Abstract

A good pseudo random number generator (PRNG) should have a sufficiently long period, fast generating speed, high-dimensional uniformity, and stellar empirical performance. In addition to these requirements, the strong demand of parallel processing in today's large-scale computing environment requires a mechanism for a PRNG to generate suitable streams of pseudo random numbers for multiple processors in parallel. Among well-known PRNGs, maximum-period multiple recursive generators (MRGs) with many nonzero terms would be such a great PRNG if only it could be implemented efficiently and in parallel. In this talk, we will show how to automatically generate numerous maximum-period MRGs with many nonzero terms, and then implement them efficiently and in parallel. (Joint work with Jyh-Jen Horng Shiau,  Horng-Shing Lu and Bryan Winter)

Anyone who is interesting in relative topics are welcome to join. People who are far away from the NCTS are also welcome to join the forum via Skype.

Please contact:  Ms. Ejan Chen(02)33668816  ejanchen@ncts.ntu.edu.twfor participation via Skype.