Relative Universality On classes of Real Computation Hector Zenil University of Lille 1 Université de Paris 1 (Panthéon-Sorbonne)

Slides:



Advertisements
Similar presentations
Completeness and Expressiveness
Advertisements

Lecture 3 Universal TM. Code of a DTM Consider a one-tape DTM M = (Q, Σ, Γ, δ, s). It can be encoded as follows: First, encode each state, each direction,
Lecture 16 Deterministic Turing Machine (DTM) Finite Control tape head.
CS 461 – Nov. 9 Chomsky hierarchy of language classes –Review –Let’s find a language outside the TM world! –Hints: languages and TM are countable, but.
CSCI 4325 / 6339 Theory of Computation Zhixiang Chen Department of Computer Science University of Texas-Pan American.
1 COMP 382: Reasoning about algorithms Unit 9: Undecidability [Slides adapted from Amos Israeli’s]
Lecture 6 Hyperreal Numbers (Nonstandard Analysis)
Computability and Complexity 4-1 Existence of Undecidable Problems Computability and Complexity Andrei Bulatov.
Complexity 15-1 Complexity Andrei Bulatov Hierarchy Theorem.
Prof. Busch - LSU1 Decidable Languages. Prof. Busch - LSU2 Recall that: A language is Turing-Acceptable if there is a Turing machine that accepts Also.
Courtesy Costas Busch - RPI1 A Universal Turing Machine.
CPSC 411, Fall 2008: Set 12 1 CPSC 411 Design and Analysis of Algorithms Set 12: Undecidability Prof. Jennifer Welch Fall 2008.
Lecture 9 Recursive and r.e. language classes
Fall 2004COMP 3351 Recursively Enumerable and Recursive Languages.
1 Module 9 Recursive and r.e. language classes –representing solvable and half-solvable problems Proofs of closure properties –for the set of recursive.
Lecture 8 Recursively enumerable (r.e.) languages
1 Undecidability Andreas Klappenecker [based on slides by Prof. Welch]
Recursively Enumerable and Recursive Languages
1 The Chomsky Hierarchy. 2 Unrestricted Grammars: Rules have form String of variables and terminals String of variables and terminals.
1 Uncountable Sets continued Theorem: Let be an infinite countable set. The powerset of is uncountable.
Fall 2004COMP 3351 The Chomsky Hierarchy. Fall 2004COMP 3352 Non-recursively enumerable Recursively-enumerable Recursive Context-sensitive Context-free.
Fall 2004COMP 3351 Reducibility. Fall 2004COMP 3352 Problem is reduced to problem If we can solve problem then we can solve problem.
1 Foundations of Software Design Fall 2002 Marti Hearst Lecture 29: Computability, Turing Machines, Can Computers Think?
Fall 2005Costas Busch - RPI1 Recursively Enumerable and Recursive Languages.
Fall 2004COMP 3351 A Universal Turing Machine. Fall 2004COMP 3352 Turing Machines are “hardwired” they execute only one program A limitation of Turing.
Courtesy Costas Busch - RPI1 Reducibility. Courtesy Costas Busch - RPI2 Problem is reduced to problem If we can solve problem then we can solve problem.
CS5371 Theory of Computation Lecture 4: Automata Theory II (DFA = NFA, Regular Language)
1 Reducibility. 2 Problem is reduced to problem If we can solve problem then we can solve problem.
Formal Language Finite set of alphabets Σ: e.g., {0, 1}, {a, b, c}, { ‘{‘, ‘}’ } Language L is a subset of strings on Σ, e.g., {00, 110, 01} a finite language,
Turing Machines A more powerful computation model than a PDA ?
CMPS 3223 Theory of Computation Automata, Computability, & Complexity by Elaine Rich ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Slides provided.
Discrete dynamical systems and intrinsic computability Marco Giunti University of Cagliari, Italy
Ding-Zhu Du Office: ECSS 3-611, M 3:15-4:30 Lecture: ECSS 2.311, MW 12:30-1:45.
1 Undecidability Reading: Chapter 8 & 9. 2 Decidability vs. Undecidability There are two types of TMs (based on halting): (Recursive) TMs that always.
1 1 CDT314 FABER Formal Languages, Automata and Models of Computation Lecture 15-1 Mälardalen University 2012.
1 The Halting Problem and Decidability How powerful is a TM? Any program in a high level language can be simulated by a TM. Any algorithmic procedure carried.
Great Theoretical Ideas in Computer Science about AWESOME Some Generating Functions Probability Infinity Computability With Alan! (not Turing) Mind-
A Universal Turing Machine
Interactive proof systems Section 10.4 Giorgi Japaridze Theory of Computability.
Remaining Discussions from Previous Class Please be precise in your writing –Specially because some of the proofs are written in plain English Queue automata.
1 Linear Bounded Automata LBAs. 2 Linear Bounded Automata (LBAs) are the same as Turing Machines with one difference: The input string tape space is the.
CS 208: Computing Theory Assoc. Prof. Dr. Brahim Hnich Faculty of Computer Sciences Izmir University of Economics.
1 Turing’s Thesis. 2 Turing’s thesis: Any computation carried out by mechanical means can be performed by a Turing Machine (1930)
Fundamentals of Informatics Lecture 3 Turing Machines Bas Luttik.
Strings Basic data type in computational biology A string is an ordered succession of characters or symbols from a finite set called an alphabet Sequence.
NP-Completness Turing Machine. Hard problems There are many many important problems for which no polynomial algorithms is known. We show that a polynomial-time.
Lecture 21 Reducibility. Many-one reducibility For two sets A c Σ* and B c Γ*, A ≤ m B if there exists a Turing-computable function f: Σ* → Γ* such that.
Recursively Enumerable and Recursive Languages
Complexity 24-1 Complexity Andrei Bulatov Interactive Proofs.
1 Linear Bounded Automata LBAs. 2 Linear Bounded Automata (LBAs) are the same as Turing Machines with one difference: The input string tape space is the.
Overview of the theory of computation Episode 3 0 Turing machines The traditional concepts of computability, decidability and recursive enumerability.
Turing Machines CS 130 Theory of Computation HMU Textbook: Chap 8.
The Church-Turing Thesis Chapter Are We Done? FSM  PDA  Turing machine Is this the end of the line? There are still problems we cannot solve:
1 Recursively Enumerable and Recursive Languages.
Recursively Enumerable and Recursive Languages. Definition: A language is recursively enumerable if some Turing machine accepts it.
1 A Universal Turing Machine. 2 Turing Machines are “hardwired” they execute only one program A limitation of Turing Machines: Real Computers are re-programmable.
Chapters 11 and 12 Decision Problems and Undecidability.
A Universal Turing Machine
Recursively Enumerable and Recursive Languages
CIS Automata and Formal Languages – Pei Wang
Linear Bounded Automata LBAs
Theory of Computation Lecture 22: Turing Machines III
LIMITS OF ALGORITHMIC COMPUTATION
Hierarchy of languages
Decidable Languages Costas Busch - LSU.
Proposed in Turing’s 1936 paper
Formal Languages, Automata and Models of Computation
Recall last lecture and Nondeterministic TMs
More Undecidable Problems
Theory of Computation Lecture 23: Turing Machines III
Presentation transcript:

Relative Universality On classes of Real Computation Hector Zenil University of Lille 1 Université de Paris 1 (Panthéon-Sorbonne)

Real numbers Definitions: A real number is of the form: X.{0,1}* {0,1}* can be always seen as a subset of natural numbers. Lets say that a subset S in R is complete if S contains at least one representative member of each complexity degree of R. by instance any non-empty open or closed ball [x,y) in R with x!=y

Arithmetical Hierarchy A formula is in Sigma_1 if it is of the form: En_1,En_2,…,En_mQ(n_1,n_2,…n_i) with Q a free quantifier recursive formula. A formula is in Pi_1 if it is of the form: An_1,An_2,…,An_mQ(n_1,n_2,…n_i) with Q a free quantifier recursive formula. Inductively: A formula is in Sigma_n if it is of the form: E^m Q_n-1 where Q has n-1 quantifiers alternations. And in Pi_n if A^m Q_n-1 where Q has n-1 quantifiers alternations. Delta_n=Sigma_n Intersection Pi_n

Oracle Turing machine A Turing machine with an aditional tape called oracle tape and three new states: q_?, q_y and q_n. With a Turing machine gets into the state q_? The oracle answers using the oracle tape and entering into the “yes” -q_y- or “no” states -q_n-. The oracle tape is made by 0’s and 1’s depending if the answer to q_? is yes or no.

Post Theorem Establishes a close connection between the arithmetical hierarchy of sets of natural numbers and the Turing degrees. A formula F is in Sigma_n iff F is a recursively enumerable set with an oracle O^n

Real numbers and oracles If O is an oracle machine O=r=X.r_1,r_2,… for some r real number since f(r_n)=1 if O answers yes for a question concerning the n-th string of a language, and f(r_n)=1 otherwise. And where f is the characteristic function for O and a language L.

Degree of computability of a real number The Turing degree of a real number r_L is defined as the Turing degree of its binary expansion since it can be seen as a subset of natural numbers

Model of Computation A model of computation M={D,F}. A domain of operation D and a set of operators F. An automata T in M is an abstract machine taking inputs from D and applying a set of Functions F’ in F. So if M is closed under F, T is closed under F’ in M.

Intrinsic Universality We are going to say that a class L is intrinsically universal if there exists an abstract machine U in L capable to behave as any other abstract machine M in L given the transition table and input for M. Intrinsic universality = universality in the Turing sense The class of recursive functions is evidently intrinsically universal

Assuming a Delta_n^m set of real numbers D: D allows Delta_n^m-universality In other words, each level of the arithmetical and the hyper arithmetical hierarchies admit a universal abstract machine for that level. Relative universality

Universality Hierarchy Because the arithmetical and hyper arithmetical hierarchies do not collapse, there is no Delta_n universal machine able to behave as any other Delta_n+1 machine Unless…

Universality Jump Operator Let be F the set of functions allowed in a model of computation M with inputs at most in Delta_n^m for some n and m. Lets call f a universal jump operator if a model of computation M allows at least one function f such that: f(i)=j with deg_T(i) <deg_T(j) In other words f converts inputs of certain degree into outputs of higher Turing degree. Then M is non-closed under its operators

Collapsing the hierarchies If jump functions are allowed, the hierarchies or at least some parts of them collapse under those models of computation. Lets call a universality jump a complete universal jump if at least one function f is allowed such that for an input i and deg_T(i)<Delta_n^m : f(i)=j with deg_T(j)= Delta_n’^m’ for n’ and m’ arbitrary non-negative numbers. In other words M reaches any AH and hyper AH level through f. Both AH and hyper AH collapse under M

Automata Complexity The complexity of an automata denoted by C(A) is the measure defined by the set of maximal* Turing degrees of the set of all possible outputs O in D -either M is closed under F or not-. If M is not closed under F deg_T(O)>deg_T(D). If M is closed the complexity of A is at most the maximal Turing degree of D. * Because the Turing degrees are a partial order

Universality Loss Finally, if a model of computation M is intrinsically universal then: 1.The scope of operation of M is arbitrary bound up to a certain computability degree in the AH 2.M computes at most the class of the recursive functions (M is Turing equivalent) (a case of 1) 3.M is non-closed under its functions F=f_1,f_2,…,f_n because a subset G in F performs a complete universal jump*. * Another possible case would be F as infinite, however a finite set F’ could be built from F composing all the functions from F into a new function f’ in F’. The result would be a non-recursive function f’ and a finite set F which would falls into the case number 3.

Relative Universality (sketch proof) Take any level of the AH e.g. Delta_n. If M is a closed model of computation: (building the Delta_n-universal machine) –Take the usual universal Turing machine and an oracle O^n. Because the Turing degree of an automata M with domain Delta_n is at most in Delta_n F:Delta_n->Delta_n and applying Post’s theorem O^n computes x for all x real number in Delta_n, then M is universal.

Universality Loss (sketch proof) (lack of absolute universality in R) Assuming the complete set of real numbers denoted by R (or any complete subset as defined before) we claim that: There is no absolute universal machine in R 1.Suppose U is such universal machine. Then U is capable to behave as any other machine M in R 2.Lets take MAX the set of maximal Turing degrees of U which defines C(U). Then take r a real number such that deg_T(r)>C(U), then build M an abstract automata able to compute r, then U is not able to compute “Mx”, where x is the input for M. Thus U is not a universal machine for R. Therefore there is no possible U in R. Unless…

Universality conditions in real computation The model of real computation can be intrinsically universal only if it allows: A set of functions F of arbitrary power going through all the AH and hyper AH reaching any possible level. Only in such case M can be closed under its operators and intrinsically universal. In other words, the model is not a field since it is not closed under the set of functions F unless it allows the whole complexity class of R.

Conclusions Three possible scenarios for intrinsic universality are possible: 1.The class of recursive functions 2.The class of any level of the AH (an infinite number of models of different power) with non-recursive functions up to that level in the AH (case 1 is naturally in the first level of this case) 3.The class of real computation with access to all real numbers and functions able to reach any level of the AH and the hyper AH.

Conclusions 2 There is a universal model of computation for each level of the AH and the hyper AH picking the correct set of functions F for each level. In a Discrete/Continuum dichotomy the election of an special set F and an arbitrary level in the AH does not seem natural. If the Continuum is taken as R -as usual- only arbitrary powerful functions makes the model intrinsically universal otherwise the model does not allow intrinsic universality.

Conclusions 3 In other words: The appropriate hierarchy for computations with real numbers are beyond the scope of the AH and the HAH. There is no universal machine for R at any level of the AH or the HAH, therefore the notion of real computation is not well-founded in the sense of lack of an abstract universality machine. Hyper-models do not converge like those Turing-equivalent which supports the Church- Turing thesis. Therefore there is no a single Church-Turing thesis for real computation but an infinite hierarchy of Church-Turing type thesis based on each of the universal devices at each level of the AH and HAH.