Fast and Exact Simulation of Long-Memory Processes
Peter Craigmile and Tilmann Gneiting
| We develop software for the
fast and exact simulation of long-memory processes, such as fractional
Gaussian noise, fractional Brownian motion, fractionally differenced,
power-law, and Cauchy processes. Random functions with long-range
dependence are useful in characterizing dependence in time series (for
example, global temperature data, or pollutant concentrations) as well
as in modeling heteorogeneities in spatial data (for example, porosity
and hydraulic conductivity in aquifers). The code will be publicly
available at the NRCSE website. |
|
|