| Author |
Rounding BigDecimal Calculation Yielded Wrong Result
|
James Gordon
Ranch Hand
Joined: Aug 09, 2002
Posts: 106
|
|
Hello All, The first println yielded 4.15 The second println yielded 4.980 Doing it manually, the result should be indeed 4.98 Does this imply that we should not use BigDecimal for calculations? Thanks in advance. [ September 22, 2004: Message edited by: James Gordon ]
|
 |
Joyce Lee
Ranch Hand
Joined: Jul 11, 2003
Posts: 1392
|
|
Hi James, I think it's still possible to use BigDecimal for the calculation if the scale value is set higher. Then round off the final value to 2 or 3 decimal places. Joyce  [ September 22, 2004: Message edited by: Joyce Lee ]
|
 |
David Hibbs
Ranch Hand
Joined: Dec 19, 2002
Posts: 374
|
|
Keep in mind that you have yourself specified the number of significant digits to carry. BigDecimal has another divide method that does not require a scale:
public BigDecimal divide( BigDecimal val, int roundingMode)Returns a BigDecimal whose value is (this / val), and whose scale is this.scale(). If rounding must be performed to generate a result with the given scale, the specified rounding mode is applied.
I have yet to see this method fail in standard calculations with a known-to-be-precise answer. HOWEVER... *You can always cover yourself with a NumberFormat that only allows 2 decimal places. *BigDecimal also provides many rounding modes to choose from, *You can always modify operator order to clarify what the result should be *Your operands define the inital scale of the result Play around with the following code some. Note that I don't specify a scale. If you pass "415" without the ".0" you'll get 5 as a result -- one of the operands had a zero scale. Anyway, good luck and have fun.
|
"Write beautiful code; then profile that beautiful code and make little bits of it uglier but faster." --The JavaPerformanceTuning.com team, Newsletter 039.
|
 |
 |
|
|
subject: Rounding BigDecimal Calculation Yielded Wrong Result
|
|
|