require('../math/numeric');
var gaussian = require('./gaussian');
var LOG_2PI = numeric.LOG_2PI;
function score(mu, sigma, x) {
    var logx = ad.scalar.log(x);
    return ad.scalar.sub(ad.scalar.sub(ad.scalar.neg(logx), ad.scalar.log(sigma)), ad.scalar.mul(0.5, ad.scalar.add(LOG_2PI, ad.scalar.div(ad.scalar.mul(ad.scalar.sub(logx, mu), ad.scalar.sub(logx, mu)), ad.scalar.mul(sigma, sigma)))));
}
function sample(mu, sigma) {
    var gaussian_sample = gaussian.sample(mu, sigma);
    return Math.exp(gaussian_sample);
}
var LogNormal = base.makeDistributionType({
    name: 'LogNormal',
    desc: 'Distribution over non-negative reals where the log of the random ' + 'variable is normally distributed.',
    params: [
        {
            name: 'mu',
            desc: 'mean',
            type: types.unboundedReal
        },
        {
            name: 'sigma',
            desc: 'standard deviation',
            type: types.positiveReal
        }
    ],
    wikipedia: 'Log-normal_distribution',
    mixins: [base.continuousSupport],
    sample: function () {
        return sample(ad.value(this.params.mu), ad.value(this.params.sigma));
    },
    score: function (x) {
        return score(this.params.mu, this.params.sigma, x);
    },
    support: function () {
        return {
            lower: 0,
            upper: Infinity
        };
    }
});
module.exports = {
    LogNormal: LogNormal,
    sample: sample,
    score: score
};