NOTES ON RAMANUJAN'S SINGULAR MODULI. Bruce C. Berndt and Heng Huat Chan. 1. Introduction

Similar documents
Bruce C. Berndt, Heng Huat Chan, and Liang Cheng Zhang. 1. Introduction

Some theorems on the explicit evaluations of singular moduli 1

Ramanujan and the Modular j-invariant

SOME MODULAR EQUATIONS IN THE FORM OF SCHLÄFLI 1

HASSE INVARIANTS FOR THE CLAUSEN ELLIPTIC CURVES

ON THE LEAST SIGNIFICANT p ADIC DIGITS OF CERTAIN LUCAS NUMBERS

#A45 INTEGERS 12 (2012) SUPERCONGRUENCES FOR A TRUNCATED HYPERGEOMETRIC SERIES

Congruences modulo 3 for two interesting partitions arising from two theta function identities

Pacific Journal of Mathematics

BOUNDS FOR THE SIZE OF SETS WITH THE PROPERTY D(n) Andrej Dujella University of Zagreb, Croatia

Diophantine Equations and Congruences

RAMANUJAN WEBER CLASS INVARIANT G n AND WATSON S EMPIRICAL PROCESS

MATH 2710: NOTES FOR ANALYSIS

#A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS

Math 4400/6400 Homework #8 solutions. 1. Let P be an odd integer (not necessarily prime). Show that modulo 2,

ON THE DISTRIBUTION OF THE PARTIAL SUM OF EULER S TOTIENT FUNCTION IN RESIDUE CLASSES

CERIAS Tech Report The period of the Bell numbers modulo a prime by Peter Montgomery, Sangil Nahm, Samuel Wagstaff Jr Center for Education

Infinitely Many Quadratic Diophantine Equations Solvable Everywhere Locally, But Not Solvable Globally

Elliptic Curves Spring 2015 Problem Set #1 Due: 02/13/2015

Ramanujan s modular equations and Weber Ramanujan class invariants G n and g n

New weighing matrices and orthogonal designs constructed using two sequences with zero autocorrelation function - a review

Extension of Karamata inequality for generalized inverse trigonometric functions

MATH 361: NUMBER THEORY EIGHTH LECTURE

Representing Integers as the Sum of Two Squares in the Ring Z n

Solvability and Number of Roots of Bi-Quadratic Equations over p adic Fields

Class Numbers and Iwasawa Invariants of Certain Totally Real Number Fields

1. Introduction. = (2n)! (n N)

f(r) = a d n) d + + a0 = 0

A CRITERION FOR POLYNOMIALS TO BE CONGRUENT TO THE PRODUCT OF LINEAR POLYNOMIALS (mod p) ZHI-HONG SUN

When do Fibonacci invertible classes modulo M form a subgroup?

Pacific Journal of Mathematics

Applicable Analysis and Discrete Mathematics available online at HENSEL CODES OF SQUARE ROOTS OF P-ADIC NUMBERS

Additive results for the generalized Drazin inverse in a Banach algebra

1. INTRODUCTION. Fn 2 = F j F j+1 (1.1)

SOME THEOREMS ON THE ROGERS RAMANUJAN CONTINUED FRACTION IN RAMANUJAN S LOST NOTEBOOK

On Character Sums of Binary Quadratic Forms 1 2. Mei-Chu Chang 3. Abstract. We establish character sum bounds of the form.

Quadratic Reciprocity

A p-adic analogue of a formula of Ramanujan

Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various Families of R n Norms and Some Open Problems

Legendre polynomials and Jacobsthal sums

( 1) n q n(3n 1)/2 = (q; q). (1.3)

On the Diophantine Equation x 2 = 4q n 4q m + 9

ERRATA AND SUPPLEMENTARY MATERIAL FOR A FRIENDLY INTRODUCTION TO NUMBER THEORY FOURTH EDITION

x 2 a mod m. has a solution. Theorem 13.2 (Euler s Criterion). Let p be an odd prime. The congruence x 2 1 mod p,

Small Zeros of Quadratic Forms Mod P m

On the irreducibility of a polynomial associated with the Strong Factorial Conjecture

6 Binary Quadratic forms

An Estimate For Heilbronn s Exponential Sum

Zhi-Wei Sun Department of Mathematics, Nanjing University Nanjing , People s Republic of China

Department of Mathematics


p-adic Properties of Lengyel s Numbers

Primes - Problem Sheet 5 - Solutions

Mobius Functions, Legendre Symbols, and Discriminants

ON THE SUPERCONGRUENCE CONJECTURES OF VAN HAMME

RECIPROCITY LAWS JEREMY BOOHER

ARITHMETIC PROGRESSIONS OF POLYGONAL NUMBERS WITH COMMON DIFFERENCE A POLYGONAL NUMBER

ACONJECTUREOFEVANS ON SUMS OF KLOOSTERMAN SUMS

#A47 INTEGERS 15 (2015) QUADRATIC DIOPHANTINE EQUATIONS WITH INFINITELY MANY SOLUTIONS IN POSITIVE INTEGERS

On the expansion of Ramanujan's continued fraction of order sixteen

DISCRIMINANTS IN TOWERS

QUADRATIC RECIPROCITY

Generalizing Gauss s Gem

Cryptanalysis of Pseudorandom Generators

Research Article A Parameter for Ramanujan s Function χ(q): Its Explicit Values and Applications

COMBINATORIAL PROOFS OF RAMANUJAN S 1 ψ 1 SUMMATION AND THE q-gauss SUMMATION

QUADRATIC RECIPROCITY

Arithmetic Consequences of Jacobi s Two-Squares Theorem

An Interesting q-continued Fractions of Ramanujan

A quasi-theta product in Ramanujan s lost notebook

The Hasse Minkowski Theorem Lee Dicker University of Minnesota, REU Summer 2001

ON THE SET a x + b g x (mod p) 1 Introduction

On the Rank of the Elliptic Curve y 2 = x(x p)(x 2)

SOME SUMS OVER IRREDUCIBLE POLYNOMIALS

PARTITIONS AND (2k + 1) CORES. 1. Introduction

When do the Fibonacci invertible classes modulo M form a subgroup?

THE LEAST PRIME QUADRATIC NONRESIDUE IN A PRESCRIBED RESIDUE CLASS MOD 4

PARTITION IDENTITIES AND RAMANUJAN S MODULAR EQUATIONS

Pacific Journal of Mathematics

M. S. Mahadeva Naika, S. Chandankumar and N. P. Suman (Received August 6, 2012)

A CONCRETE EXAMPLE OF PRIME BEHAVIOR IN QUADRATIC FIELDS. 1. Abstract

HENSEL S LEMMA KEITH CONRAD

INCOMPLETE ELLIPTIC INTEGRALS IN RAMANUJAN S LOST NOTEBOOK. 1. Introduction

QUADRATIC RECIPROCITY

16.2. Infinite Series. Introduction. Prerequisites. Learning Outcomes

arxiv: v1 [math.nt] 9 Sep 2015

Piotr Blass. Jerey Lang

AN IMPROVED BABY-STEP-GIANT-STEP METHOD FOR CERTAIN ELLIPTIC CURVES. 1. Introduction

Almost All Palindromes Are Composite

A note on the Lebesgue Radon Nikodym theorem with respect to weighted and twisted p-adic invariant integral on Z p

COMBINATORICS OF RAMANUJAN-SLATER TYPE IDENTITIES. James McLaughlin Department of Mathematics, West Chester University, West Chester, PA 19383, USA

A FEW EQUIVALENCES OF WALL-SUN-SUN PRIME CONJECTURE

Some congruences for Andrews Paule s broken 2-diamond partitions

Nanjing Univ. J. Math. Biquarterly 32(2015), no. 2, NEW SERIES FOR SOME SPECIAL VALUES OF L-FUNCTIONS

2. ETA EVALUATIONS USING WEBER FUNCTIONS. Introduction

MA3H1 TOPICS IN NUMBER THEORY PART III

arxiv: v5 [math.nt] 22 Aug 2013

Math 701: Secant Method

GENERALIZED NORMS INEQUALITIES FOR ABSOLUTE VALUE OPERATORS

MATH 361: NUMBER THEORY ELEVENTH LECTURE

Transcription:

NOTES ON RAMANUJAN'S SINGULAR MODULI Bruce C. Berndt Heng Huat Chan 1. Introduction Singular moduli arise in the calculation of class invariants, so we rst dene the class invariants of Ramanujan Weber. Set 1Y (a q) 1 = (1? aq n ) jqj < 1 n=0 If (q) = (?q q ) 1 : (1.1) q = ex(? n) (1.) where n is a ositive integer, the class invariants G n g n are dened by G n :=?1=4 q?1=4 (q) g n :=?1=4 q?1=4 (?q): (1.3) In the notation of Weber [8], G n =:?1=4 f(?n) g n =:?1=4 f1(?n): As usual, in the theory of ellitic functions, let k := k(q) 0 < k < 1 denote n the modulus. The singular modulus k n is dened by k n := k(e? ) where n is a natural number. Following Ramanujan, set n = k : n The algebraic natures of G n g n n are described in the following theorem of Chan S.{S. Huang [4]. Theorem 1. (a) If n 1 (mod 4) then G n n are units. (b) If n 3 (mod 8) then?1=1 G n n are units. (c) If n 7 (mod 8) then?1=4 G n 4 n are units. (d) If n (mod 4) then g n n are units. Lastly, we dene a modular equation in the sense of Ramanujan. Let F1( 1 1 1 x) be the ordinary or Gaussian hyergeometric function, which, for brevity, we denote by F1(x): Suose that, for some ositive integer n for = k = ` where k ` are moduli, n F1(1? ) F1() = F1(1? ) : (1.4) F1() Tyeset by AMS-TEX 1

BRUCE C. BERNDT AND HENG HUAT CHAN Then a modular equation of degree n is a relation between which is induced by (1.4). We often say that has degree n over : If, as customary, in the theory of ellitic functions, q := ex(? F1(1? )=F1()) =: F () (1.5) then [1,. 14] (q) = 1=6 f(1?)=qg?1=4 (?q) = 1=6 (1?) 1=1 (=q)?1=4 : It follows from (1.1), (1.3), (1.5) that G n = f4 n (1? n )g?1=4 g n = 4 n (1? n )??1=4 : (1.6) Thus, if G n or g n is known, then n can be easily determined from (1.6) by solving a quadratic equation. However, usually, the exression for n that one obtains in this manner is not very attractive does not reect the algebraic nature of n described in Theorem 1. Thus, further algorithms which more readily yield elegant reresentations of n are highly desirable. In his rst notebook [7], Ramanujan oers without roofs over 30 values for n briey indicates some formulas which are useful in calculating n when n is even. The authors L.{C. Zhang [3] roved these formulas of Ramanujan, develoed some new formulas for calculating n when n is odd, consequently established all of Ramanujan's values for n : Ramanujan obscurely describes two further methods for calculating n in his rst notebook [7]. In the rst, Ramanujan indicates that n may be calculated by solving a certain tye of modular equation of degree n: For several rime values of n the desired tye of modular equation exists many of these modular equations can be found in Ramanujan's notebooks are roved in [1]. This very novel method is described in Section, it is the only known method that does not require a riori the value of gn: Thus, the method is a new, valuable tool in the comutation of n: In the second, Ramanujan discloses a method for determining 3n arising from the denition of a modular equation of degree n: In Section 3 we give a rigorous formulation of this formula rove it by using a device the authors Zhang introduced in [] to calculate certain class invariants. Lastly, in Section 4, we follow our methods in Section 3 derive a similar formula for 5n: Although we utilize results of Ramanujan from his notebooks in our roof, this formula was not given by Ramanujan.. The determination of n from modular equations In the middle of age 9 in his rst notebook, Ramanujan claims that, \Changing to 4B=(1 + B) to 1? B we get an equation in 4B(1? B)=(1 + B) the value of B is for e? n :" We now state rove a rigorous formulation of this assertion. Theorem. Let have degree n over suose are related by a modular equation of the form F (() r f(1? )(1? )g r ) = 0 (.1)

NOTES ON RAMANUJAN'S SINGULAR MODULI 3 for some olynomial F some ositive rational number r: If we relace by 1? x by 4x=(1 + x) then (.1) becomes an equation of the form G(x) := g (f4x(1? x)=(1 + x)g r ) = 0 for some olynomial g: Furthermore, x = n Proof. Under the designated substitutions, F (() r f(1? )(1? )g r ) =F =F =0: Hence, the rst art of Theorem follows. Next, setting = 4x=(1 + x) we nd that F1(1? ) F1() F1 = F1 is a root of this equation. (1? x r 4x ) (1 + x) x 1? r 4x(1? x) 1 4x(1? x) 1 + x 4 r 1 + x 1? 4x (1 + x) = 4x (1 + x) 1 r 4x (1 + x) r! F1(1? x ) F1(x (.) ) by a fundamental transformation for F (x) [1,. 93], which actually arises from a secial case of Pfa's transformation. With relaced by 1? x we nd from (1.4) (.) that Therefore, n F1(x ) F1(1? x ) = 1 F1(1? x ) F1(x ) F1(1? x ) F1(x : ) = n: Now recalling the denition of a singular modulus (1.5), we deduce that x = n the roof is comlete. In each examle below we use the following denesting theorem in our calculations. If a? db = c then q r r a + c a? c a b d = : (.3) Examle.1. Let n = 3: Then () 1=4 + f(1? )(1? )g 1=4 = 1 (.4) which is originally due to Legendre was rediscovered by Ramanujan [1,. 30, 3]. With Ramanujan's substitutions, (.4) takes the form u + 1 u = 1 (.5)

4 BRUCE C. BERNDT AND HENG HUAT CHAN where u = 1=4 4x 1? x : (.6) 1 + x Solving (.5), we nd that u = 3? 1: Then solving (.6), we nd that x = 3? 3? + 6: Using two dierent modes of calculation, we nd that 6 = x = (? 3) ( 3? ) = 3 + 6? 3? 3 + 6 + 3 + : It should be remarked that there are two roots of (.6), namely, 6 =3: For if we set = 4x=(1 + x) = 1? x at the beginning of our roof of Theorem, we nd that x still satises (.1). However, in this case, F1(1? x ) F1(x ) = r n : That we have selected the correct root can be veried by a numerical check. The former reresentation for 6 two other reresentations as well, can be found in Ramanujan's rst notebook, they were roved by the authors Zhang in [3, Theorem.1]. Examle.. Let n = 5: Then we have Ramanujan's modular equation of degree 5 [1,. 80], () 1= + f(1? )(1? )g 1= + f16(1? )(1? )g 1=6 = 1: (.7) Using Ramanujan's substituions, we nd that (.7) can be ut in the shae where u + 1 4 u + u = 1 4 u + 3u = 1 (.8) u = 4x 1 1=? x : (.9) 1 + x Solving (.8), we deduce that u =?6 + 10: Next, solving (.9), we nd that x = 3 10? 9? 4 5 + 6 : Lastly, by two distinct routes for calculation, 10 = x = ( 10? 3) (3? ) = 3 10 + 6? 9? 4 5 3 10 + 6 + 9 + 4 5 : The former reresentation was given by Ramanujan in his rst notebook roved by the authors Zhang in [3, Theorem.1]. Examle.3. Let n = 11: The modular equation () 1=4 + f(1? )(1? )g 1=4 + f16(1? )(1? )g 1=1 = 1 (.10)

NOTES ON RAMANUJAN'S SINGULAR MODULI 5 is indeendently due to Schroter Ramanujan [1,. 363{364]. With Ramanujan's substitutions, (.10) can be ut in the form where u + 6u? = 0 (.11) u = 1=4 4x 1? x : (.1) 1 + x Solving (.11), we nd that u =?3 + 11: Then solving (.1), we nd that x = 30 11? 99? 70 + 1 : Hence, = x = (10? 3 11) (3 11? 7 ) = 30 11 + 1? 99? 70 30 11 + 1 + 99 + 70 : In his rst notebook Ramanujan oered the former value, which was roved in [3, Theorem.1]. 3. A formula for 3n Let q be given by (1.), suose that has degree n over : Thus, (1.4) holds. Now suose also that has degree 3 over 1? =: 0 : Then, by (1.4), Hence,, from (3.1) (3.), 3 F1() F1(1? ) = F1(1? ) F1() = n F1(1? ) : (3.1) F1() F1() F1(1? ) = r n 3 (3.) F1(1? ) F1() r n = 3 3 = 3n: (3.3) Next, from [1,. 37], since has degree 3 over 0 we have the arametrizations 3 + 0 = = 3 + (3.4) 1 + 1 + where 0 < < 1: It follows from (3.4) that (1? 0 ) = (1? )3 (1 + ) (1 + ) 3 3 + 1 + = 3 1? (1 + )( + ) 1 + 1 + (3.5) (1? ) 0 = 3 (1? )(1 + )3 + = 1? (1 + ) 1 + 1 + 3 (1 + )( + ) : (3.6) 1 +

6 BRUCE C. BERNDT AND HENG HUAT CHAN Next, set observe that It follows from (3.5){(3.8) that Now, set (1? t) = t = 1? 1 + (3.7) (1 + )( + ) : (3.8) 1 + = (1? 0 ) = 16t 3 (1? t) (3.9) (1? )(1? ) = (1? ) 0 = 16t(1? t) 3 : (3.10) Observe from (3.9), (3.10), (1.6), (3.), (3.3) that k = k = 4t(1? t): (3.11) G 6 n=3 G6 3n?1 : (3.1) We determine (3n) as a function of k: From (3.11), we nd that, with no loss of generality in choosing the minus sign in the rst equality below, t = 1? 1? k Thus, by (3.9) (3.10), we nd that 1? t = 1 + 1? k : (1? 0 ) = (1? 1? k) 3 (1 + 1? k) = k(1? 1? k) (3.13) (1? ) 0 = (1? 1? k)(1 + 1? k) 3 = k(1 + 1? k) : (3.14) Subtracting (3.13) from (3.14), we nd that 0 = 4k 1? k + : (3.15) Substituting (3.15) into (3.14), we deduce that + (4k 1? k? 1) + k(1? 1? k) = 0: Thus, q 3n = = 1? 4k 1? k (4k 1? k? 1)? 4k(1? 1? k) = 1? 4k 1? k (1? k) 1? 4k r 1? 4k 1? = 1? k (1? k)(1? 4k) : (3.16)

NOTES ON RAMANUJAN'S SINGULAR MODULI 7 This last formulation was that given by Ramanujan. We now resolve the sign ambiguity in (3.16). From (3.13) (3.14), it is clear that both 0 satisfy the equation y + y(4k 1? k? 1) + k(1? 1? k) = 0: (3.17) Since 0 = 1? it follows from (3.) (3.3) that the solutions of (3.17) are 3n 3=n: Thus, it suces to show that for n > 1: If then clearly 3=n > 3n (3.18) '(q) = 1X k=?1 q k ' (e? 3=n ) > ' (e? 3n ): From Entry 6 in Chater 17 of Ramanujan's second notebook [1,. 101], it follows that F1(3=n) > F1(3n): Since F1(x) is increasing on (0 1) (3.18) follows. Thus, we have roved the following theorem. Theorem 3. Let q be given by (1.), suose that has degree n over let have the arametrization (3.4), dene t by (3.7). Then, if k is dened by (3.11), 3n has the reresentations given in (3.16), where the minus sign must be chosen. Ramanujan's formulation of Theorem 3 at the bottom of age 310 in his rst notebook is a bit dierent. He rst gives (3.9) (3.10), but with the left sides switched. He then states (3.7), followed by the equality F 3 + = e? 3n 1 + which is a consequence of (3.3) (3.4), where F is dened in (1.5). He concludes by dening k in (3.11) by claiming that a more comlicated, somewhat ambiguous version of the right side of (3.16) equals e? 3n i.e., he forgot to write \F " in front of the right side of (3.16). (In recording secic values of F ( n ) Ramanujan frequently omitted arentheses about the arguments.) Examle 3.1. Let n = 1: Then from Weber's tables [8,. 71], G3 = 1=1 = G1=3 since G n = G1=n [5], [6,. 3]. Hence, by (3.1), k = 1=4 by (3.16), 3 =? 3 : 4 See our aer [3, Theorem 3.] for another roof further references.

8 BRUCE C. BERNDT AND HENG HUAT CHAN Examle 3.. Let n = 5: Then, from Weber's tables [8,. 71], G15 =?1=1 ( 5+ 1) 1=3 : It can also be veried that G5=3 =?1=1 ( 5? 1) 1=3 : Hence, it is easily seen from (3.1) that k = 1=16: Therefore, from (3.16), which is simler than the formula 15 = 16? 7 3? 15 3 15 = 1 16 5? 1! 4 (? 3) (4? 15) which was given by Ramanujan in his second notebook. For its roof further references, see [3, Theorem 3.]. 4. A formula for 5n We roceed as in Section 3 the calculations are easier, however. Let q be given by (1.), suose that has degree n over degree 5 over 1? =: 0 : Then, by the same argument as in Section 3, but with 3 relaced by 5, we nd that F1() F1(1? ) = r n 5 F1(1? ) F1() = 5n: (4.1) Now, from (4.1), (1.6), Entry 14(ii) in Chater 19 of Ramanujan's second notebook [1,. 88], if 0 < < (5 5? 11)= is dened by then Therefore, 1 + = m := F1() F1() (4.) k := (G n=5g5n)?4 = f4(1? )4(1? )g 1=6 =? 1 + : (4.3) Solving (4.3) for we nd that 1? k = 1 + 1 + : (4.4) = 1? k? k? 3k + 1: (4.5) Also, from [1,. 89, Entry 14(iii)], 1? 5n = (1 +? 1 + 1= ) : (4.6) 1 + Hence, using (4.4) (4.5) in (4.6), we nd that, after a modicum of calculation, 5n = 1? 1? k k? k + Thus, we have established the following theorem. 1? k k? 3k + 1 : (4.7)

NOTES ON RAMANUJAN'S SINGULAR MODULI 9 Theorem 4. Let q be given by (1.), let have degree n over dene by (4.), dene k by (4.3). Then 5n has the reresentation (4.7). Examle 4.1. Let n = 1: From Weber's tables [8,. 71], G5 =! 1=4 5 + 1 = G1=5 since G n = G1=n: Thus, from (4.3), k = (3? 5)=: Noting that k? 3k + 1 = 0 we nd from (4.7) that 5 = 1! 3= 5? 1? which can be comared with the value 5 = 1 5? 1! 3 0 @ s 3 + 5 4? s 5? 1 given by Ramanujan in his rst notebook roved by the authors Zhang in [3, Theorem 3.]. Examle 4.. Let n = 5: From Weber's tables [8,. 7], 1 A 4 G5 = 5 + 1 so, since G1 = 1 from (4.3), k = (7? 3 5)=: Thus, q k? 3k + 1 = 14? 6 5 = 3? 5 by an alication of (.3). After a moderate amount of calculation the use of (.3), we nd that (4.7) yields This can be comared with the value 5 = 1 (161? 7 5) 5 = 1 + 6 51=4 (4 5? 9): 0s @ 5 + 5 4? s 1 + 5 4 imrecisely given in Ramanujan's rst notebook established in [3, Theorem 3.]. The authors are very grateful to the referee for very carefully reading our manuscrit, making valuable suggestions, uncovering some errors. 1 A 8

10 BRUCE C. BERNDT AND HENG HUAT CHAN References 1. B. C. Berndt, Ramanujan's Notebooks, Part III, Sringer{Verlag, New York, 1991.. B. C. Berndt, H. H. Chan, L.{C. Zhang, Ramanujan's class invariants, Kronecker's limit formula, modular equations, Trans. Amer. Math. Soc. 349 (1997), 15{173. 3. B. C. Berndt, H. H. Chan, L.{C. Zhang, Ramanujan's singular moduli, The Ramanujan Journal 1 (1997), 53{74. 4. H. H. Chan S.{S. Huang, On the Ramanujan{Gordon{Gollnitz continued fraction, The Ramanujan Journal 1 (1997), 75{90. 5. S. Ramanujan, Modular equations aroximations to, Quart. J. Math. (Oxford) 45 (1914), 350{37. 6. S. Ramanujan, Collected Paers, Chelsea, New York, 196. 7. S. Ramanujan, Notebooks ( volumes), Tata Institute of Fundamental Research, Bombay, 1957. 8. H. Weber, Lehrbuch der Algebra, dritter B, Chelsea, New York, 1961. Deartment of Mathematics, University of Illinois, 1409 West Green Street, Urbana, IL 61801, USA School of Mathematics, Institute for Advanced Study, Princeton, NJ 08540, USA