A bipartite graph is an ordered triple G=(X,Y,E)such that X and Y are sets,
,
and
.
The vertices of G are the elements of
.
The edges of G are the elements of E.
A covering of G is a set
such that
every edge of G has a vertex in C,
i.e. we have
for all
.
A matching in G is
a pairwise disjoint set
.
Here pairwise disjointness
means that no two edges in M have a common vertex,
i.e. we have
for all
such that
.
For any set S we use |S| to denote the cardinality of S,
i.e. the number of elements in S.
If G is any bipartite graph and C is any covering of Gand M is any matching in G, then clearly
.
The König duality theorem [7] asserts that,
for any finite bipartite graph G,
there exist a covering C of Gand a matching M in G such that |C|=|M|.
In other words,
Clearly if (C,M) is a König covering of G then |C|=|M|. König [7] showed that every finite bipartite graph has a König covering. From this the König duality theorem follows immediately.
Podewski, Steffens and Aharoni extended the König duality theorem to infinite bipartite graphs. In order to make such extensions meaningful, they considered König coverings. Podewski and Steffens [12] showed that every countably infinite bipartite graph has a König covering. Aharoni [1] showed that every uncountable bipartite graph has a König covering. We refer to the Podewski-Steffens theorem (respectively Aharoni's theorem) as the König duality theorem for countable (respectively uncountable) bipartite graphs.
Aharoni, Magidor and Shore [2] considered the following logical or foundational question: Which set existence axioms are needed to prove the König duality theorem for countable bipartite graphs? Aharoni, Magidor and Shore obtained a partial answer to this question, but they did not answer it completely. The purpose of this paper is to finish the work which was begun by Aharoni, Magidor and Shore.
The general question of which set existence axioms are needed
to prove specific mathematical theorems
is of basic importance for the foundations of mathematics.
This general question has been studied fruitfully
in the context of subsystems of second order arithmetic.
For this purpose,
five of the most important subsystems of second order arithmetic
are
,
,
,
,
and
-
.
It is known that these five systems
are of strictly increasing strength as regards their
ability to prove mathematical theorems.
Moreover,
for many particular mathematical theorems, it turns out that
one can determine the weakest natural
subsystem of second order arithmetic
in which the given mathematical theorem is provable.
Such results are established
by showing that the given mathematical theorem
is logically equivalent to the axioms of the specified subsystem of
second order arithmetic, the equivalence being proved in
a weaker system.
Consider for example the Bolzano-Weierstrass theorem:
every bounded sequence of real numbers has a convergent subsequence.
It is known that the weakest subsystem of second order arithmetic
in which the Bolzano-Weierstrass theorem is provable is
.
This is established by showing that
the Bolzano-Weierstrass theorem is logically equivalent
to the axioms of
,
the equivalence being proved in the weaker system
.
For a survey of subsystems of second order arithmetic and their role in foundational studies, see my article [16]. A fuller treatment will appear in [17]. For additional results and open problems concerning logical and foundational aspects of combinatorics, see the articles in Logic and Combinatorics [8], especially [3].
Aharoni, Magidor and Shore [2] made a major contribution
to the foundational program of [16].
They obtained two important results.
First, the König duality theorem for countable bipartite graphs (i.e. CKDT)
is provable in
-
.
Second, CKDT logically implies the axioms of
,
this implication being provable in the weak system
.
(Aharoni, Magidor and Shore also obtained results concerning logical aspects of
some other infinitistic variants of the König duality theorem.)
The main result of the present paper is that
the König duality theorem for countable bipartite graphs is provable in
.
This is established in Section 3 below.
Combining this with the results of Aharoni, Magidor and Shore, we see that
CKDT is logically equivalent to the axioms of
,
the equivalence being provable in
.
Thus
is the weakest natural
subsystem of second order arithmetic
in which CKDT is provable.