Using keyword arguments, default arguments, with and without Unitful

General Usage
1

Hi, I have a complicated function which has enough arguments that I want to use keyword args, and I want to have some default values, and in the early stages I like to use Unitful to help me catch dimensional errors, but at other times I would like to be able to just use SI throughout and drop all the clutter of the units.

I can do that in the following way:

f_(a,b) = a+b # imagine a bit more complexity here!
f(a,b=1.0) = f_(a,b)
f(a::Quantity,b::Quantity = 1.0u"m") = f_(a,b)

and then f(2.0e-3) and f(2.0u"mm") both do what I would want.

I feel like there should be a way for multiple dispatch to solve this for me, but in my attempts I’ve only figured out new ways to generate a stack overflow.

[EDIT: simplified MWE]

2

you can use Length to make sure the quantities will be able to add up:

import Unitful: Length
f(a::Length, b::Length = 1.0m) = f_(a,b)

you can get rid of the units with ustrip (don’t forget to give the reference unit)

julia> ustrip(2mm)
2
julia> ustrip(m, 2mm)
1//500
julia> ustrip(Float64, m, 2mm)
0.002
3

After noodling some more and being reminded that Julia’s multiple dispatch doesn’t allow multiple keyword-only methods of the same function, I think I’ll do the following for now:

f(a,b) = a+b # imagine a bit more complexity here!
f(;a=1.0,b=1.0) = f(a,b)
uf(;a::Quantity = 1.0u"m", b::Quantity = 1.0u"m") = f(a,b)

Edit: On the other hand, keyword args don’t play well with dot fusion which tips it towards my initial solution.

4

Keywors arguments don’t participate in Dispatch at all, that’s why only one set per function works.

1 Like