rnc

[J P Morgan]

 I have R writing xml files of designs that include everything listed under 
"statistical properties" EXCEPT "pairwise_variance_dist" (which I will likely 
rename "pairwise_variances"). The holdup is that this is a distribution as 
envisaged in other parts of the rnc, using the element "dist_entry", but 
currently in the rnc "dist_entry" is commented as "pending."

1. There is no need to use dist_entry now if it is to be changed or removed. Can 
a decision be reached on this today? Alternatively I can just define another 
element for use in the current context.

2. If we stick with dist_entry, I need to be reminded of the meaning of one of 
its attributes. Specifically, I will generate v-choose-2 pairwise variances. 
These can be thought of as a lex-indexed vector, so attribute "format" is clear. 
Does attribute "index" take value v(v-1)/2?

3. Pairwise variances will be floating point numbers. This requires the last 
line of dist_entry be changed from (n|unknown) to (n|d|unknown). Any problem 
with this?

JP