MULTILEVEL MONTE CARLO (MLMC) SIMULATIONS: PERFORMANCE RESULTS FOR SPE10 (XY SLICES)
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
In this report we first describe a generic multilevel Monte Carlo method and then illustrate its superior performance over a traditional single-level Monte Carlo method for second order elliptic PDEs corresponding to two-dimensional layers in (x, y)-direction of the Tenth SPE Comparative Solution project (SPE 10) which gives high-contrast permeability coefficients. The SPE10 data set is used as a coarse level in the Monte Carlo method and the respective permeability coefficient k (provided in the SPE10 dataset) is used as a mean in the simulation. The actual coefficients are drawn based on a KL-expansion assuming that the log-mean is perturbed by a log-normal distributed samples.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC52-07NA27344
- OSTI ID:
- 1240930
- Report Number(s):
- LLNL-TR-684244
- Country of Publication:
- United States
- Language:
- English
Similar Records
Multilevel Hierarchical Decomposition of Finite Element White Noise with Application to Multilevel Markov Chain Monte Carlo
Multilevel Sequential Monte Carlo Samplers for Normalizing Constants
Multilevel Sequential Monte Carlo with Dimension-Independent Likelihood-Informed Proposals
Journal Article
·
Tue Jun 08 00:00:00 EDT 2021
· SIAM Journal on Scientific Computing
·
OSTI ID:1240930
Multilevel Sequential Monte Carlo Samplers for Normalizing Constants
Journal Article
·
Thu Aug 24 00:00:00 EDT 2017
· ACM Transactions on Modeling and Computer Simulation
·
OSTI ID:1240930
+1 more
Multilevel Sequential Monte Carlo with Dimension-Independent Likelihood-Informed Proposals
Journal Article
·
Tue Mar 14 00:00:00 EDT 2017
· SIAM/ASA Journal on Uncertainty Quantification
·
OSTI ID:1240930
+2 more