RECHERCHES Womcodes constructed with projective geometries «Womcodes» construits à partir de géométries projectives Frans MERKX (') École Nationale Su

Similar documents
Physical Layer: Outline

SOME CONSTRUCTIONS OF OPTIMAL BINARY LINEAR UNEQUAL ERROR PROTECTION CODES

Chapter 3 Digital Transmission Fundamentals

Department of Electrical Engineering, University of Waterloo. Introduction

CHAPTER 3 INEQUALITIES. Copyright -The Institute of Chartered Accountants of India

CS 477/677 Analysis of Algorithms Fall 2007 Dr. George Bebis Course Project Due Date: 11/29/2007

Pre-Calculus Individual Test 2017 February Regional

1996 Engineering Systems Design and Analysis Conference, Montpellier, France, July 1-4, 1996, Vol. 7, pp

and the Doppler frequency rate f R , can be related to the coefficients of this polynomial. The relationships are:

Chapter Summary. Mathematical Induction Strong Induction Recursive Definitions Structural Induction Recursive Algorithms

Pattern Recognition 2014 Support Vector Machines

Differentiation Applications 1: Related Rates

Physics 2010 Motion with Constant Acceleration Experiment 1

Chapter 3: Cluster Analysis

Determining the Accuracy of Modal Parameter Estimation Methods

(2) Even if such a value of k was possible, the neutrons multiply

Homology groups of disks with holes

NUROP CONGRESS PAPER CHINESE PINYIN TO CHINESE CHARACTER CONVERSION

CHAPTER 4 DIAGNOSTICS FOR INFLUENTIAL OBSERVATIONS

Lyapunov Stability Stability of Equilibrium Points

Department of Economics, University of California, Davis Ecn 200C Micro Theory Professor Giacomo Bonanno. Insurance Markets

READING STATECHART DIAGRAMS

initially lcated away frm the data set never win the cmpetitin, resulting in a nnptimal nal cdebk, [2] [3] [4] and [5]. Khnen's Self Organizing Featur

k-nearest Neighbor How to choose k Average of k points more reliable when: Large k: noise in attributes +o o noise in class labels

THE LIFE OF AN OBJECT IT SYSTEMS

Sequential Allocation with Minimal Switching

ENSC Discrete Time Systems. Project Outline. Semester

COMP 551 Applied Machine Learning Lecture 9: Support Vector Machines (cont d)

1 The limitations of Hartree Fock approximation

On small defining sets for some SBIBD(4t - 1, 2t - 1, t - 1)

Keysight Technologies Understanding the Kramers-Kronig Relation Using A Pictorial Proof

Admissibility Conditions and Asymptotic Behavior of Strongly Regular Graphs

Bootstrap Method > # Purpose: understand how bootstrap method works > obs=c(11.96, 5.03, 67.40, 16.07, 31.50, 7.73, 11.10, 22.38) > n=length(obs) >

Support-Vector Machines

Lab 1 The Scientific Method

The blessing of dimensionality for kernel methods

Floating Point Method for Solving Transportation. Problems with Additional Constraints

MODULE 1. e x + c. [You can t separate a demominator, but you can divide a single denominator into each numerator term] a + b a(a + b)+1 = a + b

BASD HIGH SCHOOL FORMAL LAB REPORT

Chapter 3 Digital Transmission Fundamentals

Medium Scale Integrated (MSI) devices [Sections 2.9 and 2.10]

SUPPLEMENTARY MATERIAL GaGa: a simple and flexible hierarchical model for microarray data analysis

You need to be able to define the following terms and answer basic questions about them:

2.161 Signal Processing: Continuous and Discrete Fall 2008

Dead-beat controller design

From Signals to Packets. Outline : Introduction to Telecommunication Networks Lectures 5: Datalink layer. Error Control.

Dataflow Analysis and Abstract Interpretation

CONSTRUCTING STATECHART DIAGRAMS

A New Evaluation Measure. J. Joiner and L. Werner. The problems of evaluation and the needed criteria of evaluation

Section 6-2: Simplex Method: Maximization with Problem Constraints of the Form ~

A Simple Set of Test Matrices for Eigenvalue Programs*

CESAR Science Case The differential rotation of the Sun and its Chromosphere. Introduction. Material that is necessary during the laboratory

Review Problems 3. Four FIR Filter Types

Preparation work for A2 Mathematics [2017]

o o IMPORTANT REMINDERS Reports will be graded largely on their ability to clearly communicate results and important conclusions.

Source Coding Fundamentals

Methods for Determination of Mean Speckle Size in Simulated Speckle Pattern

Math Foundations 20 Work Plan

Computational modeling techniques

OF SIMPLY SUPPORTED PLYWOOD PLATES UNDER COMBINED EDGEWISE BENDING AND COMPRESSION

Interference is when two (or more) sets of waves meet and combine to produce a new pattern.

Subject description processes

Chapter 8 Predicting Molecular Geometries

TRAINING GUIDE. Overview of Lucity Spatial

This section is primarily focused on tools to aid us in finding roots/zeros/ -intercepts of polynomials. Essentially, our focus turns to solving.

Chapter 3 Kinematics in Two Dimensions; Vectors

MODULE FOUR. This module addresses functions. SC Academic Elementary Algebra Standards:

An Area Efficient (31, 16) BCH Decoder for Three Errors

Kinematic transformation of mechanical behavior Neville Hogan

SPH3U1 Lesson 06 Kinematics

Computational modeling techniques

Tree Structured Classifier

The standards are taught in the following sequence.

Revision: August 19, E Main Suite D Pullman, WA (509) Voice and Fax

Trigonometric Ratios Unit 5 Tentative TEST date

Lead/Lag Compensator Frequency Domain Properties and Design Methods

FINITE BOOLEAN ALGEBRA. 1. Deconstructing Boolean algebras with atoms. Let B = <B,,,,,0,1> be a Boolean algebra and c B.

3.4 Shrinkage Methods Prostate Cancer Data Example (Continued) Ridge Regression

Public Key Cryptography. Tim van der Horst & Kent Seamons

B. Definition of an exponential

Least Squares Optimal Filtering with Multirate Observations

^YawataR&D Laboratory, Nippon Steel Corporation, Tobata, Kitakyushu, Japan

CHAPTER 2 Algebraic Expressions and Fundamental Operations

Perfrmance f Sensitizing Rules n Shewhart Cntrl Charts with Autcrrelated Data Key Wrds: Autregressive, Mving Average, Runs Tests, Shewhart Cntrl Chart

Preparation work for A2 Mathematics [2018]

, which yields. where z1. and z2

Flipping Physics Lecture Notes: Simple Harmonic Motion Introduction via a Horizontal Mass-Spring System

MAKING DOUGHNUTS OF COHEN REALS

Application Of Mealy Machine And Recurrence Relations In Cryptography

Turing Machines. Human-aware Robotics. 2017/10/17 & 19 Chapter 3.2 & 3.3 in Sipser Ø Announcement:

What is Statistical Learning?

PLEASURE TEST SERIES (XI) - 07 By O.P. Gupta (For stuffs on Math, click at theopgupta.com)

We can see from the graph above that the intersection is, i.e., [ ).

Chemistry 20 Lesson 11 Electronegativity, Polarity and Shapes

Cambridge Assessment International Education Cambridge Ordinary Level. Published

How do scientists measure trees? What is DBH?

Aristotle I PHIL301 Prof. Oakes Winthrop University updated: 3/14/14 8:48 AM

COMP 551 Applied Machine Learning Lecture 11: Support Vector Machines

A New Approach to Increase Parallelism for Dependent Loops

Transcription:

Wmcdes cnstructed with prjective gemetries «Wmcdes» cnstruits à partir de gémétries prjectives Frans MERKX (') Écle Natinale Supérieure de Télécmmunicatins (ENST), 46, rue Barrault, 75013 PARIS Étudiant à l'université de Technlgie à Eindhven (Pays-Bas) dans le dmaine des mathématiques discrètes. SUMMARY We cnsider strage media which cnsist f a number f write-nce bit psitins (wits). A wit initially cntains a "0", that may be irreversibly verwritten with a "1". It was shwn by Rivest and Shamir [5] that, by cding techniques ne can reuse such a write-nce memry (wm) up t a very high rate. We present tw new cyclic wmcdes, based n PG (2,2) and PG (3,2) respectively, which attain the RS-bund. These cdes can be decded with a decding algrithm fr Hamming cdes. Sme ther high-rate wmcdes, derived frm thse abve, are discussed. Errr-crrecting cdes, finite gemetry, numerical strage media. KEY WORDS RÉSUMÉ Nus cnsidérns des mémires cnstituées de psitins permettant l'écriture irréversible d'un bit (wits). Un wit cntient initialement un zér, qui peut être définitivement transfrmé en un. Nus utilisns des techniques de cdage pur réutiliser ces mémires à écriture unique avec un rendement élevé. Cdes crrecteurs d'erreurs, gémétrie finie, mémires numériques. MOTS CLÉS (') During this research, the authr was with the Écle Natinale Supérieure de télécmmunicatin, Paris. 1227

WOMCODES CONSTRUCTED WITH PROJECTIVE GEOMETRIES CONTENTS 1. Intrductin 2. Ntatin 3. A < 7 > 4 /7 wmcde 4. Mre wmcdes frm prjective gemetries 5. Extensins Cnclusin Acknwledgements References 1. Intrductin We cnsider strage media which cnsist f a number f write-nce bit psitins (wits). A wit initially cntains a "0", that may be irreversibly verwritten with a "1". We call such a strage medium a "writence memry" r wm. Examples f wms are punched cards r digital ptical disks. In their pineering paper n this subject, Rivest and Shamir [5] shwed that it is pssible t use a wm several times, by using "wmcdes". They gave many examples f wmcdes, and shw that the "capacity" (defined later) f a wmcde is greater than the number f wits. They als derive asympttic results fr this capacity. The fllwing cding scheme was their prime "mtivating example". Example 1. 1 : We write tw times 2 bits in a memry f 3 wits, as fllws : Message x First writing r (x) Secnd writing r' (x) 00 000 111 01 100 011 10 010 101 11 001 110 This scheme must be interpreted as fllws : The first time we receive a message x, we write r (x). Later, we will receive fr the secnd time a message, say y. If x =y we dn't change the memry, if x*y we change the memry state t r' (y), by nly changing 0's t 1's. Withut cding, we wuld have t use 4 wits. Remark that after the secnd writing we lst the infrmatin n the first message. 2. Ntatin We call a cding scheme that uses n wits t represent t times ne ut f v messages (i. e. write nce and change t - 1 times) a < v > t/n wmcde. Mre generai, with a < v l, v 2i..., vt >/n wmcde the message in the i'th generatin can lie ne f a set V, f vi messages. Such a wmcde must have the fllwing prperties : 1. Each memry cntent that ccurs must determine uniquely the last received message, and 2. fr each memry cntent x e { 0, 11" ccuring in the s'th generatin (s < t) it must be pssible t encde all sequences f messages K+1,.., m t ), mi e V i, such that x, :g x, 1 <... < x t (cmpnentwise). The wmcde is determined by giving fr all pssible memry states and new messages the new memry state (update functin [5]). w (< v t, v 2,..., vt >) dentes the least n fr which a ( v l, v 2,..., vt i/n wmcde exists, and, f curse, w(<v> t)=w (<v, v,..., v>). Rivest and Shamir [5] derive a lwer bund fr w (< v > 1 ). This can be easily extended t a lwer bund fr w (< v 1, v2,..., v t i), see [4], which we shall refer t as the RS-bund. The capacity C and the rate R f a < u t, v 2,..., vt >/n wmcde are defined as C :=1g (v 1.v 2 v t), R : =C/n. Fr example, the < 4 > 2/3 wmcde f example 1. 1 has C=4 and R=1.33... (=1.33... bit per wit). The fllwing tw examples f wmcdes are described in [5]. Example 2. 1 : A < 5 > 3 /5 cyclic wmcde (Rate= 1.39... ), cnstructed by D. Klarner. Message 1 is represented by 10000 in the first generatin, either 01001 r 00110 in the secnd, and ne f 01111, 10110 r 11001 in the third generatin. Message i, 1 <= i <_ 5 is represented by a cyclic shift ver i-1 psitins f the wrds fr the first message (since 5 is prime they are all distinct). Example 2.2 : A < 7 >'/7 cyclic wmcde (rate =1.60... ), cnstructed by D. Leavitt, by extending the methd f example 2. 1 (t appear). We will present a new cyclic < 7 >'/7 wmcde. The first, secnd and furth generatins are equivalent t these f Leavitt's cde, but the thid generatin is different (private cmmunicatin). We will use a prjective gemetry PG (2, 2) r 2-(7, 3, 1) symmetric blck design (Steiner triple system) r Fan plane. 1228

L These are different names fr the same bject, described in Figure 2. 3. Fig. 2.3. - The Fan plane. This picture shuld be interpreted as fllws : The plane cnsists f seven pints (numbered 1, 2,..., 7) and seven lins, each cntaining three pints (the six lines tgether with the circle in Figure 2. 3). Remark that any pair f pints is n exactly ne fine, and any tw lines intersect in exactly ne pint. With the enumeratin as in Figure 2. 3, the incidence vectrs f lines with the pints f the plane are the cyclic shifts f 1101000. In what fllws, we shall identify the seven pints f the plane with the wits f a seven-wit memry. Figure 3. 1 shuld be interpreted as fllws : pssible memry states f the wmcde are described by their crrespnding cnfiguratins [The encircled pints in (3. 1)]. Since there are seven messages, they are identified with the seven pints in the Fan plane. Fr each cnfiguratin the message (pint) represented is indicated by an arrw. Using the prperties f the Fan plane given after (2. 3) it is easy t check that Figure 3. 1 describes indeed a < 7 >'/7 wmcde : 1. Each memry state determines uniquely a message pint, and 2. Fr each cnfiguratin in generatin i, and fr each message pint received, it is pssible, by adding ne r tw pints, t find a cnfiguratin in generatin i+1 representing the received message pint. (NB : if the same message is received twice, the memry state is nt changed.) Example 3. 2: Suppse we receive the message sequence 2, 5, 3, 7. Using the wmcde f (3. 1), we btain the fllwing sequence f cnfiguratins representing them: As a cnsequence, every memry cntent can be identified with a cnfiguratin f pints in the plane (i. e. the pints fr which the crrespnding wit cntains a 1). 3. A < 7 >'/7 wmcde The < 7 >'/7 wmcde which we prpse is described in terms f cnfiguratins f pints in the Fan plane in Figure 3. 1 belw. generatin Ouater cnfiguratins Crrespnding t the sequence f memry cntents 0100000, 0110000, 1111000, 1111110. 1.O (a pint) DECODING 3 4 (tw pints) (a line and a pint) C. r c (a plane with ne r tw pints missing) Fig. 3. 1. - A descriptin f a < 7) /7 wmcde. Fr the decding, we use the fact that the cdewrds f the [7, 4] binary perfect Hamming cde are all the linear cmbinatins mdul 2 f the lins f the Fan plane (see, fr instance [3]). S the cde wrds are 0', lines, symmetric differences f tw lins, and the whle plane. Cnfiguratins f the wmcde in (3. 1) are never Hamming cdewrd, s they are at Hamming distance 1 frm exactly ne Hamming cde wrd. Mrever, by inspectin f (3. 1) we sec that this Hamming cdewrd is btained by adding (md 2) the message pint t the cnfiguratin. S, if we dente the memry cntent as a Hamming cdewrd, the errr vectr yields exactly the message 229

WOMCODES CONSTRUCTED WITH PROJECTIVE GEOMETRIES (i. e. using syndrme decding, each message crrespnds with ne f the seven pssible nn-zer syndrmes). Fr sme f these cnstraints the exact value f t is determined. If turns ut that, using nly the minimal weight cnstraint, des nt always (and prbably nly fr n = 3 r n = 7) yield the maximal t. 4. Mre wmcdes frm prjective gemetries Example 4. 1 : Cnsider the prjective gemetry PG (3, 2). It cntains 15 pints, 35 lines f 3 pints and 15 Fan planes f 7 pints. The [15, 11] Hamming cde can be seen as the cllectin f all linear cmbinatins md 2 f lins in PG (3, 2). Nw the apprach f sectins 3 can be generalised : it is pssible t cnstruct a < 15 >'/15 wmcde which can be "decded" by the methd described in sectin 3, i. e. the message represented by each memry state in the wmcde is just ils Hamming cde errr, and is btained by cmputing the syndrme (see [4] fr details). Example 4. 2 : Fix a line in the Fan plane. The cnfiguratins f the first tw generatins in (3. 1), restricted t thse n this line, describe a < 3 > 2 /3 wmcde. Since a line is a PG (1, 2), this wmcde culd be cnsidered as the first cde f a class f wmcdes, based n PG (n, 2), all having the prperty that they can be dented with syndrme decding fr the Hamming cde. Of curse, the secnd cde f this class is the cde in sectin 3, the third is (4. 1). These three cdes are ptimal in the sense that w(<3>2)>_3, w(<3>3)>_4, w(< 7 > 4)>7, w(<7>5)>_8, w(<15> 7 )>_15, w(<15>')>_16, all by the RS-bund. The next cde in this class is a < 31 >731 wmcde. The RS-bund yields w (< 31 > 14) >_ 31, w (< 31 > 15) >_ 32. We have cnstructed a < 31 > 10 /31 wmcde, s that sme sequences f length 11 cannt be encded. Hwever, we think that by further selectin f the cnfiguratins it is pssible t cnstruct at least a < 31 > 12 /31 wmcde. An imprtant feature f the cdes described is that messages crrespnd t a cset f a linear errrcrrecting cde with (n-k) x n parity check matrix H. Encding a new message m in a memry with state x c { 0, 1 }n is equivalent t finding a y e { 0, 1 }" such that m =y. HT and y >_ x, cmpnentwise. This is als described in [2], tgether with a dynamic prgramming algrithm fr finding a y with minimal weight. Hwever, ther cnstraints fr y than having minimal weight, can be psed, such as "y must be ne f the cnfiguratins f (3. 1)". In [4] sme cnstraints are described t cnstruct < 2n -k _ l > t/2 n-k - l wmcdes based n the crrespnding Hamming cdes. 5. Extensins Fr mre details f these extensins, see [4]. Extensin 5. 1 : Cnsider again the cde f (4.2). If we allw a 4-th message, represented by the empty set and the whle line, in the first resp. secnd generatin, a < 4 > 2/3 wmcde is btained. This is the wmcde f example 1. 1. Extensin 5.2: Cnsider the cde f sectin 3. If, in the first, third and furth generatin, we allw an cighth message, represented by the empty set, a line r the whle plane respectively, we btain a < 8, 7, 8, 8 )/7 wmcde (rate is 1.69...). Extensin 5. 3 : Adding t the cnfiguratins in the third generatin all thse cnsisting f three nnclinear pints, it is pssible t represent three mre messages, giving an < 8, 7, 11, 8 >17 wmcde (rate is 1.75...). 11 is best pssible here. Extensin 5. 4 : By extending the memry f the cde f sectin 3, respectively example 4. 1 with ne wit, it is pssible t cnstruct an < 8, 14, 11, 8 >/8 (rate is 1.66...) and a < 16 >'/16 (rate is 1.75...) wmcde, respectively. Cnclusin We have cnsidered memries (wms) which cnsist f a number f Write-Once bit psitins (wits). Using prjective gemetries, we have cnstructed cdes, which make it pssible t use these wms several times. It turned ut that the message, represented by the memry-cntent, can be seen as the syndrme f the binary Hamming cde. Acknwledgements We wuld like t thank Henk Hllmann fr his great help. 12301

REFERENCES [1] A. FIAT and A. SHAMIR, Generalized Write-Once Memries, IEEE Transactins n Infrmatin Thery, IT-30, 1984, n 3, May 1984, pp. 470-480. [2] C. HEEGARD, An efficient encder fr ptical disk cdes, preprint. [3] J. H. VAN Lira, Intrductin t cding thery, Springer Verlag, New Yrk, 1982. [4] F. MERKX, Cdes fr Write-Once Memries, Rapprt Interne, Écle Natinale Supérieure des Télécmmunicatins, Paris, 1984. [5] R. L. RIVEST and A. SHAMIR, (1982), Hw t reuse a Write-Once Memry, Infrmatin and Cntrl, 55, 1982, pp. 1-19. 231