Difference between revisions of "Least-number operator"
(Importing text file) |
(TeX) |
||
Line 1: | Line 1: | ||
− | '' | + | {{TEX|done}} |
+ | ''$\mu$-operator, minimization operator'' | ||
− | A device for constructing new functions out of others, as follows. Let | + | A device for constructing new functions out of others, as follows. Let $g$ be an $(n+1)$-ary arithmetic function, i.e. a function with arguments and values in the set of natural numbers. It is assumed here that $g$ is a partial function, i.e. it is not necessarily defined for all values of its arguments. One says that the $n$-ary arithmetic function $f$ is obtained from $g$ by the least-number operator if, for any natural numbers $k_1,\ldots,k_n,k$, |
− | + | $$f(k_1,\ldots,k_n)=k$$ | |
− | if and only if the values of | + | if and only if the values of $g(k_1,\ldots,k_n,l)$ are defined and are not equal to zero for all $l<k$, while $g(k_1,\ldots,k_n,k)$ is defined and equals zero. If $f$ is obtained from $g$ by the least-number operator, one writes |
− | + | $$f(x_1,\ldots,x_n)=\mu y[g(x_1,\ldots,x_n,y)=0].$$ | |
− | An important property of the least-number operator is the following: If | + | An important property of the least-number operator is the following: If $g$ is a [[Computable function|computable function]], then the above function $f$ is always partially computable. In fact, if there is an algorithm computing $g$, then the value of $f(x_1,\ldots,x_n)$ can be computed as follows. Compute $g(x_1,\ldots,x_n,0)$. If the computation process ends, i.e. if $g(x_1,\ldots,x_n,0)$ is defined, and if $g(x_1,\ldots,x_n,0)=0$, put $f(x_1,\ldots,x_n)=0$; but if $g(x_1,\ldots,x_n,0)\neq0$, begin to compute $g(x_1,\ldots,x_n,1)$. If the process ends and $g(x_1,\ldots,x_n,1)=0$, put $f(x_1,\ldots,x_n)=1$; but if $g(x_1,\ldots,x_n,1)\neq0$, proceed to compute $g(x_1,\ldots,x_n,2)$; etc. The computation will come to an end if there exists a $y$ such that, for all $z<y$, $g(x_1,\ldots,x_n,z)$ is defined and not zero, while $g(x_1,\ldots,x_n,y)$ is defined and equal to zero. Then $f(x_1,\ldots,x_n)=y$. |
The least-number operator plays an important role in the definition of the class of partial recursive functions (cf. [[Partial recursive function|Partial recursive function]]). | The least-number operator plays an important role in the definition of the class of partial recursive functions (cf. [[Partial recursive function|Partial recursive function]]). |
Latest revision as of 07:06, 12 August 2014
$\mu$-operator, minimization operator
A device for constructing new functions out of others, as follows. Let $g$ be an $(n+1)$-ary arithmetic function, i.e. a function with arguments and values in the set of natural numbers. It is assumed here that $g$ is a partial function, i.e. it is not necessarily defined for all values of its arguments. One says that the $n$-ary arithmetic function $f$ is obtained from $g$ by the least-number operator if, for any natural numbers $k_1,\ldots,k_n,k$,
$$f(k_1,\ldots,k_n)=k$$
if and only if the values of $g(k_1,\ldots,k_n,l)$ are defined and are not equal to zero for all $l<k$, while $g(k_1,\ldots,k_n,k)$ is defined and equals zero. If $f$ is obtained from $g$ by the least-number operator, one writes
$$f(x_1,\ldots,x_n)=\mu y[g(x_1,\ldots,x_n,y)=0].$$
An important property of the least-number operator is the following: If $g$ is a computable function, then the above function $f$ is always partially computable. In fact, if there is an algorithm computing $g$, then the value of $f(x_1,\ldots,x_n)$ can be computed as follows. Compute $g(x_1,\ldots,x_n,0)$. If the computation process ends, i.e. if $g(x_1,\ldots,x_n,0)$ is defined, and if $g(x_1,\ldots,x_n,0)=0$, put $f(x_1,\ldots,x_n)=0$; but if $g(x_1,\ldots,x_n,0)\neq0$, begin to compute $g(x_1,\ldots,x_n,1)$. If the process ends and $g(x_1,\ldots,x_n,1)=0$, put $f(x_1,\ldots,x_n)=1$; but if $g(x_1,\ldots,x_n,1)\neq0$, proceed to compute $g(x_1,\ldots,x_n,2)$; etc. The computation will come to an end if there exists a $y$ such that, for all $z<y$, $g(x_1,\ldots,x_n,z)$ is defined and not zero, while $g(x_1,\ldots,x_n,y)$ is defined and equal to zero. Then $f(x_1,\ldots,x_n)=y$.
The least-number operator plays an important role in the definition of the class of partial recursive functions (cf. Partial recursive function).
Comments
References
[a1] | H. Rogers jr., "Theory of recursive functions and effective computability" , McGraw-Hill (1967) pp. 164–165 |
Least-number operator. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Least-number_operator&oldid=32848