2 3 live&learn&rejoice: stochastic

April 20, 2011


 . . . from the Greek στόχος for aim or guess, means random according to wikipedia.

....a synonym for random. in statistics, the only sample that can be truly representative of a population is a random sample per http://www.csa.com/discoveryguides/lang/gloss.php

and more from http://www.statslab.cam.ac.uk/~rrw1/stoc_sched/:

stochastic scheduling is concerned with scheduling problems in which the processing times of tasks are modelled as random variables. Thus a job's processing time is not known until it is complete. Scheduling may be preemptive or non-preemptive, occur on one or on many processors, and be concerned with various optimization criteria.

a typical result in this area is that if n jobs have processing times that are exponentially distributed with different means and are to be processed by m identical machines operating in parallel, then LEPT (longest expected processing time first) minimizes the expected makespan (the time at which all jobs are complete.)

and then comes the fun part:

consider the differences between stochastic programming and deterministic programming at
http://stoprog.org/index.html?spintroduction.html. note that stochastic programming "prevents decisions that anticipate future uncertain events."

No comments:

Post a Comment