Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Save to My Library
Look up keyword
Like this
4Activity
0 of .
Results for:
No results containing your search query
P. 1
Kolmogorov-Loveland Randomness and Stochasticity

Kolmogorov-Loveland Randomness and Stochasticity

Ratings: (0)|Views: 273|Likes:
Published by Feflos

More info:

Published by: Feflos on Mar 13, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

07/25/2010

pdf

text

original

 
Kolmogorov-Loveland Randomness andStochasticity
Wolfgang Merkle – Institut f¨ur Informatik, Universit¨at HeidelbergJoseph S. Miller – Department of Mathematics, Indiana UniversityAndr´e Nies Department of Computer Science, University of AucklandJan Reimann – Institut f¨ur Informatik, Universit¨at HeidelbergFrank Stephan – Departments of Mathematics and of Computer Science,National University of SingaporeFebruary 18, 2006
Abstract
An infinite binary sequence
is Kolmogorov-Loveland (or KL)random if there is no computable non-monotonic betting strategy thatsucceeds on
in the sense of having an unbounded gain in the limitwhile betting successively on bits of 
. A sequence
is KL-stochasticif there is no computable non-monotonic selection rule that selectsfrom
an infinite, biased sequence.One of the major open problems in the field of effective randomnessis whether Martin-L¨of randomness is the same as KL-randomness.Our first main result states that KL-random sequences are close toMartin-L¨of random sequences in so far as every KL-random sequencehas arbitrarily dense subsequences that are Martin-L¨of random. Akey lemma in the proof of this result is that for every effective splitof a KL-random sequence at least one of the halves is Martin-L¨of random. However, this splitting property does not characterize KL-randomness; we construct a sequence that is not even computablyrandom such that every effective split yields two subsequences thatare 2-random. Furthermore, we show for any KL-random sequence
A
that is computable in the halting problem that, first, for any effectivesplit of 
A
both halves are Martin-L¨of random and, second, for anycomputable, nondecreasing, and unbounded function
g
and almostall
n
, the prefix of 
A
of length
n
has prefix-free Kolmogorov complexityat least
n
g
(
n
). Again, the latter property does not characterize KL-randomness, even when restricted to left-r.e. sequences; we construct
1
 
1 INTRODUCTION 
2
a left-r.e. sequence that has this property but is not KL-stochastic, infact, is not even Mises-Wald-Church stochastic.Turning our attention to KL-stochasticity, we construct a non-empty Π
01
class of KL-stochastic sequences that are not weakly 1-random; by the usual basis theorems we obtain such sequences thatin addition are left-r.e., are low, or are of hyperimmune-free degree.Our second main result asserts that every KL-stochastic sequencehas effective dimension 1, or equivalently, a sequence cannot be KL-stochastic if it has infinitely many prefixes that can be compressed bya factor of 
α <
1. This improves on a result by Muchnik, who hasshown that were they to exist, such compressible prefixes could notbe found effectively.
1 Introduction
The major criticism brought forward against the notion of Martin-L¨of ran-domness is that, while it captures almost all important probabilistic laws, itis not completely intuitive, since it is not characterized by
computable
mar-tingales but by
recursively enumerable
ones (or by an equivalent r.e. testnotion).This point was issued first by Schnorr [26,27], who asserted that Martin- of randomness was too strong to be regarded as an
effective notion 
of ran-domness. He proposed two alternatives, one defined via coverings with mea-sures which are computable real numbers (not merely left-r.e.), leading tothe concept today known as
Schnorr randomness
[27]. The other concept is based on the unpredictability paradigm; it demands that no
computable
betting strategy should win against a random sequence. This notion is com-monly referred to as
computable randomness
[27].If one is interested in obtaining stronger notions of randomness, closer toMartin-L¨of randomness, without abandoning Schnorr’s paradigm, one mightstay with computable betting strategies and think of more general ways thosestrategies could be allowed to bet. One possibility is to remove the require-ment that the betting strategy must bet on a given sequence in an order thatis
monotonic on the prefixes of that sequence
, that is, the strategy itself deter-mines which place of the sequence it wants to bet against next. The resultingconcept of 
non-monotonic betting strategies
is a generalization of the conceptof monotonic betting strategies. An infinite binary sequence against which nocomputable non-monotonic betting strategy succeeds is called
Kolmogorov-Loveland random 
, or KL-random, for short. The concept is named after Kol-mogorov [9] and Loveland [14], who studied non-monotonic selection rules to define accordant stochasticity concepts, which we will describe later.
 
1 INTRODUCTION 
3The concept of KL-randomness is robust in so far as it remains the sameno matter whether one defines it in terms of computable or partial com-putable non-monotonic betting strategies [18]; in terms of the latter, the concept has been introduced by Muchnik, Semenov, and Uspensky[20] in 1998. They showed that Martin-of randomness implies KL-randomness,but it is not known whether the two concepts are different. This questionwas raised by Muchnik, Semenov, and Uspensky [20] and by Ambos-Spiesand Kuˇcera [1]. It is still a major open problem in the area. A proof that both concepts are the same would give a striking argument against Schnorr’scriticism of Martin-L¨of randomness.Most researchers conjecture that the notions are different. However, a re-sult of Muchnik [20]indicates that KL-randomness is rather close to Martin- of randomness.Recall that it is possible to characterize Martin-L¨of randomness as incom-pressibility with respect to prefix-free Kolmogorov complexity K: A sequence
A
is Martin-L¨of random if and only if there is a constant
c
such that for all
n
the prefix-free Kolmogorov complexity of the length
n
prefix
A
n
of 
A
is atleast
n
c
. It follows that a sequence
A
cannot be Martin-L¨of random if there is a function
h
such thatK(
A
h
(
c
)
)
h
(
c
)
c
for every
c.
(1)On the other hand, by the result of Muchnik [20] a sequence
A
cannot beKL-random if (1) holds for a
computable
function
h
. So, the difference be-tween Martin-L¨of randomness and KL-randomness appears, from this view-point, rather small. Not being Martin-L¨of random means that for any givenconstant bound there are infinitely many initial segments for which the com-pressibility exceeds this bound. If, moreover, we are able to detect such initialsegments efficiently (by means of a computable function), then the sequencecannot even be KL-random.In this paper we continue the investigations by Muchnik, Semenov, andUspensky, and give additional evidence that KL-randomness is very close toMartin-L¨of randomness.In Section4we refine a splitting technique that Muchnik used in orderto obtain the result mentioned above. We show the following: if 
A
is KL-random and
is a computable, infinite and co-infinite set of natural numbers,either the bits of 
A
whose positions are in
or the remaining bits form aMartin-L¨of random sequence. In fact, both do if 
A
is
02
. Moreover, in thatcase, for each computable, nondecreasing and unbounded function
g
it holdsthat K(
A
n
)
n
g
(
n
) for all but finitely many
n
.In Section5we construct counterexamples that show that two of theimplications mentioned in the preceding paragraph cannot be extended to

Activity (4)

You've already reviewed this. Edit your review.
1 hundred reads
rgomezv liked this
chongkm liked this
blablablou liked this

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->