- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Russell Standish <R.Standish.domain.name.hidden>

Date: Tue, 16 Oct 2001 09:57:11 +1000 (EST)

To try and settle this debate on uniform measures, the best definition

of measure I could find was at

http://www.probability.net/. Unfortunately, this site is rather

difficult to get into. However, a measure is a function m defined over

the subsets of the set O in question (eg O=Z in the case of integers). It

has the following two properties:

m(\empty) = 0

for all A\subsetO, A_n\subsetO such that A=\union_n=1^\infty A_n and A_i

\intersect A_j =\empty \forall i,j => m(A)=\sum_n=1^\infty A_n.

(called the countably additive property). In less formal terms, it

means you get the same number for your measure, no matter how you

slice the set into disjoint subsets.

Furthermore, if m(A)\in[0,\infty], the measure is called a positive

measure.

It can be readily seen that the cardinality function over the powerset

of the integers satisfies these properties, and hence is a

measure. Furthermore, it is correctly a uniform measure, as each set

element contributes equal weight to the set's measure.

I still think the problem has to do with confusing measure with

probability distribution, which must additionally be normalisable (ie

m(A)\in[0,1]). There is clearly no uniform probability distribution

over the integers, or any set that is not compact for that matter.

Cheers

----------------------------------------------------------------------------

Dr. Russell Standish Director

High Performance Computing Support Unit, Phone 9385 6967, 8308 3119 (mobile)

UNSW SYDNEY 2052 Fax 9385 6965, 0425 253119 (")

Australia R.Standish.domain.name.hidden

Room 2075, Red Centre http://parallel.hpc.unsw.edu.au/rks

International prefix +612, Interstate prefix 02

----------------------------------------------------------------------------

Received on Mon Oct 15 2001 - 17:11:58 PDT

Date: Tue, 16 Oct 2001 09:57:11 +1000 (EST)

To try and settle this debate on uniform measures, the best definition

of measure I could find was at

http://www.probability.net/. Unfortunately, this site is rather

difficult to get into. However, a measure is a function m defined over

the subsets of the set O in question (eg O=Z in the case of integers). It

has the following two properties:

m(\empty) = 0

for all A\subsetO, A_n\subsetO such that A=\union_n=1^\infty A_n and A_i

\intersect A_j =\empty \forall i,j => m(A)=\sum_n=1^\infty A_n.

(called the countably additive property). In less formal terms, it

means you get the same number for your measure, no matter how you

slice the set into disjoint subsets.

Furthermore, if m(A)\in[0,\infty], the measure is called a positive

measure.

It can be readily seen that the cardinality function over the powerset

of the integers satisfies these properties, and hence is a

measure. Furthermore, it is correctly a uniform measure, as each set

element contributes equal weight to the set's measure.

I still think the problem has to do with confusing measure with

probability distribution, which must additionally be normalisable (ie

m(A)\in[0,1]). There is clearly no uniform probability distribution

over the integers, or any set that is not compact for that matter.

Cheers

----------------------------------------------------------------------------

Dr. Russell Standish Director

High Performance Computing Support Unit, Phone 9385 6967, 8308 3119 (mobile)

UNSW SYDNEY 2052 Fax 9385 6965, 0425 253119 (")

Australia R.Standish.domain.name.hidden

Room 2075, Red Centre http://parallel.hpc.unsw.edu.au/rks

International prefix +612, Interstate prefix 02

----------------------------------------------------------------------------

Received on Mon Oct 15 2001 - 17:11:58 PDT

*
This archive was generated by hypermail 2.3.0
: Fri Feb 16 2018 - 13:20:07 PST
*