choose function for large values

Question

As part of a formula I am computing, I need to compute large b inomial coefficients

This is something like :

choose(1598,999)

but the result from R is just inf. Is there some way to get around that? Or some way to approximate the result in any way? It is just that I am trying to work with a formula which has sample size as n (1598) and another parameter as p (in this casw 999).

So I want to compute

with n->16000
b->1000
c()->w
c()->w1
    for (k in 2:(n-1)){
      k*(k-1)*chooseZ((n-k),(b-1))->w
    c(w1,w)->w1
    }

While there is no error the result is wrong:

sum(w1)
Error in sum(w1) : invalid 'type' (raw) of argument
    > head(w1)
    [1] 01 00 00 00 01 00

Answer 1

You could use the print function with the digits specified. Don't know if it'll work for this though.

The syntax would be

print(choose(1598,999),digits = 20)

I think print caps it at 22 digits. Check if this works

Answer 2

Try using chooseZ from the gmp package.

library(gmp)
chooseZ(1598,999)
Big Integer ('bigz') :
[1] 24278696426445719206411636658492530369104863587241559663553520984424347733955902209013400793410551248995223418598838685432844582570031038690080807851906250009888240115104285225539059084804496423089545062262737596184912679476923113553162874564650288993564978319447910715500162623621305510917593005744632186322124862050089421200736125546497281717999352152426164243210315176391085915104025111403152439452509792706140328536251214887318149223708091336739500080000

Note that the result is a bigz class object and larger than the numeric class can represent.