###### Kofi Gumbs

# Lambda Calculus

Lambda Calculus is often cited as the first functional programming language.
Though there have been many attempts to derive a formal programming language
from Lambda Calculus, the original language in its purest form
**does not support** the following:

- named functions
- primitives
- objects
- functions without a return value
- functions with multiple parameters

Here is what the syntax does allow:

- Defining an anonymous function

`λx.x`

defines the identity function`λ`

marks the start of a function definition`.`

marks the start of the function body (right-associative)

- Applying a function

`xy`

means “apply`x`

to`y`

”

Some things to notice:

- Everything in Lambda Calculus is a function with one parameter and one return value
- The “multiple parameters problem” is solved by the fact that functions always return functions
`λx.λy.yx`

defines a function that takes a function`x`

, then returns a function, which takes a function`y`

, then returns`y`

applied to`x`

- Here’s how you would write that in Clojure:
`(fn [x] (fn [y] (y x)))`

- We use currying to
*express*multiple parameters

Strictly speaking, Lambda Calculus does not support any non-function values. However, in order to “evaluate” an expression, we need to define a base case. Therefore, we will define the following “integer” representations:

- 0:
`λs.λz.z`

function that ignores its parameter and returns the identity function - 1:
`λs.λz.sz`

function that applies its parameter once to the parameter of the function it returns - 2:
`λs.λz.ssz`

function that applies its parameter twice to the parameter of the function it returns - etc. These definitions only serve to help us “evaluate” expressions!

Let’s look at how we would define Clojure’s `inc`

function in Lamda Calculus:

```
λn.λs.λz.s(n s z) #=> parenthesis help us describe order of operations
λn.λs.λz.s(n s z) 3 #=> let's apply this function to the value 3
λs.λz.s(3 s z) #=> starting from the outside, we replace 3 for n
λs.λz.s((λa.λb.aaab) s z) #=> 3 as a function call can be replaced for its "function representation"
λs.λz.s((λb.sssb) z) #=> apply "3" to s...
λs.λz.s((λz.sssz)) #=> ...then to z
λs.λz.ssssz #=> at this point, we can extract the nested function
4 #=> we arrive at the "function" representation of 4!
```