There & Never Back Again

Simulation software is used in a multitude of industry and academic fields, in an assortment of scopes. One might be interested in gravitation of celestial bodies, or field structures of molecular interactions. The scale to which these simulations can grow demands an equally scalable computational mechanism. Traditional and even accelerated solvers suffer from a lack of general scalability, depending on multiple input aspects and recursive refinement. Monte Carlo solvers offer a highly scalable computational mechanism, as it is not prone to the curse of dimensionality and the error can be driven down simply by taking more samples. In the course of refactoring such a Monte Carlo simulation software artefact, several anomalies were noted in its implementation structure. Through attempts at remediating the undesirable behaviour, a general problem of susceptibility was discovered for the Walk-on-Subdomains family of algorithms.