Forum Home
Press F1
 
Thread ID: 71205 2006-07-29 10:10:00 large number in excel bpt2 (6653) Press F1
Post ID Timestamp Content User
474570 2006-07-29 10:10:00 What's the biggest number that I can calculate in excel. When I tried 111,111,111^2 I got 12,345,678,987,654,300 instead of 12,345,678,987,654,321. bpt2 (6653)
474571 2006-07-29 10:48:00 bpt2

For what it's worth, I tried the same in Openoffice 2.0 under Xandros (Linux) and got the same answer as you did. It seems as if it is rounding the answer over those last three figures.

Jim
Hhel (8073)
474572 2006-07-29 22:30:00 In Excel Help type in Excel specifications and limits, then look under calculation specifications. The biggest number is 9.99999999999999E+307.

regards,
Graham
Parry (5696)
474573 2006-07-30 00:14:00 Hi Graham
I believe the original poster was referring to the accuracy of numbers in excess of 15 digits - most numbers greater than 10^15 are incorrect given the 16th and greater digits (counting from the left) are rounded to zero. That said, I don't think it is an Excel issue - I have seen the same thing with the Windows calculator. That's why I wrote a programme to correctly handle numbers > 10^15.
Cheers, Andrew
andrew93 (249)
474574 2006-07-30 07:36:00 {snip}
I believe the original poster was referring to the accuracy of numbers in excess of 15 digits

Ah I see, thanks Andrew. Its all to do with floating point precision and is a problem with computers in general it seems. For some dry reading see support.microsoft.com :D

regards,
Graham
Parry (5696)
474575 2006-07-30 09:21:00 Thanks for that bpt2 (6653)
474576 2006-07-31 05:52:00 Of course it's only a problem if you have a need for that precision. I challenge you to measure anything to that precision in the real world.

Floating point calculations are always a trap for people who trust computer output.

Calculate 10/3 and multiply the answer by 3. For extra fun, compare the final result with 10.000 . :D

There's a whole branch of mathematics devoted to "Numerical Methods". A Google search for "Ieee 754" will find lots about the major standard used for computing.

If you want "exact" values, there are "infinite precision" systems, but they are often slow. Matlab will do it.
Graham L (2)
1