Random number generator ordering by extreme event probability calculations

Abstract

This collection contains aggregated data used to investigate the effects of various random number generators on the quality and performance of extreme event simulations. The analysis was carried out using Quantum Dice's quantum random number generator (QRNG), the Mersenne Twister pseudo-random number generator (PRNG) and multiple linear congruential generators (LCG) in a sequential use case. All archives contain Snakemake workflows documenting the generation and evaluation process as well as simulation parameters. Jupyter notebooks within the archives demonstrate how to analyse the respective data. Archives referencing a mass contain workflows and aggregated data for a system of masses loosely coupled by harmonic springs. The data is used to analyse the probability of extreme sagging of the system below a predefined tolerance, resulting in a failure condition. In addition the archive contains the sample notebook used to evaluate the data. The archive referencing an SDE contains the collection of probabilities of a non-linear stochastic differential equation to evolve from a fixed initial position to an extremal configuration. The data consists of multiple workflows, each consisting of 100 repetitions of the SDE simulation for varying combinations of the threshold parameter and the number of trajectories simulated. A notebook used to compare the RNGs using the absolute approximation error is provided with the dataset.