unitty

Unitty (pronounced 'unity') provides units when you need them, not when you don't. It's based on a phiolosophy that:

• Units are useful for inputs
• Units are useful for outputs (e.g. reports, plots)
• Calculations are done in SI units, so units are not needed unless checking.
• Unittests are used to encompass dimensional checks. Dimensional checks themselves are insufficient.
• Quantities are named adequately in the code (e.g. 'area', 'length', 'mass'). The kind of quantity they represent is clear from their names, and so representing it elsewhere is typically redundant.

Unitty provides a flexible, lightweight package to:

• Automatically switch outputs to a different unit system using a single command, while keeping meaningful units (e.g. for force, pressure, etc.). This capability was a core reason to write unitty.
• Apply units intuitively.
• Convert numbers to and from different units.
• Fully customise units and unit systems. The user needs full flexibility of the systems and units used.
• Bundle values with units into quantities for downstream use (e.g. reports, plots).
• Not lock the user into objects with units attached. By default, units are not attached to calculation objects - the results are simply floats or arrays. This behaviour can be changed automatically to do dimensional checks during unittesting.
• Use meaningful units (e.g. for force, pressure, etc.) while also allowing dimensionality reduction and checking. Automatic dimensionality reduction in some other packages can be frustrating.

It is built on the idea that units provide scale factors to convert numbers to and from values in SI base units. Since SI base units all have magnitudes of 1.0, unit information need not be in the calculations themselves.

Quick start

Load the default units and unit systems:

import unitty
u = unitty.get_units()

Getting units

unitty supports attribute and string methods to get units.

u.kg
# 1 kg

u['kg']

Strings can be complex, like this:

u['kg/(s2.m)']

There are a few rules to formatting the strings:

• Use only one divide symbol ('/')
• Use only one pair of brackets in the dividend, if needed
• Use a period ('.') to signify multiplication
• Do not include an exponent symbol ('^'). For example, for square meters, write 'm2'.

Basic calculations

Use the units to get an input into a base unit system:

v = 5 * u.ft
v
# 1.524

We get 1.524, which is in the defalut base unit for length, which is meters. We can multiply and divide units in sensible ways:

v2 = q = 5 * u.lbs / u.ft2
v2
# 24.4121

This gives 24.4121, which is in kg/m^2 (we'll omit power symbol and write this as 'kg/m2').

Making quantities

Now we'll create quantities like this:

q = 5 << u.lbs / u.ft2
q
# 24.4121 kg/m2

# Alternatively:
q = 5 << u['lbs/ft2']

val, s = q.in_sys()
val
# 24.4121
s
# 'kg/m2'

q.str_in_sys()
# 24.4121 kg/m2

Switching unit systems

The Quantity q displays as '24.4121 kg/m2', since it's a Quantity that includes unit information. Now, let's change our unit system to another preloaded one and look at it again:

unitty.set_system('US')
q
# 5 lbs/ft2

val, s = q.in_sys()
val
# 4.999999999999999
s
# 'lbs/ft2'

q.str_in_sys()
# 5 lbs/ft2

Now, q shows as '5 lbs/ft2'.

We can use the in_sys method to get the value and units in this new unit system. But it's still the same underlying value, which we can see via:

val, s = q.in_base()
val
# 24.4121
s
# 'kg/m2'

q.str_in_base()
# 24.4121 kg/m2

We can make a new Quantity while in this unit system:

q2 = 7 << u.lbs / u.ft2
q2
# 7 lbs/ft2

Importantly, the value of q2 is still in base units:

q2.str_in_base()
# 34.177 kg/m2
q2.value
34.177

We can switch back to metric (the default unit system), and take a look at out quantities again:

unitty.set_system('metric')
q
# 24.4121 kg/m2
q2
0.34177 kg/cm2

Named quantity types

Notice that in the above, q2 displays in different units. That's because by default, it guesses the best available combination of units in the unit system to display in a friendly way. Often, though, there are particular units we want to display in, which depend on the unit system we want to us. For one-off cases, we can do this:

val, s = q2.in_units(u['kg/m2'])
val
# 34.17699345
s
# 'kg/m2'

If we have many such quantities, we can do this automatically. We can define some named quantity types in a csv file, like this one that we'll call `example.csv':

ref	metric	US
widget_length	mm	in
complex.value	kg.s2/m	lbs.s2/ft

Then we apply it like this:

s = unitty.get_systems() # The object that looks after different unit systems
s.set_refs('example.csv')

Now we can name the quantity types like this:

q2.set_ref('complex.value')

A shorthand way is to add the 'ref' (quantity reference) when getting the unit:

q2 = 7 << u['lbs/ft2', 'complex.value']

Now, the display of the unit automaticallymatches the units we've specified.

unitty.set_system('US')
q2
# 7 lbs/ft2

unitty.set_system('metric')
q2
# 34.17699345 kg/(m2)

val, s = q.in_sys()
val
# 34.17699345
s
# kg/(m2)

Alternatives

Several other packages might be better suited to your particular needs. Here are some to consider, along with some notes. It is believed that none of these other packages allow to automatically switch outputs between unit systems with a single command.

• numericalunits: Units are values. Simple.
• astropy.units: Great, but locks units into calculation values - can't get back to simple floats or arrays.
• sympy.physics.units: Solid.
• pint: A very powerful units package.
• unyt: An excellent and capable package.
• quantiphy: Seems a bit awkward to use.
• Buckingham: A bit awkward to use.
• DimPy: Very old.
• Magnitude: Clunky to use.
• Python-quantities: A good package.
• physipy: Another good package
• SciMath Units: Large range of units.
• cf_units: Suggested replacement of old udunitspy package. Clunky.
• Units
• Unum
• quantities
• physical-quantities
• parampy
• pynbody
• misu
• pysics