Massachusetts Institute of Technology (MIT) researchers have created an algorithm to produce random numbers with the best mix of speed, accuracy, and low memory currently available for certain tasks.
The Fast Loaded Dice Roller (FLDR) algorithm simulates the roll of dice to generate random integers.
The dice can have any number of sides, and are loaded to make some sides more likely to come up than others, constraining randomness to meet a preset probability distribution.
FLDR ensures perfectly loaded dice that exactly achieve the specified probabilities.
MIT's Vikash Mansinghka thinks FLDR could boost the efficiency of Monte Carlo simulation and inference techniques.
From MIT News
View Full Article
Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA