Thread View: sci.math.stat
1 messages
1 total messages
Started by Keith Decker
Fri, 02 Oct 1992 15:37
Combinatorics of `limited partitions'?
Author: Keith Decker
Date: Fri, 02 Oct 1992 15:37
Date: Fri, 02 Oct 1992 15:37
46 lines
2551 bytes
2551 bytes
Hello again, Since my last query generated some great help, I thought I'd post the other little problem that has come up in recent work. I actually have, I think, a set of recurrence equations to solve this, but I wanted to know if it is a solved problem so I can double check my answer, and if it has a `standard' name by which I can refer to it. I've been calling it the problem of `limited partitions'. It is known that the number of partitions of n into exactly k parts is [n-1 choose k-1]. What I want to know is the partition of n into k OR LESS parts WHERE THE MAX VALUE = m. I call this (until someone tells me it's real name) a `limited partition' P(n,k,m). I've solved this by breaking it into simpler problems, i.e., we first choose some number i (i>1, and n<im) of the k slots to get the max value m (there are [k-(i-1) choose 1] ways to do this), and then for each compute P'(n-im,k-i,m-1), where P'(n,k,m) = number of partitions of n into k or less values each <= m. P'(n,k,m) is in turn computed by choosing another j<k and referring to a final partition, P*(n,k,m), which is the partition of n into EXACTLY k values where each value is <= m. In any case, I wondered if any of these were standard solved problems so I could call them by the right names and check my derivations (maybe they even have closed form solutions). Thanks again, /Keith decker@cs.umass.edu Cooperative Distributed Problem Solving Lab University of Massachusetts at Amherst P.S. In case anyone wants to know, the application of the result is to distributed problem solving. Say we want to analyze how long it will take A agents to solve n problems that initially arrive randomly at agents (like tossing n balls into A boxes) The termination time for the system is based on which agent gets the *most* problems (which box gets the most balls). The expected termination time is then the termination time if the agent with the most balls got exactly m balls, times the probability of that occuring, which can be deduced directly from the above limited partition results.
Thread Navigation
This is a paginated view of messages in the thread with full content displayed inline.
Messages are displayed in chronological order, with the original post highlighted in green.
Use pagination controls to navigate through all messages in large threads.
Back to All Threads