Discussion Topics
- Tests of Serial and Parallel RNGs
- Are the current mix of random number tests adequate for the
needs of the Monte Carlo and the random number generation communities?
What other sorts of tests should be made widely available? Should
these two communities work together in developing a suite of tests
based on applications to create a tool that will (one hopes) predict
undesirable interactions between generators and real applications?
- Applications
- What are the applications requirements that have not been
addressed in the current generation of pseudorandom number generation
tools for high performance machines? What sorts of future requirements
need to be considered in the further development of pseudorandom
number generation tools for high performance machines?
- Quasi-random Numbers
- What are the quasirandom number needs of the Monte Carlo
community? Can quasirandom numbers be implemented with enough support
to allow their often subtle application be used effectively by
relative novices? What are the quasirandom number generation
techniques that should be implemented for the Monte Carlo community?
Are there fundamental differences between providing parallel
quasirandom numbers and providing parallel pseudorandom numbers?
- Theoretical RNG issues
- What recent developments in pseudorandom number generation stand to benefit the the
Monte Carlo community the most? What are some requirements of the Monte Carlo
community that need to be addressed in the search for new pseudorandom number
generators?
- RNG Libraries
- What are the roadblocks that vendors face in adopting the
parallel random number generation tools presented in this workshop
into their own scientific software? What are vendor requirements that
have not been addressed in the parallel random number generation tools
presented in this workshop?
ashoks@ncsa.uiuc.edu
Last modified:2 Mar 1998