On Completeness of Pseudosimple Sets
Vadim Bulitko
(Odessa State University, Ukraine
ul. Perekopskoj divizii 67-40
Odessa 270062, UKRAINE
bulitko.odessa@REX.iasnet.com)
Abstract:
The paper contains completeness criterions for pseudosimple sets. Those
criterions are constructed using effectivization of the definitions as well
as extensionally bounded functions. Keywords:
Completeness, Pseudosimple set, Effectivization, Extensionally bounded function Category:
G
1 On Completeness of Pseudosimple Sets
A recursively enumerable (r.e.) set is called complete (Turing-complete) if
any r.e. set can be reduced to that set using Turing reducibility.
Optionally complete sets can have some interesting properties, for instance
they can be simple, pseudosimple, etc. Problems of constructing completeness
criterions for simple sets by an effectivization of definitions have been
investigated by McLaughlin, Smullyan, Lachlan, Arslanov, etc. History and
systematical description of that can be found in [Arslanov 87]. Simple sets fall into class of r.e. set classification by
Uspenskij and Dekker-Myhill [Rogers 67]. For creative sets the
criterions have been constructed even before. However we do not know any
specific criterion for pseudosimple and pseudocreative sets.
In this paper we propose some completeness criterions for
pseudosimple sets. Most of our notation follows [Rogers 67] also r. means recursive, r.e.
means recursively enumerable and i.r.e. means
infinite recursively enumerable. We use and
as ' under definition'. Sometimes we use 'number of' as
'index of'. Let us recall the following definitions from [Rogers 67]: Definition 1. A is pseudosimple (1) A is r.e. but not r.;
(2)
. Such C we will denote . Definition 2. A is pseudosimple with center C (1)
A is pseudosimple;
(2) Above definitions allow the following effectivization:
Definition 1'. A is weakly effectively
pseudosimple (1) A is pseudosimple;
(2) Definition 2'. A is weakly effectively
pseudosimple with center C (1) A is
pseudosimple with center C; (2)
First prove two complementary lemmas. Lemma 1. For any complete set A and any r.e. set M the following
function is total and recursive in A: Page 151
Proof. because
is an r.e. set. Therefore f is computable with oracle A. Lemma 2. Let A be complete and g be partial recursive in A. Then
the following function is partial recursive in A: Proof. Let us make computation of as follows. Given z we
start computation of g(z) and if it ends, we enumerate numbers of r.e.
sets
until some of them fall into which
is recursive in A. Let it happen when i=n so we set equal to .
Note we use uniform
recursiveness of in respect to complete A. Theorem 1. (A is pseudosimple and complete) (A
is weakly effectively pseudosimple and ). Proof. Let A be pseudosimple and complete then
obviously . Define total f as follows: f is computable with oracle A by lemma 1,2, and because is an
r.e. set, and is immune. Thus A is weakly
effectively pseudosimple. Using f, which is recursive in A, and possibility to
write out elements of in increasing order with
oracle A, construct initial segment of enumeration of
in increasing order. An r.e. index of X(z) can be computed with oracle A
because X(z) is finite and recursive in A. Let g(z) be that r.e.
index. So . Thus A
is complete by the Arslanov theorem [Arslanov 87]. (The Arslanov theorem used
here and below states that for any r.e.s. A (A is complete)
. Another completeness criterion can be obtained introducing concept of
special function. Definition 3. Let A be weakly effectively pseudosimple. We call
total and recursive in A function f special for A if Theorem 2. (A is pseudosimple and complete) (A
is weakly effectively pseudosimple and there is f special for A). Proof. Define f as follows: f can be computed with oracle A by lemmas 1,2, and because
is an r.e. set, and is immune. So A is weakly
effectively pseudosimple with special function f. Let g be defined as follows: Page 152
g can be computed with oracle A because f is recursive in A. Also g
is total and obviously does not have fixed points so we can apply the
Arslanov theorem that causes A is complete. There is also another approach to effectivization of the definitions. The
approach is based on use of functions evaluating complexity characteristics
of set other than the number of elements. For instance, Lachlan in [Lachlan 68]
used the number of value changes of characteristic function for a given
r.e. set on a cohesive set building sufficient conditions for complete
maximal sets. Completeness criterion for maximal sets [Bulitko 92] constructed
using Kolmogorov complexity of set initial segments is another example. Next
step in that direction is use of extensionally bounded functions to
construct completeness criterions for special classes of simple sets. That
approach has been posed by V.K.Bulitko in [Bulitko 95]. In present paper we use
that approach to describe pseudosimple and complete sets. Definition 4. Total f is extensionally bounded (e.b.)
Theorem 3. (A is pseudosimple and complete) (A
is r.e. but not r. & such that
the following function f:
is e.b. and recursive in A). Proof. Obviously A is r.e. but not r. f is
computable with oracle A by lemma 1 and because is immune.
For any number i of r.e. f(i) is bounded by cardinality of X that is
finite because A is pseudosimple. f is equal to 0 for a number of an
r.e. set which is not a subset of . Therefore f
is e.b. Assume .
Then taking in
account infiniteness of index set we get f is not bounded on
index set. That contradicts with the definition of e.b. function. Therefore
our assumption is wrong and A is pseudosimple. Completeness of A can be
proved in the same way as in theorem 2. Let us consider pseudosimple sets with center. Theorem 4. (A is pseudosimple with recursive center C and
complete) (A is weakly effectively pseudosimple with
recursive center C). Proof. Define f as follows: f is computable with oracle A by lemmas 1,2 and .
Therefore A is weakly effectively pseudosimple with recursive
center. We can get total function that does not have
fixed points doing in the same way as in the first theorem. Using the
Arslanov theorem we obtain completeness of A. Page 153
Situation regarding case of non-recursive (in general) center is described
by the following theorems and remarkable enough. Trying to get the criterion
we succeeded in way of using e.b functions only. Theorem 5. (A is weakly effectively pseudosimple with center C
recursive in A) (A is pseudosimple with center C and
complete). Proof. Follows from the first theorem. Definition 5. Let A be weakly effectively pseudosimple with center
C. Call total and computable with oracle A function f bounding for A
if Theorem 6. (A is weakly effectively pseudosimple with center C
and there is bounding for A function f) (A is
pseudosimple with center C and complete). Proof. Define g as follows: g can be computed with oracle A because f is recursive in A. Also g
is total and does not have fixed points. Thus A is complete by the
Arslanov theorem. Theorem 7. (A is pseudosimple with center C and complete)
(A is r.e. but not r.; ;
the following f is e.b. and computable with oracle A): Proof. Obviously A is r.e. but not r. f is
computable with oracle A by lemma 1 and because having oracle A we are
able to write out all of elements not belonging to C and situated in
f is e.b. because f is bounded on numbers of r.e.
which is finite
because A is pseudosimple with center, and f is equal to 0 on numbers of
an r.e. set which is not a subset of . A is pseudosimple with center C because f is e.b. and
therefore
(otherwise we would get unbounded f
taking in account that any r.e. set has an infinite number of its
indexes). Completeness of A can be proved in the same way as in theorem 6.
References
[Arslanov 87] Arslanov, M.M.: 'Local theory of unsolvability
degrees and sets'; KGU press / Kazan, (1987).(In Russian).
[Rogers 67] Rogers, H. Jr.: 'Theory of recursive functions and
effective computability'; McGrow-Hill / N.-Y. (1967).
[Lachlan 68] A.H.Lachlan, A. H.: 'Complete recursively
enumerable sets'; Proc. Amer. Math. Soc., 19, (1968), 99-102.
[Bulitko 92] Bulitko, V. K.: 'Subturing reducibilities with
bounded complexity'; Izv. VUZov, ser. mat., 1 (1992), 27-37. (In Russian).
[Bulitko 95] Bulitko, V. K.: 'On some complexity
characteristics of Immune sets'; MLQ, 41, 3 (1995) (in press). Page 154
|