Accounting Calculation formulas not adding correctly

Attached, please find a simple spreadsheet where if you carry the digits out 10 or more, the balance (which should be zero), isn't zero.  Has anyone seen this before?
Recon.xlsx
Who is Participating?

Commented:
0

Commented:
you may need to change the format of cells in some columns to use more/less decimal places.
0

Commented:
That's not really an Excel bug - it's a computer 'bug' generally. Round the numbers to whatever precision concerns you.
0

Author Commented:
Thanks, rowansmith.  We would have never noticed it until the decimals were very extended.  I've not run across this before.

ill:
For small formulas, the variance may be insignificant.  However, this is a company that forecasts forest data with millions of cell data.  The small error would make a significant difference.

Either of you have any idea how to fix this issue or if there is a patch out there?  I cannot find one.
0

Commented:
It is a bug and it isn't a bug.  It depends on your perspective.  If it's not doing what you expect then it's a bug.  For me I expect this and so it is not a bug :-)

The problem is unavoidable in the sense that there are decimal numbers which have no exact, finite representation in binary but they have to be stored with a limited number of digits.

The problem is that the binary representation of the decimal 0.1 is 0.0001100110011... where the 0011 repeats forever.  The floating point representation of 0.1 is inherently inaccurate.  In a more familiar setting, if we had a machine that could store a two decimal digit mantissa, then 1/3 would be stored as .33 and when added to itself two times would yield .99 rather than 1.0.  Displayed values are often rounded so that the last couple of digits are not displayed so that the underlying representation is again different from the displayed value. [REF: http://www.lawrence.edu/fast/jordanj/FloatingPoint.html]

To avoid seeing it, you need to use Excel's "ROUND" function to limit your numbers to some significant number of digits (say 10 or 12).  The other solution is to use something, like APL or J (see www.jsoftware.com) with the notion of comparison tolerance built in. [REF: http://www.lawrence.edu/fast/jordanj/FloatingPoint.html]

-Rowan
0

Author Commented:
The explaination is clear and a potential solution has been presented for this type of formula.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.