Re: exercise in double number arithmetic

On 08/07/2024 13:30, Krishna Myneni wrote:

On 7/7/24 09:33, Gerry Jackson wrote:

On 07/07/2024 14:17, Krishna Myneni wrote:

...

The whole D>F and floating point operations can be dispensed with if we have D/MOD -- a double length version /MOD.
>
-- Krishna
>
>

>
There is a +D/MOD in this link
>
http://www3.cs.stonybrook.edu/~algorith/implement/random-number/distrib/r250.seq
>

 Thanks, Gerry. It looks like D/MOD is not essential to computing the binomial coefficients with double length numbers. We can use M*/ for single computations, or simply D+ to build up a table of coefficients. But, it's good to have a reference for D/MOD.
 I remember that I wanted to update the FSL R250 code to generate 64-bit random numbers, but I never got around to it. There is also a permutations and combinations module in FSL which may benefit from what has been demonstrated in this thread.
 -- Krishna
 

I didn't realise the paper I linked to was in the FSL as I've never had much to do with the FSL. I found I had another definition for D/MOD from Forth Dimensions Vol 14 Issue 6, p 27 "Math - Who Needs It"
https://www.forth.org/fd/FD-V14N6.pdf
I've no idea how good it is
--
Gerry