Monte Carlo Convergence Analysis for Anisotropic Sampling Power Spectra

1Dartmouth College

Tech. Report, Dartmouth College, Computer Science, 2016

Teaser
Illustration of (a) an expected anisotropic Fourier power spectrum for 2D n-rook (N = 256) point samples (left). The accompanying radial power spectra correspond to the radial average (top), radial behavior along the horizontal arrow (middle), and radial behavior along the diagonal arrow (bottom) of the 2D n-rooks power spectrum. Similarly, on the right, (b) we show the expected Fourier power spectrum for 3D parallel line samples (generated horizontally, orthogonal to the plane containing the power) using isotropic Poisson disk samples (N = 4096), with radial power spectra corresponding to radial averaging (top) within the 3D domain (power scaled for visualization) and radial behavior along two directions (middle and bottom) within the subspace containing power.

Abstract

Traditional Monte Carlo (MC) integration methods use point samples to numerically approximate the underlying integral. This approximation introduces variance in the integrated result, and this error can depend critically on the sampling patterns used during integration. Most of the well known samplers used for MC integration in graphics, e.g. jitter, Latin hypercube (n-rooks), multi-jitter, are anisotropic in nature. However, there are currently no tools available to analyze the impact of such anisotropic samplers on the variance convergence behavior of Monte Carlo integration. In this work, we propose a mathematical tool in the Fourier domain that allows analyzing the variance, and subsequently the convergence rate, of Monte Carlo integration using any arbitrary (anisotropic) sampling power spectrum. We apply our analysis to common anisotropic point sampling strategies in Monte Carlo integration, and extend our analysis to recent Monte Carlo approaches relying on line samples which have inherently anisotropic power spectra. We validate our theoretical results with several experiments using both point and line samples.

Downloads

Text Reference

Gurprit Singh, Wojciech Jarosz. Monte Carlo Convergence Analysis for Anisotropic Sampling Power Spectra. Tech. Report, Dartmouth College, Computer Science, August 2016.

BibTex Reference

@techreport{singh16monte,
    author = "Singh, Gurprit and Jarosz, Wojciech",
    title = "Monte Carlo Convergence Analysis for Anisotropic Sampling Power Spectra",
    institution = "Dartmouth College, Computer Science",
    address = "Hanover, NH",
    number = "TR2016-816",
    year = "2016",
    month = "August"
}