Gregory Ewing <greg.ew...@canterbury.ac.nz> writes: > Pete Forman wrote: >> However I am coming from scientific measurements where 1.0 is the >> stored value for observations between 0.95 and 1.05. > > You only know that because you're keeping some extra information in > your head about what the 1.0 stored in your computer represents. It's > not inherent in the value itself.
No, that is a real case. Floating point hardware can hold intermediate values in 80 bit registers before writing the double precision result to 8 bytes of memory. There are compiler switches available to enforce strict IEEE conformance and discard the intermediate guard digits. By adhering to those rules the results are predictable (repeatable on other hardware) but less accurate mathematically speaking. -- Pete Forman -- https://mail.python.org/mailman/listinfo/python-list
