simulation@uflorida.cis.ufl.edu (Moderator: Paul Fishwick) (12/06/88)
Volume: 6, Issue: 9, Tue Dec 6 09:26:45 EST 1988 +----------------+ | TODAY'S TOPICS | +----------------+ (1) Protein Modeling at JvNC (2) Random Numbers * Moderator: Paul Fishwick, Univ. of Florida * Send topical mail to: simulation@uflorida.cis.ufl.edu * Archives available via FTP to bikini.cis.ufl.edu, login as 'anonymous', use your last name as the password, change directory to pub/simdigest. ----------------------------------------------------------------------------- Date: Thu, 01 Dec 88 15:16:31 EST From: David Salzman <SALZMAN@pucc.princeton.edu> Subject: Protein Modeling at JvNC, Dec. 5 (Please Post To: comp-simulation@rutgers.edu Richard Feldmann National Institutes of Health The Role of Intuition in The Modeling of Proteins Monday, December 5, 1988 4:00 PM (Refreshments will be served at 3:40 PM) Physical intuition about molecular models can be formalized as sophisticated computer models, but the converse is not necessarily true. Experience has shown that computer models with visual representations yield more physical insight than models with purely numerical output, but the exact form of the visual representation matters further. Feldmann will present protein models in a variety of media--stereoscopic, motion picture, tactile, etc.--as illustrations of the principle. Stereo glasses will be provided. Richard Feldmann has been involved with issues of presenting macromolecular structure since he joined the National Institutes of Health in 1967. He now works in the NIH's Division of Computer Research and Technology. He received his degrees from the University of Dayton and the Polytechnic Institute of New York. Talks are free and open to the public. The John von Neumann National Supercomputer Center (JvNC) is supported by the National Science Foundation and the New Jersey Commission on Science and Technology. Talks are held at 665 College Road East, Princeton Forrestal Center, Plainsboro NJ. For further information, call (609)520-2000. ------------------------------ Date: 30 Nov 88 14:21:02 GMT (Wednesday) To: simulation%uflorida.cis.ufl.edu@NSS.CS.UCL.AC.UK From: Peter Dickman <pwd%computer-lab.cambridge.ac.uk@NSS.CS.UCL.AC.UK> Subject: Random Numbers George Bier points to the recent CACM article by Stephen Park & Keith Miller on 'multiplicative linear congruence' pseudo-random number generators and asks whether anyone knows of any theory which allows for the fact that these algorithms generate numbers without replacement. In Knuth Vol. 2, 2nd Edition (pp 26 -- 28) an additive generator devised by Mitchell & Moore is described. It is efficient (no multiplies, it requires an addition MOD (2^n) & two pointer decrements plus possibly a resetting of a pointer if it hits zero) and has a period of (2^f)(2^55 - 1) for some f: 0 <= f < n. Knuth says that the theory behind the generator is not well-known though & proofs of good randomness are lacking. The generator does require a little more state than the MLC approaches though and is still working without replacement (if one views all 55 words of state plus the two pointers as all combining together to describe the current random number - when their values are all precisely repeated the sequence will start again). Anyway, while people are thinking about the lack of replacement, does anyone know if the randomness of this generator has been investigated? In particular how does it compare with the 'minimal standard generator' proposed in the CACM article (Vol 31 No 10 pp 1192 -- 1201)? Disclaimer: I'm not a theorist, I just want a good pseudo-random number supply. Thanks, Peter ARPA: pwd@cl.cam.ac.uk UUCP: {Atlantic Ocean}!mcvax!ukc!cam-cl!pwd ------------------------------ +--------------------------+ | END OF SIMULATION DIGEST | +--------------------------+