µΩ - Just a little bit of resistance ...

A prototype of a SI unit system implementation using the new Numpy extensible data type feature that aims to solve the cross-library incompatibility problem that other unit system implementations, like unyt, suffer.

At a high level, the implementation consists of a Quantity scalar object and a QuantityDType Numpy DTypeMeta. The scalar object contains all of the unit system logic and scalar arithmetic. The dtype object allows the native Numpy ndarray to represent floating point values in the array as Quantity objects.

Quantity converts the input quantity to the equivalent base SI units. This is how Mathcad works.

>>> from microohm import Quantity
>>> v = Quantity(1, "mV")
>>> v
Quantity('0.001 V')
>>> v.display(".1f mV")
1.0 mV
>>> i = Quantity("1 kA")
>>> i
Quantity('1000.0 A')
>>> r = v/i
>>> r
Quantity('1e-06 Ω')
>>> r.display(".1f µΩ")
1.0 µΩ

Use QuantityDType to instantiate an array of floating point values with a given unit.

>>> from microohm import QuantityDType
>>> import numpy as np
>>> v_arr = np.array([1.0, 2.0], dtype=QuantityDType("mV"))
>>> i_arr = np.array([1.0, 2.0], dtype=QuantityDType("kA"))
>>> r_arr = v_arr/i_arr
>>> r_arr
array([Quantity('1e-06 Ω'), Quantity('1e-06 Ω')],
      dtype=QuantityDType('mV/(kA)'))
>>> r_arr[0].display(".1f µΩ")
1.0 µΩ

There is support for non-SI system units as defined in NIST Special Publication 811.

>>> d1 = Quantity(1, "mil")
>>> d2 = Quantity(1, "mm")
>>> d3 = d1 + d2
>>> d3
Quantity('0.0010254 m')
>>> d3.display(".6f mil")
40.370079 mil

This approach solves the cross-library incompatibility problem.

>>> import pandas as pd
>>> arr = np.array([1, 2], dtype=QuantityDType("mV/Hz**(1/2)"))
>>> s1 = pd.Series(arr)
>>> s1.iloc[0].display(".1f mV/Hz**(1/2)")
1.0 mV/Hz**(1/2)
>>> s2 = 2 * s1
>>> s2.iloc[1].display(".1f mV/Hz**(1/2)")
2.0 mV/Hz**(1/2)

Developer workflow

The QuantityDType has to be implemented using the Numpy C-API and compiled as a C-extension module. The following flow works on Ubuntu 22.04; other platforms may work, but haven't been tested.

$ git clone [email protected]:l-johnston/microohm.git
$ cd microohm
$ python -m venv .venv
$ source .venv/bin/activate
(.venv) $ python -m pip install -U pip
(.venv) $ pip install -U setuptools
(.venv) $ pip install -e .[dev]