AFICS, print precision only affect floating number and have no effect on integer
or extended integer.
*/1+i.10050 will overflow to double, but */1+i.10050x will not
max precision of ieee double is between 15 and 16 digits.
10^.2^52
15.65356
Arie Groeneveld wrote:
It's because print precision is to 'low' .
9!:11[6
ts 'it=:|.&.":@(5{.":&(|.&.":))@*/1+i.10050'
|ill-formed number: ts
| it=: |.&.":@(5{.":&(|.&.":))@*/1+i.10050
Setting print precision to
9!:11 [10
gives
ts 'it=:|.&.":@(5{.":&(|.&.":))@*/1+i.10050'
0.141714 135040
it
69696
=@@i
Geoff Canyon schreef:
I'm trying to find the last five non-zero digits of a large factorial.
So for:
!10x = 3628800
the answer would be 36288 while for:
!20x = 2432902008176640000
the answer would be 17664.
I'm trying to do this by calculating /* 1+i.1000 without calculating
large numbers by losing the trailing zeroes at each step and the
leading digits more than five. Here's what I have so far:
(((|.&.":)@(5{.":&(|.&.":)))@*)/ 1+i.20x
The truncating code:
-- Converts to string, transposes, and converts back to a number.
That loses trailing zeroes (because they're now leading zeroes).
-- Converts to string and grabs the first five characters (which were
the last five non-zero digits).
-- Converts to string again -- not sure why this is necessary, but it
doesn't work otherwise -- transposes, and converts back to a number.
This whole thing is performed atop *, and the resulting verb is
inserted into the list from 1 to a large number.
This seems to work for large numbers up to 10049x. At 10050x I get an
ill-formed number error.
Where did I go wrong?
regards,
Geoff
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
