This python package allows you to manipulate physical quantities, basically considering in the association of a value (scalar, numpy.ndarray and more) and a physical unit (like meter or joule).
>>> from physipy.quickstart import nm, hp, c, J
>>> E_ph = hp * c / (500 * nm)
>>> print(E_ph)
3.9728916483435158e-19 kg*m**2/s**2
>>> E_ph.favunit = J
>>> print(E_ph)
3.9728916483435158e-19 JFor a quickstart, check the quickstart notebook on the homepage
Get a live session at
pip install physipy
- Few LOC
- Simple architecture, with only 2 classes (namely Dimension and Quantity)
- High numpy compatibility
- Human-readable syntax (fast syntax !)
- Define scalar and arrays of physical quantities
- Compute operation between them : add, sub, mul, div, pow, and so on
- Display physical quantities in various “units”
The implementation is pretty simple :
- a Dimension object represents a physical dimension. For now, these dimension are based on the SI unit. It is basically a dictionary where the keys represent the base dimensions, and the values are the exponent these dimensions.
- a Quantity object is simply the association of a value, scalar or array (or more!), and a Dimension object. Note that this Quantity classe does not sub-class numpy.ndarray (although Quantity objects are compatible with numpy's ufuncs). Most of the work is done by this class.
- By default, a Quantity is displayed in term of SI untis. To express a Quantity in another unit, just set the "favunit", which stands for "favourite unit" of the Quantity :
my_toe_length.favunit = mm. - Plenty of common units (ex : Watt) and constants (ex : speed of light) are packed in. Your physical quantities (
my_toe_length), units (kg), and constants (kB) are all Quantity objects.
Numpy is almost fully and transparently handled in physipy : basic operations, indexing, numpy functions and universal functions are handled. There are more than 150 functions implemented ! Some limitation still exist but can be can be circumvented. See the dedicated notebook : https://github.com/mocquin/physipy/blob/master/docs/notebooks/Numpy.ipynb.
Matplotlib allows defining a physical units interface, which can be turned on using just setup_matplotlib, all plot involving a physical quantity will automatically label the axis accordingly :
import numpy as np
import matplotlib.pyplot as plt
from physipy import s, m, units, setup_matplotlib
setup_matplotlib() # make matplotlib physipy's units aware
mm = units["mm"] # get millimiter
ms = units["ms"] # get millisecond
y = np.linspace(0, 30) * mm
x = np.linspace(0, 5) * s
y.favunit = mm # no need to call ax.yaxis.set_units(mm)
x.favunit = ms # no need to call ax.xaxis.set_units(ms)
fig, ax = plt.subplots()
ax.plot(x, y)
Checkout the dedicated notebook on matplotlib support.
Some ipywidgets are provided to make your physical researches and results more interactive :
Checkout the dedicated notebook on ipywidgets.
See the dedicated notebook.
Benchmark results using asv are available at https://mocquin.github.io/physipy/ :
See also the corresponding notebook at : https://github.com/mocquin/physipy/blob/master/docs/notebooks/Benchmarking%20with%20AirSpeedVelocity.ipynb.
See : https://www.bipm.org/en/CGPM/db/20/8/. Astropy's base units : https://docs.astropy.org/en/stable/units/standard_units.html#enabling-other-units
A quick performance benchmark show that physipy is just as fast (or faster) than other well-known physical packages, both when computing scalars (int or float) and numpy arrays :
For a more in-depth comparison, checkout this repository (not maintenained be it should!) : https://github.com/mocquin/quantities-comparison :
There are plenty of python packages that handle physical quantities computation. Some of them are full packages while some are just plain python module. Here is a list of those I could find (approximately sorted by guessed-popularity) :
- astropy
- sympy
- pint
- forallpeople
- unyt
- python-measurement
- Unum
- scipp
- magnitude
- physics.py : there are actually several packages based on the same core code : ipython-physics (python 2 only) and python3-physics (python 3 only)
- ScientificPython.Scientific.Physics.PhysicalQuantities
- numericalunits
- dimensions.py (python 2 only)
- buckingham
- units
- quantities
- physical-quantities
- brian
- quantiphy
- parampy
- pynbody
- python-units
- natu
- misu
- and finally pysics from which this package was inspired
If you know another package that is not in this list yet, feel free to contribute ! Also, if you are interested in the subject of physical quantities packages in python, check this quantities-comparison repo and this talk. Also check this comparison table and this takl.
Some C/C++ alternatives :
This project is licensed under the MIT License - see the LICENSE.md file for details
Thumbs up to phicem and his pysics package, on which this package was highly inspired. Check it out !