Conventions

SI units

All too often mismatched units are to blame for outright errors, or at least tedious extra work to get data into the correct format. Therefore, even if it goes against common usage patterns, units are assumed to be SI everywhere. This is also consistent with e.g. scipy.constants.

Time

Throughout pysmo you will notice extensive use of Python's datetime module. Points in time are always datetime.datetime objects with the tzinfo attribute set (ideally to UTC). Time calculations are straightforward with the datetime module, so things like relative times (e.g. for a begin time in a seismogram) are avoided. Similarly, time deltas (e.g. sampling interval) are always datetime.timedelta objects.

SciPy and NumPy parameters

In places where scipy or numpy functions are used, pysmo will adhere to the original parameter definitions (and ranges). For example, scipy.signal.windows.tukey takes a parameter alpha which ranges from 0 to 1 whereas similar "taper" functions in seismology often use an alpha that lies between 0 and 0.5 (whereby 0.5 does the same thing as 1 in the SciPy version).

Avoid None type

Pysmo types are meant to be specific and narrowly defined. Unless absolutely necessary, using the None type is avoided. Consider for example a Coordinates type that has a latitude and a longitude attribute. As coordinates always have a latitude and a longitude, it should not be possible to set either of those attributes to None.

Prefer functions over methods

It is generally better to use functions than methods, or at the very least the number of methods should be kept at a minimum. This is even more so when code re-usability becomes a factor. Hence pysmo types almost never contain methods.