Weihrauch reducibility explained
In computable analysis, Weihrauch reducibility is a notion of reducibility between multi-valued functions on represented spaces that roughly captures the uniform computational strength of computational problems. It was originally introduced by Klaus Weihrauch in an unpublished 1992 technical report.[1]
Definition
A represented space is a pair of a set
and a
surjective partial function
.
Let
and
be represented spaces and let
f:\subsetX\rightrightarrowsY
be a partial multi-valued function. A
realizer for
is a (possibly partial) function
such that, for every
,
\deltaY\circF(p)=f\circ\deltaX(p)
. Intuitively, a realizer
for
behaves "just like
" but it works on names. If
is a realizer for
we write
.
Let
be represented spaces and let
f:\subsetX\rightrightarrowsY,g:\subsetZ\rightrightarrowsW
be partial multi-valued functions. We say that
is
Weihrauch reducible to
, and write
, if there are
computable partial functions
such that
where
\Psi\langleid,G\Phi\rangle:=\langlep,q\rangle\mapsto\Psi(\langlep,G\Phi(q)\rangle)
and
denotes the join in the
Baire space. Very often, in the literature we find
written as a binary function, so to avoid the use of the join. In other words,
if there are two computable maps
such that the function
is a realizer for
whenever
is a solution for
. The maps
are often called
forward and
backward functional respectively.
We say that
is
strongly Weihrauch reducible to
, and write
, if the backward functional
does not have access to the original input. In symbols:
Notes and References
- Weihrauch . Klaus . The Degrees of Discontinuity of some Translators between Representations of the Real Numbers . Informatik-Berichte . 129 . FernUniversität in Hagen . 1992 .