|
|
|
|
|
|
 |
 |
|
Mathematica Numerics: Controlling the Effects of Numerical Errors in Computation
|
|
|
|
|
|
| Organization: | Wolfram Research, Inc. |
|
|
|
|
|
|
1992 Mathematica Developer Conference
|
|
|
|
|
|
Boston, MA
|
|
|
|
|
|
A common misconception regarding errors in numerical computation is that the error in the final result is the simple accumulation of the errors made in the intermediate results. It is certainly true that if you reduce the errors in the intermediate results, the error in the final result will also be reduced. By using high-precision arithmetic you can generally make the error in the final result arbitrarily small, but such an approach to error reduction can be very much more expensive than is necessary. Reprint from the Mathematica Conference, June 1992, Boston. 29 pages.
|
|
|
|
|
|
|
|
|
|
|
|
documentation, tutorial, numerical methods, machine precision, machine arithmetic, machine accuracy, condition numbers, condition of an algorithm, heuristics, inexact arithmetic, big number arithmetic, SetPrecision, Set Accuracy, data structures, postscript, Numerics.ps
|
|
|
|
|
|
| Numerics.ps (146.8 KB) - Postscript file |
|
|
|
|
|
|
|
| | | | | |
|
For the newest resources, visit Wolfram Repositories and Archives »
