Next: About this document
Up: Random WalksMarkov Chains
Previous: Markov chains
- When used with care, Monte Carlo methods can be used to solve problems
that are otherwise intractible.
- The choice of the random number generator must be carefully considered.
- The built in generator in your compiler library should not be used for
serious applications
- There is no universal, single best generator.
- For some problems quasi-random numbers provide more accurate results
than pseudo-random numbers.
- For the estimation of definite integrals in high dimension (above about 5),
a Monte Carlo method using Quasi-random is the best approach to use.
- Markov chains can be used for solving some very difficult problems.
Skip Carter
Tue Jan 9 12:38:54 PST 1996