GEOMETRY OF BINOMIAL COEFFICIENTS. STEPHEN WOLFRAM The Institute jor Advanced Study, Princeton NJ 08540

Similar documents
Dynamics of finite linear cellular automata over Z N

Words generated by cellular automata

Examples of Finite Sequences (finite terms) Examples of Infinite Sequences (infinite terms)

Cellular Automata as Models of Complexity

Decomposition of Pascal s Kernels Mod p s

Regular Sequences. Eric Rowland. School of Computer Science University of Waterloo, Ontario, Canada. September 5 & 8, 2011

McGill University Faculty of Science. Solutions to Practice Final Examination Math 240 Discrete Structures 1. Time: 3 hours Marked out of 60

PUTNAM TRAINING NUMBER THEORY. Exercises 1. Show that the sum of two consecutive primes is never twice a prime.

2x 1 7. A linear congruence in modular arithmetic is an equation of the form. Why is the solution a set of integers rather than a unique integer?


Prime Numbers in Generalized Pascal Triangles

NORTHWESTERN UNIVERSITY Thrusday, Oct 6th, 2011 ANSWERS FALL 2011 NU PUTNAM SELECTION TEST

Numbers and their divisors

X(^+(2^i) = i + X02^r + i.

Module 4: Equations and Inequalities in One Variable

12x + 18y = 50. 2x + v = 12. (x, v) = (6 + k, 2k), k Z.

A PERIODIC APPROACH TO PLANE PARTITION CONGRUENCES

Know the Well-ordering principle: Any set of positive integers which has at least one element contains a smallest element.

New Constructions for De Bruijn Tori

Solutions of Exam Coding Theory (2MMC30), 23 June (1.a) Consider the 4 4 matrices as words in F 16

27 th Annual ARML Scrimmage

NFC ACADEMY COURSE OVERVIEW

Mathematical Modelling Lecture 14 Fractals

CHAPTER 3. Congruences. Congruence: definitions and properties

A MATRIX GENERALIZATION OF A THEOREM OF FINE

Sponsored by: UGA Math Department and UGA Math Club. Team Round / 1 hour / 210 points October 22, 2016 WITH SOLUTIONS

MATH 433 Applied Algebra Lecture 4: Modular arithmetic (continued). Linear congruences.

Sponsored by: UGA Math Department and UGA Math Club. Written test, 25 problems / 90 minutes. October 24, 2015 WITH SOLUTIONS

Advanced Math. ABSOLUTE VALUE - The distance of a number from zero; the positive value of a number. < 2 indexes draw 2 lines down like the symbol>

nonadjacent angles that lie on opposite sides of the transversal and between the other two lines.

Computer Investigation of Difference Sets

Pascal s Triangle: Cellular Automata and Attractors

Miller Objectives Alignment Math

SMSU Mathematics Course Content

History of Mathematics

Unit 3 Vocabulary. An algebraic expression that can contains. variables, numbers and operators (like +, An equation is a math sentence stating

Section 5.1 Polynomial Functions and Models

q-pell Sequences and Two Identities of V. A. Lebesgue

Recommended questions: a-d 4f 5 9a a 27.

Cellular Automata. Introduction

Discrete Math, Spring Solutions to Problems V

Math Requirements for applicants by Innopolis University

Pacing Guide Algebra 1

2x 1 7. A linear congruence in modular arithmetic is an equation of the form. Why is the solution a set of integers rather than a unique integer?

How to write polynomials in standard form How to add, subtract, and multiply polynomials How to use special products to multiply polynomials

STUDY GUIDE Math 20. To accompany Intermediate Algebra for College Students By Robert Blitzer, Third Edition

YOU CAN BACK SUBSTITUTE TO ANY OF THE PREVIOUS EQUATIONS

Scott%County%Public%Schools%

MATH 433 Applied Algebra Lecture 22: Review for Exam 2.

Unit 2, Ongoing Activity, Little Black Book of Algebra II Properties

Name (please print) Mathematics Final Examination December 14, 2005 I. (4)

Pascal s Triangle. Jean-Romain Roy. February, 2013

The cocycle lattice of binary matroids

MATH The Derivative as a Function - Section 3.2. The derivative of f is the function. f x h f x. f x lim

August 2018 ALGEBRA 1

Ch1 Algebra and functions. Ch 2 Sine and Cosine rule. Ch 10 Integration. Ch 9. Ch 3 Exponentials and Logarithms. Trigonometric.

Fractals. Justin Stevens. Lecture 12. Justin Stevens Fractals (Lecture 12) 1 / 14

Continuing discussion of CRC s, especially looking at two-bit errors

The Cups and Stones Counting Problem, The Sierpinski Gasket, Cellular Automata, Fractals and Pascal s Triangle

Cellular Automaton Growth on # : Theorems, Examples, and Problems

Math 10-C Polynomials Concept Sheets

UNCORRECTED SAMPLE PAGES. Extension 1. The binomial expansion and. Digital resources for this chapter

(Primes and) Squares modulo p

6.1 Using Properties of Exponents 1. Use properties of exponents to evaluate and simplify expressions involving powers. Product of Powers Property

Unit 1 Vocabulary. A function that contains 1 or more or terms. The variables may be to any non-negative power.

SPLITTING FIELDS AND PERIODS OF FIBONACCI SEQUENCES MODULO PRIMES

Daily Skill Builders:

1 2 3 style total. Circle the correct answer; no explanation is required. Each problem in this section counts 5 points.

Math "Matrix Approach to Solving Systems" Bibiana Lopez. November Crafton Hills College. (CHC) 6.3 November / 25

Definition 6.1 (p.277) A positive integer n is prime when n > 1 and the only positive divisors are 1 and n. Alternatively

MATH 433 Applied Algebra Lecture 21: Linear codes (continued). Classification of groups.

MATH98 Intermediate Algebra Practice Test Form B

How do we analyze, evaluate, solve, and graph quadratic functions?

Algebra. Table of Contents

MCR3U Unit 7 Lesson Notes

20A. Build. Build and add. Build a rectangle and find the area (product). l e s s o n p r a c t i c e 1. X X X 2 + 6X X

Number, Number Sense, and Operations Data Analysis and Probability

CHINO VALLEY UNIFIED SCHOOL DISTRICT INSTRUCTIONAL GUIDE ALGEBRA II

Check boxes of Edited Copy of Sp Topics (was 261-pilot)

The Sierpiński Sieve of Nim-varieties and Binomial Coefficients

Aviezri Fraenkel Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot, Israel 76100

Int Math 3 Midterm Review Handout (Modules 5-7)

MATH98 Intermediate Algebra Practice Test Form A

THE CUBIC PUBLIC-KEY TRANSFORMATION*

CURRICULUM UNIT MAP 1 ST QUARTER. COURSE TITLE: Applied Algebra 1 GRADE: 9

A connection between number theory and linear algebra

Math 75 Mini-Mod Due Dates Spring 2016

Area Formulas. Linear

Integrated Math II Performance Level Descriptors

Final Exam Review MAT-031 (Algebra A) Spring 2013

Intermediate Algebra

CHARACTERISTIC AND MINIMAL POLYNOMIALS OF LINEAR CELLULAR AUTOMATA

Algebra I Unit Report Summary

Algorithms CMSC Homework set #1 due January 14, 2015

Sierpinski's Triangle and the Prouhet-Thue-Morse Word DAVID CALLAN. Department of Statistics University Ave. November 18, 2006.

DIVISIBILITY OF BINOMIAL COEFFICIENTS AT p = 2

SERIES

College Algebra with Corequisite Support: Targeted Review

A technique for DDA seed shifting and scaling

Latest Syllabus - NMO

Transcription:

Reprinted from the AMERICAN MATHEMATICAL MONTHLY Vol. 91, No.9, November 1984 GEOMETRY OF BINOMIAL COEFFICIENTS STEPHEN WOLFRAM The Institute jor Advanced Study, Princeton NJ 08540 This note describes the geometrical pattern of zeroes and ones obtained by reducing modulo two each element of Pascal's triangle formed from binomial coefficients. When an infinite number of rows of Pascal's triangle are included, the limiting pattern is found to be "self-similar," and is characterized by a "fractal dimension" log 2 3. Analysis of the pattern provides a simple derivation of the result that the number of even binomial coefficients in the nth row of Pascal's triangle is 2#,(n), where #l(n) is a function which gives the number of occurrences of the digit 1 in the binary representation of the integer n. Pascal's triangle modulo two appears in the analysis of the structures generated by the

1984] NOTES 567 evolution of a class of systems known as "cellular automata" (See [1], [2], [3] for further details and references.) These systems have been investigated as simple mathematical models for natural processes (such as snowflake growth) which exhibit the phenomenon of "self organization." The self-similarity of the patterns discussed below leads to self-similarity in the natural structures generated. 2 3 3 t -- 4 6 4 t t 5 to to 5 t 6 45 20 t5 6 t 7 21 35 35 2t 7 FIG.!. The first few lines of Pascal's triangle modulo two. FIG. 2. The first sixty-four lines of Pascal's triangle modulo two (black squares indicate ones, white squares indicate zeroes). Fig. 1 shows the first few rows of Pascal's triangle, together with the figure obtained by reducing each element 'modulo two, and indicating ones by black squares and zeroes by white (blank) squares. Fig. 2 gives sixty-four rows of Pascal's triangle reduced modulo two. A regular pattern of inverted triangles with various sizes differing by powers of two is clear. Large inverted triangles spanning the whole of Pascal's triangle begin at rows n = 2J. Consider the pattern down to the beginning of one such large inverted triangle (say down to the sixty-third row). A striking feature of the pattern is that the largest upright triangle contains three smaller triangles whose contents are similar (except at the scale of very small triangles) to those of the largest triangle, but reduced in size by a factor of two. Inspection of each of these three smaller triangles reveals that each is built from three still smaller similar triangles. This "self similarity" continues down to the smallest triangles. At each stage, one upright triangle from the pattern could be magnified by one or more factors of two to obtain essentially the complete pattern. The pattern obtained differs from the original complete pattern at the scale of very small triangles. If, however, Pascal's triangle were extended to an infinite number of rows, then for all finite triangles this effect would disappear, and the original and magnified patterns would be identical. In fact, triangles of any size could be reproduced by taking smaller triangles and then magnifying them. The limiting pattern

568 STEPHEN WOLFRAM [November obtained from Pascal's triangle modulo two is thus" self similar" or "scale invariant," and may be considered to exhibit the same structure at all length scales. Many examples of other "self similar" figures are given in [4], [5]. If the number of inverted triangles with base length i is denoted 1';, then Fig. 2 indicates that 1';/ 2 = 31';. For large i, therefore (1 ) The exponent log23 :: 1.59 appearing here gives the "fractal dimensionality" [4], [5] of the self-similar pattern. Consider a (" filled in") square. Reduce the square by a factor of two in each of its linear dimensions. Four copies of the resulting reduced square are then required to cover the original square. Alternatively, one may write that the number of squares S; with side length i contained in the original square satisfies S, /2 = 4S;, so that S; - i - 2. The exponent two here gives the usual dimensionality of the square. One may then by analogy identify the exponent :: 1.59 in Equation (1) as the generalized or "fractal" dimension of the figure formed from Pascal's triangle modulo two. Fig. 2 suggests that the number N (n) of ones in the nth row of Pascal's triangle modulo two (or, equivalently, the number of odd binomial coefficients of the form (';) is a highly irregular function of n. However, when n is of the form 2;, the simple result N (2;) = 2 is obtained. This can be considered a consequence of the algebraic relation ( ~ J ) = 0 mod p for 0 < i < pi and all primes p, which may be proved by considering the base p representations of factorials. Algebraic methods [6]-[12] have been used to obtained the general result (2) The function #1 (n) gives the number of occurrences of the digit 1 in the binary representation of the integer n. Hence, for example, #1(1) = I, #1(2) = #1(10 2 ) = I, #1(3) = #1(11 2) = 2, #1 (4) = 1) and so on. A graph of #1 (n) for n up to 128 is given in Fig. 3. Note that although the function is defined only for integer n, values at successive integers have been joined by straight Ii nes on the graph. For n > 0, 1 ~ # I ( n) ~ r log2 11 1. The lower bound is reached when n is of the form 2i; the upper one when n = 2' - 1. Clearly #1(2i n) = #dn) (since multiplication by 2i simply appends zeroes, not affecting the number of 1 digits), and for n < 2i, #I(n + 2i) = #1(11) + 1 (since the addition of 2' in this case prepends a single I, without affecting the remaining digits). The result (2) for N (n) may be obtained by consideration of the geometrical pattern of Fig. 2, continued for 2f log,,, 1 rows, so as to include the complete upright triangle containing the nth row. By construction, the nth row corresponds to a line which crosses the lower half of the largest upright triangle. Each successive digit in the binary decomposition of n determines whether the line crosses the upper (0) or lower (1) halves of successively smaller upright triangles. The upper halves always contain one upright triangle smaller by a factor two; the lower halves contain two such smaller triangles. The total number of triangles crossed by the line corresponding to the nth row is thus multiplied by a factor of two each time the lower half is chosen. The total number of ones in the nth row is therefore a product of the factors of two associated with each 1 digit in the binary representation of n, as given by Equation (2). There are several possible extensions and generalizations of the results discussed above. One may consider Pascal's triangle reduced modulo some arbitrary integer k. Fig. 4 shows the resulting patterns for a few values of k. In all cases, a self-similar pattern is obtained when sufficiently many rows are included. For k prime, a very regular pattern is found, with fractal dimension (k+l) k Dk = 10gkL i = 1 + logk - 2-,, - I

1984] NOTES 569 8~------------~1--------------'-1------' 6-4... r' - 2 - O~ ~I ~I ~ o n 128 FIG. 3. The number of ones in the binary representation of the integer II. so that D) = 1 + log) 2 :::: 1.631, Ds :::: 1.683, and so on. In general, for large k, one finds that Dk - 2-1jlog2k; when k -+ 00, the elements of Pascal's triangle modulo k become ordinary integers, which are all nonzero by virtue of the nonzero values of binomial coefficients. By a simple generalization of Equation (2), the number of entries with value r in the nth row of Pascal's triangle modulo k is found to be N(r)(n) = 2#~kl(n), where now #!k)(n) gives the number of occurrences of the digit r in the base-k representation of the integer n. One may also consider the generalization of Pascal's triangle to a three-dimensional pyramid of trinomial coefficients. Successive rows in the triangle are generalized to planes in the pyramid, with each plane carrying a square grid of integers. The apex of the pyramid is formed from a single 1. In each successive plane, the integer at each grid point is the sum of the integers at the four neighbouring grid points in the preceding plane. When the integers in the resulting three-dimensional array are reduced modulo k, a self-similar pattern is again obtained. With k = 2, the fractal dimension of the pattern is log25 :::: 2.32. In general, the pattern obtained from the d-dimensional generalization of Pascal's triangle, reduced modulo two, has fractal dimension log2(2d+ 1). References 1. S. Wolfram, Statistical mechanics of cellular automata, Rev. Modem Phys., 55 (1983) 60l. 2. O. Martin, A. Odlyzko and S. Wolfram. Algebraic properties of cellular automata, Bell Laboratories report (January 1983). to be published in Comm. Math. Phys.

VI C3 k = 5 ~..:::::....4:"""...-X....»..-;;.:.....,...::: ~.. ~...,-y.... W... /k....-x.-...-:jf....-x.-... «<::....;;...;......,('».....do.-..... -x.-......»... -x.-... -x.-... -x.-..:. ~ ~ ~.-x.-.....-x....;;.«""" #...&.. ~.-;"".:::::..... ~..:::::.. /»...... /k. h...;........... k=9 k - 10 FIG. 4. Patterns obtained by reducing Pascal's triangle modulo k for several values of k. White squares indicate zeroes; progressively blacker squares indicate increasing values, up to k - 1.

NOTES 571 3. S. Wolfram, Universality and complexity in cellular automata, Institute for Advanced Study preprint (May 1983), to be published in Phys. D. 4. B. Mandelbrot, Fractals: Form, Chance and Dimension, Freeman, 1977. 5. B. Mandelbrot, The Fractal Geometry of Nature, Freeman, 1982. 6. J. W. L. Glaisher, On the residue of a binomial theorem coefficient with respect to a prime modulus, Quart. J. Math., 30 (1899) 150. 7. N. J. Fine, Binomial coefficients modulo a prime, this MONTHLY, 54 (1947) 589. 8. S. H. Kimball et al., Odd binomial coefficients, this MONTHLY, 65 (1958) 368. 9. J. B. Roberts, On binomial coefficient residues, Canad. J. Math., 9 (1957) 363. 10. R. Honsberger, Mathematical Gems II, Dolciani Math. Expositions, Mathematical Association of America. 1976, p. 1. n. K. B. Stolarsky, Power and exponential sums of digital sums related to binomial coefficient parity. SIAM 1. Appl. Math., 32 (1977) 717. 12. M. D. McIlroy, The numbers of l's in binary int~gers: bounds and extremal properties, SIAM J. Comput.. 3 (1974) 255..