math/ndiff 2.00_5

ndiff is a utility for comparing putatively similar files, ignoring small numeric differences. The utility is written by Nelson H. F. Beebe and covered by the GNU General Public License (GPL), version 2. It may be built with arbitrary precision support (more powerful) or using built-in floating point precision, see Makefile.

Assessing the consistency of a numerical program run in multiple environments (operating systems, architectures, or compilers) can be a difficult task for a human, as small differences in numerical output values are expected. File differencing utilites, such as diff(1), will generally produce voluminous output, often longer than the original files.

ndiff solves this problem. Taking two text files expected to be identical, or at least numerically similar, it allows to specify absolute and/or relative error tolerances for differences between numerical values in the two files, and then reports only the lines with values exceeding those tolerances. It also tells by how much they differ. A simple example:

% ndiff --relative-error 1.0e-3 test019.txt.1 test019.txt.2 ### Maximum relative error in matching lines = 8.64e-51 at line 129 field 4