Core Numerical

While the Num is practical, it's also not very well defined or generic enough. For example, it's heavily skewed towards types forming a Field, which are also typically signed.

The design used here is to try to break single piece of functionality into their own classe, so that you can overloaded addition operation on your types which doesn't requires an implementation for subtraction or negation that is fundamentally impossible to write.

We also strive to provide something that is enjoyable to use for computer programs purpose instead of exposing an extremely precise mathematical abstraction that is only useful for corner cases.






This is a simple type class to wrap the + operation.

It has the following member 'azero', '+' and 'scale' where:

  • azero is the identity element of the group over '+'
  • '+' is the group addition
  • 'scale' is the repeated addition of n times.

It has the following properties:

azero + azero == azero
azero + x     == x
x     + azero == x
scale 0 x == azero
scale 1 x == x

This can be effectively be seen as a Monoid where mempty is azero, and (+) is mappend.


This is a class to wrap the - operation.

However the key difference with the - available in base, is that while the operand need to be of same type, the result is not constraint to be of the same type as the operands.

The result is an associated type called Difference which is determined by the operand.

(-) :: a -> a -> Difference a

This doesn't cover all the possible scenario available, but the added flexibility allow new interesting use cases.


This is a class to wrap the multiplication * operation.

It has the following properties:

midentity * midentity == midentity
midentity * x == x
x * midentity == x
x ^ 0 == midentity
x ^ 1 == x