In computability theory the S m
n  theorem, written also as "smn-theorem" or "s-m-n theorem" (also called the translation lemma, parameter theorem, and the parameterization theorem) is a basic result about programming languages (and, more generally, Gödel numberings of the computable functions) (Soare 1987, Rogers 1967). It was first proved by Stephen Cole Kleene (1943). The name S m
n  comes from the occurrence of an S with subscript n and superscript m in the original formulation of the theorem (see below).

In practical terms, the theorem says that for a given programming language and positive integers m and n, there exists a particular algorithm that accepts as input the source code of a program with m + n free variables, together with m values. This algorithm generates source code that effectively substitutes the values for the first m free variables, leaving the rest of the variables free.

The smn-theorem states that given a function of two arguments ${\displaystyle g(x,y)}$ which is computable, there exists a total and computable function such that ${\displaystyle \phi _{s}(x)(y)=g(x,y)}$ basically "fixing" the first argument of ${\displaystyle g}$. It's like partially applying an argument to a function. This is generalized over ${\displaystyle m,n}$ tuples for ${\displaystyle x,y}$. In other words,it addresses the idea of "parametrization" or "indexing" of computable functions. It's like creating a simplified version of a function that takes an additional parameter (index) to mimic the behavior of a more complex function.

The function ${\displaystyle s_{m}^{n))$ is designed to mimic the behavior of ${\displaystyle \phi (x,y)}$ when given the appropriate parameters. Essentially, by selecting the right values for ${\displaystyle m}$ and ${\displaystyle n}$, you can make ${\displaystyle s}$ behave like for a specific computation. Instead of dealing with the complexity of ${\displaystyle \phi (x,y)}$, we can work with a simpler ${\displaystyle s_{m}^{n))$ that captures the essence of the computation.

Details

The basic form of the theorem applies to functions of two arguments (Nies 2009, p. 6). Given a Gödel numbering ${\displaystyle \varphi }$ of recursive functions, there is a primitive recursive function s of two arguments with the following property: for every Gödel number p of a partial computable function f with two arguments, the expressions ${\displaystyle \varphi _{s(p,x)}(y)}$ and ${\displaystyle f(x,y)}$ are defined for the same combinations of natural numbers x and y, and their values are equal for any such combination. In other words, the following extensional equality of functions holds for every x:

${\displaystyle \varphi _{s(p,x)}\simeq \lambda y.\varphi _{p}(x,y).}$

More generally, for any m, n > 0, there exists a primitive recursive function ${\displaystyle S_{n}^{m))$ of m + 1 arguments that behaves as follows: for every Gödel number p of a partial computable function with m + n arguments, and all values of x₁, …, x_m:

${\displaystyle \varphi _{S_{n}^{m}(p,x_{1},\dots ,x_{m})}\simeq \lambda y_{1},\dots ,y_{n}.\varphi _{p}(x_{1},\dots ,x_{m},y_{1},\dots ,y_{n}).}$

The function s described above can be taken to be ${\displaystyle S_{1}^{1))$.

Formal statement

Given arities m and n, for every Turing Machine ${\displaystyle {\text{TM))_{x))$ of arity ${\displaystyle m+n}$ and for all possible values of inputs ${\displaystyle y_{1},\dots ,y_{m))$, there exists a Turing machine ${\displaystyle {\text{TM))_{k))$ of arity n, such that

${\displaystyle \forall z_{1},\dots ,z_{n}:{\text{TM))_{x}(y_{1},\dots ,y_{m},z_{1},\dots ,z_{n})={\text{TM))_{k}(z_{1},\dots ,z_{n}).}$

Furthermore, there is a Turing machine S that allows k to be calculated from x and y; it is denoted ${\displaystyle k=S_{n}^{m}(x,y_{1},\dots ,y_{m})}$.

Informally, S finds the Turing Machine ${\displaystyle {\text{TM))_{k))$ that is the result of hardcoding the values of y into ${\displaystyle {\text{TM))_{x))$. The result generalizes to any Turing-complete computing model.

This can also be extended to total computable functions as follows:

Given a total computable function ${\displaystyle s_{m,n}:\mathbb {N} ^{m+1}\rightarrow \mathbb {N} }$ and ${\displaystyle m,n\geq 1}$ such that ${\displaystyle \forall {\vec {x))\in \mathbb {N} ^{m},\forall {\vec {y))\in \mathbb {N} ^{n))$, ${\displaystyle \forall e\in \mathbb {N} }$:

${\displaystyle \phi _{e}^{m+n}({\vec {x)),{\vec {y)))=\phi _{s_{m,n}{(e,{\vec {x)))))^{n}({\vec {y)))}$

There is also a simplified version of the same theorem (defined infact as "simplified smn-theorem", which basically uses a total computable function as index as follows:

Let ${\displaystyle f:\mathbb {N} ^{n+m}\rightarrow \mathbb {N} }$ be a computable function. There, there is a total computable function ${\displaystyle s:\mathbb {N} ^{m}\rightarrow \mathbb {N} }$ such that ${\displaystyle \forall x\in \mathbb {N} ^{m))$, ${\displaystyle {\vec {y))\in \mathbb {N} ^{n))$:

${\displaystyle f({\vec {x)),{\vec {y)))=\phi _{S({\vec {x)))}^{(n)}({\vec {y)))}$

Example

The following Lisp code implements s₁₁ for Lisp.

(defun s11 (f x)
  (let ((y (gensym)))
    (list 'lambda (list y) (list f x y))))

For example, (s11 '(lambda (x y) (+ x y)) 3) evaluates to (lambda (g42) ((lambda (x y) (+ x y)) 3 g42)).