SearcharxivSearch

arXiv subjects

Vladimir Anashin

Publications and source records attributed to Vladimir Anashin.

13 recordsLinked to original sources

Quantization causes waves:Smooth finitely computable functions are affine

Given an automaton (a letter-to-letter transducer, a dynamical 1-Lipschitz system on the space $\mathbb Z_p$ of $p$-adic integers) $\mathfrak A$ whose input and output alphabets are $\mathbb F_p=\{0,1,\ldots,p-1\}$, one visualizes word transformations performed by $\mathfrak A$ by a point set $\mathbf P(\mathfrak A)$ in real plane $\mathbb R^2$. For a finite-state automaton $\mathfrak A$, it is shown that once some points of $\mathbf P(\mathfrak A)$ constitute a smooth (of a class $C^2$) curve in $\mathbb R^2$, the curve is a segment of a straight line with a rational slope; and there are only finitely many straight lines whose segments are in $\mathbf{P}(\mathfrak A)$. Moreover, when identifying $\mathbf P(\mathfrak A)$ with a subset of a 2-dimensional torus $\mathbb T^2\subset\mathbb R^3$ (under a natural mapping of the real unit square $[0,1]^2$ onto $\mathbb T^2$) the smooth curves from $\mathbf P(\mathfrak A)$ constitute a collection of torus windings. In cylindrical coordinates either of the windings can be ascribed to a complex-valued function $ψ(x)=e^{i(Ax-2πB(t))}$ $(x\in\mathbb R)$ for suitable rational $A,B(t)$. Since $ψ(x)$ is a standard expression for a matter wave in quantum theory (where $B(t)=tB(t_0)$), and since transducers can be regarded as a mathematical formalization for causal discrete systems, the paper might serve as a mathematical reasoning why wave phenomena are inherent in quantum systems: This is because of causality principle and the discreteness of matter.

math.DS

T-functions revisited: New criteria for bijectivity/transitivity

The paper presents new criteria for bijectivity/transitivity of T-functions and fast knapsack-like algorithm of evaluation of a T-function. Our approach is based on non-Archimedean ergodic theory: Both the criteria and algorithm use van der Put series to represent 1-Lipschitz $p$-adic functions and to study measure-preservation/ergodicity of these.

cs.CR

The Non-Archimedean Theory of Discrete Systems

In the paper, we study behavior of discrete dynamical systems (automata) w.r.t. transitivity; that is, speaking loosely, we consider how diverse may be behavior of the system w.r.t. variety of word transformations performed by the system: We call a system completely transitive if, given arbitrary pair $a,b$ of finite words that have equal lengths, the system $\mathfrak A$, while evolution during (discrete) time, at a certain moment transforms $a$ into $b$. To every system $\mathfrak A$, we put into a correspondence a family $\mathcal F_{\mathfrak A}$ of continuous maps of a suitable non-Archimedean metric space and show that the system is completely transitive if and only if the family $\mathcal F_{\mathfrak A}$ is ergodic w.r.t. the Haar measure; then we find easy-to-verify conditions the system must satisfy to be completely transitive. The theory can be applied to analyze behavior of straight-line computer programs (in particular, pseudo-random number generators that are used in cryptography and simulations) since basic CPU instructions (both numerical and logical) can be considered as continuous maps of a (non-Archimedean) metric space $\mathbb Z_2$ of 2-adic integers.

math.DS

Ergodicity criteria for non-expanding transformations of 2-adic spheres

In the paper, we obtain necessary and sufficient conditions for ergodicity (with respect to the normalized Haar measure) of discrete dynamical systems $ $ on 2-adic spheres $\mathbf S_{2^{-r}}(a)$ of radius $2^{-r}$, $r\ge 1$, centered at some point $a$ from the ultrametric space of 2-adic integers $\mathbb Z_2$. The map $f\colon\mathbb Z_2\to\mathbb Z_2$ is assumed to be non-expanding and measure-preserving; that is, $f$ satisfies a Lipschitz condition with a constant 1 with respect to the 2-adic metric, and $f$ preserves a natural probability measure on $\mathbb Z_2$, the Haar measure $μ_2$ on $\mathbb Z_2$ which is normalized so that $μ_2(\mathbb Z_2)=1$.

math.DS

Automata finiteness criterion in terms of van der Put series of automata functions

In the paper we develop the $p$-adic theory of discrete automata. Every automaton $\mathfrak A$ (transducer) whose input/output alphabets consist of $p$ symbols can be associated to a continuous (in fact, 1-Lipschitz) map from $p$-adic integers to $p$ integers, the automaton function $f_\mathfrak A$. The $p$-adic theory (in particular, the $p$-adic ergodic theory) turned out to be very efficient in a study of properties of automata expressed via properties of automata functions. In the paper we prove a criterion for finiteness of the number of states of automaton in terms of van der Put series of the automaton function. The criterion displays connections between $p$-adic analysis and the theory of automata sequences.

cs.FL

Non-Archimedean Ergodic Theory and Pseudorandom Generators

The paper develops techniques in order to construct computer programs, pseudorandom number generators (PRNG), that produce uniformly distributed sequences. The paper exploits an approach that treats standard processor instructions (arithmetic and bitwise logical ones) as continuous functions on the space of 2-adic integers. Within this approach, a PRNG is considered as a dynamical system and is studied by means of the non-Archimedean ergodic theory.

math.DS

Non-Archimedean analysis, T-functions, and cryptography

These are lecture notes of a 20-hour course at the International Summer School \emph{Mathematical Methods and Technologies in Computer Security} at Lomonosov Moscow State University, July 9--23, 2006. Loosely speaking, a $T$-function is a map of $n$-bit words into $n$-bit words such that each $i$-th bit of image depends only on low-order bits $0,..., i$ of the pre-image. For example, all arithmetic operations (addition, multiplication) are $T$-functions, all bitwise logical operations ($\XOR$, $\AND$, etc.) are $T$-functions. Any composition of $T$-functions is a $T$-function as well. Thus $T$-functions are natural computer word-oriented functions. It turns out that $T$-functions are continuous (and often differentiable!) functions with respect to the so-called 2-adic distance. This observation gives a powerful tool to apply 2-adic analysis to construct wide classes of $T$-functions with provable cryptographic properties (long period, balance, uniform distribution, high linear complexity, etc.); these functions currently are being used in new generation of fast stream ciphers. We consider these ciphers as specific automata that could be associated to dynamical systems on the space of 2-adic integers. From this view the lectures could be considered as a course in cryptographic applications of the non-Archimedean dynamics; the latter has recently attracted significant attention in connection with applications to physics, biology and cognitive sciences. During the course listeners study non-Archimedean machinery and its applications to stream cipher design.

cs.CR

Wreath Products in Stream Cipher Design

The paper develops a novel approach to stream cipher design: Both the state update function and the output function of the corresponding pseudorandom generators are compositions of arithmetic and bitwise logical operations, which are standard instructions of modern microprocessors. Moreover, both the state update function and the output function are being modified dynamically during the encryption. Also, these compositions could be keyed, so the only information available to an attacker is that these functions belong to some exponentially large class. The paper shows that under rather loose conditions the output sequence is uniformly distributed, achieves maximum period length and has high linear complexity and high $\ell$-error linear complexity. Ciphers of this kind are flexible: One could choose a suitable combination of instructions to obtain due performance without affecting the quality of the output sequence. Finally, some evidence is given that a key recovery problem for (reasonably designed) stream ciphers of this kind is intractable up to plausible conjectures.

cs.CR

Ergodic Transformations of the Space of $p$-adic Integers

Let $\mathcal L_1$ be the set of all mappings $f\colon\Z_p\Z_p$ of the space of all $p$-adic integers $\Z_p$ into itself that satisfy Lipschitz condition with a constant 1. We prove that the mapping $f\in\mathcal L_1$ is ergodic with respect to the normalized Haar measure on $\Z_p$ if and only if $f$ induces a single cycle permutation on each residue ring $\Z/p^k\Z$ modulo $p^k$, for all $k=1,2,3,...$. The multivariate case, as well as measure-preserving mappings, are considered also. Results of the paper in a combination with earlier results of the author give explicit description of ergodic mappings from $\mathcal L_1$. This characterization is complete for $p=2$. As an application we obtain a characterization of polynomials (and certain locally analytic functions) that induce ergodic transformations of $p$-adic spheres. The latter result implies a solution of a problem (posed by A.~Khrennikov) about the ergodicity of a perturbed monomial mapping on a sphere.

math.DS

Pseudorandom number generation by p-adic ergodic transformations: an addendum

The paper study counter-dependent pseudorandom number generators based on $m$-variate ($m>1$) ergodic mappings of the space of 2-adic integers $\Z_2$. The sequence of internal states of these generators is defined by the recurrence law $\mathbf x_{i+1}= H^B_i(\mathbf x_i)\bmod{2^n}$, whereas their output sequence is %while its output sequence is of the $\mathbf z_{i}=F^B_i(\mathbf x_i)\mod 2^n$; here $\mathbf x_j, \mathbf z_j$ are $m$-dimensional vectors over $\Z_2$. It is shown how the results obtained for a univariate case could be extended to a multivariate case.

cs.CR

Pseudorandom number generation by $p$-adic ergodic transformations

The paper study counter-dependent pseudorandom generators; the latter are generators such that their state transition function (and output function) is being modified dynamically while working: For such a generator the recurrence sequence of states satisfies a congruence $x_{i+1}\equiv f_i(x_i)\pmod{2^n}$, while its output sequence is of the form $z_{i}=F_i(u_i)$. The paper introduces techniques and constructions that enable one to compose generators that output uniformly distributed sequences of a maximum period length and with high linear and 2-adic spans. The corresponding stream chipher is provably strong against a known plaintext attack (up to a plausible conjecture). Both state transition function and output function could be key-dependent, so the only information available to a cryptanalyst is that these functions belong to some (exponentially large) class. These functions are compositions of standard machine instructions (such as addition, multiplication, bitwise logical operations, etc.) The compositions should satisfy rather loose conditions; so the corresponding generators are flexible enough and could be easily implemented as computer programs.

cs.CR

Uniformly distributed sequences of p-adic integers, II

The paper describes ergodic (with respect to the Haar measure) functions in the class of all functions, which are defined on (and take values in) the ring of p-adic integers, and which satisfy (at least, locally) Lipschitz condition with coefficient 1. Equiprobable (in particular, measure-preserving) functions of this class are described also. In some cases (and especially for p=2) the descriptions are given by explicit formulae. Some of the results may be viewed as descriptions of ergodic isometric dynamical systems on p-adic unit disk. The study was motivated by the problem of pseudorandom number generation for computer simulation and cryptography. From this view the paper describes nonlinear congruential pseudorandom generators modulo M which produce stricly periodic uniformly distributed sequences modulo M with maximal possible period length (i.e., exactly M). Both the state change function and the output function of these generators could be, e.g., meromorphic functions (in particular, polynomials with rational, but not necessarily integer coefficients, or rational functions), or compositions of arithmetical operations (like addition, multiplication, exponentiation, raising to integer powers, including negative ones) with standard computer operations, such as bitwise logical operations (XOR, OR, AND, etc.). The linear complexity of the produced sequences is also studied.

math.NT