Smart detectors for Monte Carlo radiative transfer

TL;DR

This paper introduces smart detectors that utilize full impact location information of photon packages in Monte Carlo radiative transfer simulations, significantly reducing noise and improving efficiency with minimal computational overhead.

Contribution

The paper proposes a novel class of smart detectors that leverage impact location data to reduce noise in surface brightness estimates, enhancing Monte Carlo simulation efficiency.

Findings

Smart detectors reduce noise by about 10%.

Implementation is straightforward with negligible extra cost.

Simulation run time decreases by approximately 20%.

Abstract

Many optimization techniques have been invented to reduce the noise that is inherent in Monte Carlo radiative transfer simulations. As the typical detectors used in Monte Carlo simulations do not take into account all the information contained in the impacting photon packages, there is still room to optimize this detection process and the corresponding estimate of the surface brightness distributions. We want to investigate how all the information contained in the distribution of impacting photon packages can be optimally used to decrease the noise in the surface brightness distributions and hence to increase the efficiency of Monte Carlo radiative transfer simulations. We demonstrate that the estimate of the surface brightness distribution in a Monte Carlo radiative transfer simulation is similar to the estimate of the density distribution in an SPH simulation. Based on this similarity, a recipe is constructed for smart detectors that take full advantage of the exact location of the impact of the photon packages. Several types of smart detectors, each corresponding to a different smoothing kernel, are presented. We show that smart detectors, while preserving the same effective resolution, reduce the noise in the surface brightness distributions compared to the classical detectors. The most efficient smart detector realizes a noise reduction of about 10%, which corresponds to a reduction of the required number of photon packages (i.e. a reduction of the simulation run time) of 20%. As the practical implementation of the smart detectors is straightforward and the additional computational cost is completely negligible, we recommend the use of smart detectors in Monte Carlo radiative transfer simulations.