I want to generate a multdimensional dataset for multidimensional OLAP
experiments. The database is complied with zipfian distribution. But I'm not
clear on how to explane zipfian distribution over multidimensional space ?
Any comments are appreciated. Thank you in advance.

Jack Dull

In article <c81g7b$tuv$1 (AT) mail (DOT) cn99.com>,
"jack_dull" <jack_dull (AT) 163 (DOT) com> wrote:

Sorry, haven't got a clue, just wanted to know whether "explane"
was a pun on "multidimensional space."

--
Gerry Myerson (gerry (AT) maths (DOT) mq.edi.ai) (i -> u for email)

hi, Gerry Myerson.
Thanks for your replay. I'm Sorry. It is a typing mistake. It should be
'explain'.

The zipfian's defination is over one dimensional space originally. For
multdimensional space, no explanation I can find. The following is my
question:
1. Multidimensional space can be linearized. The linearized space is
formed by "a combination of distinct values of all dimension". The
probability for points' existances is complied to zipfian distribution. Am I
all right ?
2. Once the probability for points in linear space is determined. The
uniform distribution over [0-1] span can be generated by the computer. Then
it can be mapped to zipfian distribution. But while generating the database,
some points maybe occur more than one times. Although it will be a small
probability events. How to deal with this situation ?

jack dull

"Gerry Myerson" <gerry (AT) maths (DOT) mq.edi.ai.i2u4email> дÈëÓʼþ
news:gerry-B76631.14512114052004 (AT) sunb (DOT) ocs.mq.edu.au...