[Numpy-discussion] RFC: A proposal for implementing some date/time types in NumPy

Anne Archibald peridot.faceted@gmail....
Fri Jul 11 11:28:30 CDT 2008


2008/7/11 Pierre GM <pgmdevlist@gmail.com>:

>> A final note on time scales
>> ---------------------------
> Wow, indeed. In environmental sciences (my side) and in finances (Matt's), we
> very rarely have a need for that precision, thankfully...

We do, sometimes, in pulsar astronomy. But I think it's reasonable to
force us to use our own custom time representations. For example, I've
dealt with time series representing (X-ray) photon arrival times,
expressed in modified Julian days. The effect of representing
microsecond-scale quantities as differences between numbers on the
order of fifty thousand days I leave to your imagination. But 64 bits
was certainly not enough.

More, there are even more subtle effects relating to differences
between TAI and TT, issues relating to whether your times are measured
at the earth's barycenter or the solar system barycenter... A
date/time class that tries to do everything is quickly going to become
unfeasibly complicated.

Anne


More information about the Numpy-discussion mailing list