- Home
- Non Sequential Ray Tracing
- Sources, Splitting and Scattering
- Understanding Sobol Sampling
Understanding Sobol Sampling
- By Mark Nicholson
- Published 10 November 2005
- Sources, Splitting and Scattering
-
Rating:
Summary and References
Summary
Sobol sampling can made a significant reduction in the time taken to undertake a Monte Carlo ray-tracing simulation, and will generally produce faster convergence than truly random rays.
External References
Sobol sampling can made a significant reduction in the time taken to undertake a Monte Carlo ray-tracing simulation, and will generally produce faster convergence than truly random rays.
External References
1. Press, Flannery, Teukolsky, and Vetterling, Numerical Recipes, Cambridge Press.
Attachments
4 Responses to "Understanding Sobol Sampling" 
|
said this on 14 Nov 2005 5:01:02 AM PDT
What about the acuracy in the general case of this sobol method Is there some cases where sobol distribution yields to false results
|
Author)|
said this on 14 Nov 2005 9:04:32 AM PDT
Hi The results will never be "false" as such but with anything to do with Monte-Carlo ray-tracing you need to trace "enough" rays to have confidence in the results. Its difficult to generalize about when you have "enough" rays! But switching between random and Sobol sampling should not change the results significantly: the benefit of using Sobol is probably more in the early stages of a design where it lets you get a better signal/noise with fewer rays. As your design matures you should always increase the number of rays traced to be sure you are not just seeing a statistical artefact.
I'll amend the article to make this point more clearly and then remove your questiona nd this reply from the comment list in a week or so.
|
|
said this on 17 Nov 2005 7:14:10 PM PDT
I did a search on Sobol Method Sobol sampling in SPIE OSA electronic databases and did not advantages/disadvantages of Sobol vs. Monte Carlo. This article filled that gap for a simple assumed uniform case.
|
|
said this on 15 Mar 2007 11:22:52 AM PDT
Good information.
|