Berndt Ramanujan
Transcription
Berndt Ramanujan
R A A B C C B C I C N R H. U EH N d r e i L K K t Op e n a a , C E sq 0 mn n L a & L ,d ’ w t m k t < 1 S w ; cn i , kh : o d , ?t . u ’ s p , a t o l i o e ( S a n t ( ro o n r c F n u> ’r = p m tpeao i i pit d w l e o t i oa i r . 1 2 = m o t n r q = c - x s a l = k e a ,K t 7 a : s&rd s a m ie si e s e m l t = t , bb f o e cl ou i a uo u rle nrde e r L u hn s n csg n ep s ih s s so e d ep e - it ee )s u es t in l re vc n u h i rt u l d i t aa ed t c u o - t. , ts( od r fs r ( r 1e l o f . l i i euU l o n in asl s l d s g pu u e u s g t t c i ,e r c v ri s e u os a o n t t a ’ , e ccs b ( p ) esr , n t = a Lt ’ oI fn ) N A d o 0 <: p ) x 3 k e= , ro i c sl hi nt e n e gR c CI T ti s e n f i sm C N fh lc wl f l gi k ol fa h f L k = P o S th p k = l, wKO =ha Lp ’a 1 h d . a 2 o L ( t e di ct u t c e c , t :h , d , a U S o .q : R f D DT A H 1 a! A nu nanp r r n l o t mn f amn I sn ui a p t e. A G % bn ( n i c h f f l s h ob i w f d hf e A AT k L’ x N NA E ~Z z od i t % ’e1 =a d mt h p 1 T p od d H N N S I o & l q n 1A d fu yd a1m d s a e . o u c gl ui t. c u l e o n i t r c p y t t u n ( b e e, K’ K f op r c V RN n o nt h iu A A B A RU C L . c u s N O FD 1 I T M L 2n m w 1 d l i 2 j . t o [ a pa 3 l W s l i .m 7 l . A = ( q4 I C e, n a1o - 2) ( T t q b & , ) m: w q y l i c y + ) p i j c S e n 2 B C B T 1 p 1 R h H. , . 1 a n ( 1 x o d x F q= e S m ( t w - G h c c W ( S m W r n b f t a h s n R 2 M H f [ [ p [ t f e i h o s t p [ c c t l [ b o g ha n n 1 ao ci i u 1s i 2R f s e 4 .= 2 =, 2 7t a an a t o ei o [ a r l 2h iI o ts r t f i h m ei a1 a t Mna J e[ oR ny bi c 9 [ h i e RAU C L H N n ( - c5 . NA E I Z A G N DT A HN D N A> T GN 1 1 q X) i ) ) 6 / { q o (- c rt cc p a qn n. 6 l i h r de r ,j-11*4 - # w l uc ) l lf ( n fe i o q ui Cu aq e p g iea fn t dW h m ac R o e m nh i G mi El n uf ? aAr Gl N d7W , tsc [ .g r. ) a i ~ m b e l sn , e t(t o 1 “ e er a ) ( r) sc a dar t bli l F 7 b m e dni i m * es n l i, r to s o] r td s a re r. o l a o d e n du Go f e s o v n oo t nt d a a g f free o, d nv l n oo nvr m da u rf . em a l e ” ni r u s u et n 2 s , b, o 3 : e2 f he at ] =, s’ ( er : na 2 f i T f = dt ’ l 4 ih ii ( G )e o i ot u 1 b sRt h s . iy haa af e 7p n i m t [ta [d ) a p s2 i al m 1 p 3oln o 3 e. - n] u u , [ n CE d h t 2 x ( E HU nm n t i h h ospc o snp i i tecl[ i Pd B . B . d n n a. u . a o dt c tl l w rd S a 4c w e eH o O s e] ie m 1 dt a o 2 G eei m [ i n [i ow m h d1s th o n o h s rl no t i e es ot e a a p a m n r l o ac nn i e [ 9 R o, s c 1 l ta f o aso s s arol nd r o ei h s ti a o b e ,n tl u t w et i fn] n tfbn as1 s a o, t i . e2i [ e i i ft s 9 fun ] s c t tb o ui o c f n l O m np a t o 2o i 7 e m c o in n pn oh w m it l h s i e h hi ouna1 s th 1 bil sh h T i,s c o n d l 2 au n v ba ] f s 7 b la t nd i ti yG n oi ,h c te e n ie e] s e ar i pn . i t s v a h s e n e o s um ] a i ke j e 1 one i a , ne ] o s t bo d ok i l 9 sj a eg g l 4 sa : t o an a v o su e s f a h o n s ac s e t d tt rv c s h h hf ec na e e i i[r r ot r s r1o a t s s2u l e, 2 1 c os i p 1 1 l f n n ei h ]sao t vnn r i e sua a a2o s c s nn i r5t o ddn i e ln n g r i ha o s iv t nn is t e vaAh o ss Rc cr le t ao ni tr , m n ah e a c i 2e l lt nw ]v pc ah vcde ru s a b arW er el s t y rctea va i c bl it h i n p 3, eb sn i a ]a a tw o n pW :m s t, i t eI c at a t e n rb nw b , r c uo on] i r R ao, vv, n ps a su ea a m s n dr p a d W at s1 p 1 p 2 co7 ri 8 e 4 lnf] ui h o se l nt n ti ec ]H o sc co , of nt = 3x oa r w o3 c 2c ndn re6 b 5e ded v3 o i e w 2e nr m b c R vc a ay r s a ac i p t c c im ort n l an a ih r i a t nade n ho f p r n [i t i W a s e 1s e r a a p “t m 7 b s t p n e c p ]t o t , r r m l : o o s o p o k i 4 f h Tt n p w eh , a v ru b aWai l a o s [ iy sp abs s r c e1 n r tl o i e d6 o si as ] hG o f o i c I[ r a o c r n l~ m n u bo R lS u d n [ym ap ]x~ l d 1 m m . x, a 3 u a i l n i t e G H eH s o c I . t . tsa opn h [t e r W na ee 1e c d a dp m 8r o y t e p] n. o t e q d t or s u thu i e t a f e el n m i aR t o na v na b a i u r a i l mo n s r r ni p r t s a il h s e Iet t o ] i s c nr h t t, n [ oW eeo t 1 n a b r7 d t o o b u t an po nn l i “ i ic It ps nm t cto u t a hi a b ie r t cnc l l nn k h o vo c i d s c f on u s e ls s N + Q h( ei tea tn i b hp Rv n o ep p , a [ at e ta a r 1 nh a hp mi 1 d r ee a n e a p o o t cc s a t f h o oot nr c e n nm hb Dot bB M Wdcym rt, o . . i e .e o l n - k m A n y W l da W s a te n cti e t Rhf d lol ds n aof f s io o m u o . o t n t rga ”n h o i t [ b p 2 a i n i an ei h t r l v i 1nn f t t b p vw c aen Th e i eC tn rcC s b o , Wi ae r w l an mr a l o t s o v Io Ws h at a f l ac c e c r m t ta r t i hd s l l h b i p lao T i a c2 s v eorr oah tn ae1 aa6sik 6 f se rn l r5ta 9 t i bv r c r s c l a e oc . e u s,.n , RAMANUJAN’S CLASS INVARIANTS AND CUBIC CONTINUED FRACTION 3 77, 81, 117> 141, 145, 147, 153, 205, 213; 217, 265, 289, 301: 441, 445, 505, 553> 90, and 198. After establishing calculatcd 26 of Ramanujan’s invariants several other class invariants in [16], [17], and [18], Watson in four furthcr papers [19], [20]: [21], and [22]. Thc values of n considered by Watson depend upon the class numbers for positive definite quadratic forms of discriminant -n. In the course of his evaluations, he determined the class invariants for n = 81 [21], 147 [21], and 289 [22]. Thus, 18 class invariants from the list of 21 bave heretofore not# bcen proved. In this paper> we shall establish 441, 90, and 198. prove the remaining R,amanujan’s Not#e that for each su& 13 class invariants. class invariants for n = 117? 153, n; 9[n. In anothcr paper [3], we shall Our starting point is a relation connecting gn and ggn, found on page 318 of Ramanujan’s first notebook> but not in his second notebook. K. G. Ramanathan [9], [lO] noticed this relation in the first notebook, but apparently he never gave a proof. notebook: J. M. and P. B. Borwcin explicitly, derived a formula Also unaware of its appearance in the first [4! pp. 145, 1491, although not stating the results connc&ng gr, and ggn,, as well as a formula relating They [4, pp. 145: 1461 also derive formulas connecting Gsln with GTZ G,) and Gg?,. and Ggn, and gslTL with g?, and gg!,. In t,he ncxt section. we use one of Ramanujan’s modular cquations of degrcc 3 t,o establish t,he aforemcnt,ioned formulas connecting G,, and GgTZ; and .9?, and gg??. The former formula is net, found in thc notebooks, but it cari be proved along t,he same lines as thc latter. Of course, the ideas in this papcr cari be utilizcd to establish ot,hcr class invariants found by Ramanujan when gin, in particular> for n = 27,45.63,81,225,333,765,18, 126,522, and 630. Undoubtedly, somc of these proofs would be simpler than previous proofs? for cxample, for 81; 225> 333; 7G5, 126, 522: and 630. In particular, Watson’s proofs for n = 333,765,522, and 630 [17] were based on an “empirical method.” In fact: the Borweins [4, pp. 147> 149, 1501 employed the aforcmentioncd formulas to calculate the invariants GZT: Gsl: Gz2~ T and g522. Ramanujan’s cubic continued fra&ion G(q), first introduced by him in his second letter to Hardy [13> p. xxvii]; is dcfined for 1~1< 1 by y2 + y4 yv3 G(y) := 1 + + + T + q + . . .. Chan [6] has proved several elcgant theorems for G(y)> many of which are analogues of well-known propcrtics satisfied by t,he Rogers-R,amanujan continued fraction F(y) := z--1 +ï+T+T+.., li5 y y2 y3 In his first letter to Hardy. Ramamljan asserted that F(~-~fi) “cari be exact,ly folmd if n be any positive rational quantity.” Alt~hough this claim is correct, and the values are algcbraic. it is generally quitc diffîc~ll~~to cvalllate F(emXfi) for specific values of n. Ste papcrs of R,amanathan [7], [8]. [9], [lO] and Berndt and Chan [2] for specific evaluations. In contra& to t,his situation, wc prove in Section 4 general formulas for G(--emnfi) and G(eerfi) t,hat arise from connecting Gr,, and Ggn, and gT2and ggTZ,respectively. WC cmphasize formulas are known for F(eerfi) and F(-~-~fi). the formulas that no such 4 BRUCE C. BERNDT. 2. Formulas Theorem Proof. HENG HUAT CHAN, AND LIANG-CHENG Connecting ZHANG GTz and Gg?,, and gn and ggn 1. Let For brevity; we set G = GT, throngho~~t, thc proof. We shall employ Ent,ry 19(xii) [ll], [l, p. 2311. Let (2.3) P = {16a$?(l in Chapter 19 of Ramanujan’s Q= an d - a!)(1 - /?)}“’ second notebook (f;;;+;)1’4> where ,0 bas degree 3. Then Q+-$++=-$) (2.4) =O. Ntow from (1.5) and (1.7), when q = exp(--~fi)> G =Y GrI = {4a(1 - ti)}-1’24 GgT, = {4B1 - /?)}-1’24. an d Hc!nce, by (2.3); when q = exp(-Tfi), from (2.4). (2.5) (G/GsT)~ Set z = (G9JGj3. + (G/&JG i- 2fi ((GGsJ-~ Th us, (2.5) cari bc rewritten .x’ - C2.6) R,earranging form F = (GGgr,)-3 2fiGGz3 and using t,hc not,ation (z2 - fiG% + 2v+G-% and Q = (G/GgfTj6. - (GGA3) Thus, = 0. in the form + 1 = 0. (2.1), we find that we cari recast (2.6) in the +JI)~ = 2@* - 1) ( G* +)2. RAMANUcJAN’S CLASS INVARIANTS AND CUBIC CONTINUED FRACTION 5 Since z > l> or Noting that CE> 0 and solving for x: WC find that New, Tltus, squaring and expanding, we find that and G4(G8+2G4@=ï+~2-l) =; (P+@-z); -2~,-2dm (P~-2+14c-i) + (Il- - +~(~+~~)~~2-1~-2~-2~~ (2.10) Using (2.9) and (2.10) in (2.7), we deducc X +=/(P (2.11) + @Y) (P + @-Y) (g t,hat - 2 + J(J)2 - l)($ (g - 4 + &2 - l)(p2 - 4)). Recalling that x = (GgTI/GJ3, wc sec t,hat (2.11) is cquivalent proof is complete. We next prove thc aforememioned notebook. - 4)) to (2.2), and SO the result fomld on page 318 in R,amanujan’s first BRUCE 6 Theorem 2. C. BERNDT, HENG HUAT CHAN. AND LIANG-CHENG ZHANG Let Tlt,en YJn (2,13) Proof. , a Set 9 = ,9n, throughout thc proof. Using (1.7) and (1.4)> wc rewrite (2.5) in the form Using (1.7) and (1.4) again; we find that -(.9/~7? S&ing Rccalling form x = (9gn/,9)3, + (.99~,/.9)~ WC dedllce the notation (X2 - (2.12): h.9% 2~+(.99~Tz)-~ - 2~&~9gn)~ = 0. that we sec that the last cquation cari be rewritten in the - j))2 = 2(J12 + 1) (. 92 1 + JJ2. It shollld now bc clear that thc rcmainder of t,hc proof is complctcly th<ritfor Theorem 1, and SO wc omit thc rcst of t,he proof. analogous to The cube roots in (2.2) and (2.13) are not vcry at,tra&ve, and l~s~~allyin applications R,amanujan folmd more appcaling expressions for t,hese cube roots. R,amanujati had an amazing ability for denesting and simplifying radicals: and undoubt,edly wc do not havc thc insights into radicals t,hat R,amanujan had. Howevcr, it seems quite likely that in sevcral instances Ramamljan llscd thc following elementary resuit from Carr’s book [5! p. 521. S ince Carr docs not give a proof and sincc hc adds thc extraneous here. hypothesis Qlat, (a* - !>)1j3 bc a Perfect cube, we provide a proof R AC L 3 S e I ML . c :’ t m ( 2 (. ( 2 . 4 n ( F . = La1+ & - 3 ( r w re YJ = 2 so ( w s ( ~c = z b s u ( o - t t R i o u d n o p a ep n W a s , 3 J z AT I rj o i 2 ) & 3 5 e I t s t f C A i 3s n CI A t ) ) . co vq1’3 a 2 = d - a = z 2 ( :n f sn d ; t i 4. l ) y i J , î2 7 to 2 ,c r d s ) , ‘ f . hon Woed f n . t o1 3 ( ap &) + 3 1 e f c 1 ef - = xa12 ta m s .rl? - ? (. - & b( f , ) ,d x 8 i 3 Cls i g1 h r 4 , 7 ) 2 . e o u . c vn 7 a ) f h e , k L a m a3 i e s e t p m nl i Hm s Ro o s o a a t tt sd b am o si r n t s rr uo he c o e pt c al r c s u a i , l m ng n h e db l v e s T e e r 1 a h 2 t c i o w n c b h a t mo d , u h c a h t 1 ( 2 . T h e a n d ( 2 a 1 . l ab y m o m i la eswe ohio a tc i c t cl o o o e at d f s a1 n mc 1 ) . 5 e )+ 0 +* ) r 8 t3 2 2 ) 1a ) >e e qi b’ t a i a h u e c t s 1 u c v y an s 1 m m n = ) t o ne pwa y uc l c l gn csel i e u es a nn en l mb es rs g c e d rc l r r ot vi 9 i d ~ i i bs t s t s ( c u ob o t 2os ai y lr .frt lni f v i e 1o l sf o ea x 4r y ,p m a i u et rv s so 1 tSh o da ef m hO e fb al d oi ot rt cn u ? s dh ca uoc w o o h s e r ot s a o i ot i t u g m g i p h n l R o n 5 ra p a o c s ] rs ro m mk c o u s mt nte r C b i [ w a ’ if i aas S l m d B U S R 7N R 6 1 tO hp . 5 = z I -t (i w U C r c C. 2 ai ( a 2 B t s x+4 1 Ci? - & e oa - e7 eS tl 2 C U NS F O DA l =1 a z c S C NV d 2 P C AA T o = mow c ) aw p h a s e a ‘ t p r - bh I ( a AN l, +- t e nd B 8 C B R H. E HU S i n ( 2 . S t a s O h ( (f qt 2 A t t n ( rs e h ia( o 2 ( U T 4 P j N w sd L 2 h i ( J = f y G 117 ( g h p a d 2 3 a x 3 C F t h c. h 2 23 ea s .I i 114 ( + aa , l l+ 2 g e e o3 n o. t r e e e[ ] N A. is , v a ) e GN h J) l. . . v et h pf 7 ro r cn sc a n b og ) s e 2 v 2 s d s , ) r n r) o o b2 R . , s 4 ac pn m ) , a ) e 1 s e ’ n = g2 r l 0p vg ui e vi s /3 4 T )2 n2 c A HN re el m w o f rn te o o1 3 nF ht W p l ia d pi r l ( ne t h f2 c 2 r a 2) m 2( + )1? e r b i. DDT ) ak 2 la u - 3( = N ( . l ) ua L ? eW i e t 2T u >b c = 2 ts . Te a 3 1 L r 63 B 1 Zl,= a, + - 2. C n k2 . n = r1 [ a p e a t es NN A E I Z A G c (u . - I RAU C L H N CE i r m c2 m r r e3, a1 e ra, t1 m = . h as f y . 2 i { d/ o 1 ( p, i r v /B y iC h a f 1n n) i g i[ 4oR n s er1 r ta wi i bs hcv s ot t Rivc vr h u cai n I ta o r t hv t sn e t ec o he mh r f) 2p y e 5 FZ , . l G c + }+ G s2 ei etm f r t ] nc h ba o T eo a t 1hm d a wh ,n t i eas o. aa e )w it f 2i t )c 1 r , o eo n r ee g l nn n o ~ o RAMANUJAN’S CLASS INVARlANTS AND CUBIC CONTINUED FRACTION 9 We apply Lemma 3 with a, = Thus, c = 1. We thcrefore 11+G& 9+Gfi 2 * need to solve 4x3 - 3z = TO solvc t,his by inspection, bx and 2 it pcrhaps 11+M 2 J . is bcst, t,o square both sidcs and set x = t2. T~E, i(4i-3)2= It is not difficlllt to see that 1 = 1+ fi/4. ll+G& 2 . Hence, 1 = $ dm, y = fi/4. Th us, (3.2) follows> and the proof is complcte. Alternatively? in the notation (2.19) and (2.21); by (2.22); and> from (2.lG), we want to solve UZ(UP - 3)2 = 22 + 12& (3.3) Factoring in Z [ A/?]: WC dedllce from (2.23) that, N(,UZ)NP(,UP - 3) = N(22 + 12v5) = 2z . 13. Then zt2 = N(U2 - 3) =: ïV(A + Bd?) Choose = A2 - 3B2. A = 1 = B. We now observe that ïv(U2) = AJ(4 + fi) as required. It is eaily chccked = 13, that,, whcn u = dm; (3.3) holds. Wc then readily find that Q<=;(3++&zz). Thus, (3.2) has been shown once again. Theorem 5. (3.4) (/?+/y2 G53 Proof. Set 7~= 17 in Thcorcm paper [ll], (/y+/z-q’3. = 1. By Weber’s trcatisc [23, p. 7211: or Ramanujan’s 10 BRUCE C. BERNDT, HENG HUAT CHAN? AND LIANG-CHENG Gk = 5 + &ï? 4 + (1+ ZHANG 67J3’2 4fi ; from (2.1) WC find that Also, T~US, from Theorem 1; b 17+5Jï?+ X (19+5dïq(l3+5dE) 4 113 13+50+ (19+5dE)(l3+5mJ + 4 I However> note that (19 + 5m)(13 G 153 + 5dï?‘) = G72 + MOdï? = (20 + 4fi? = C(bmparing the equality above with (3.4); by (3.5), we sce that it, rcmains to show that (3.6) 5+&7+ 2 d 19+5mxG6. which is rather curions indeed. Note that 2 lÏ, RAMANUcJAN’S CLASS INVARIANTS AND CUBIC CONTINUED FRACTION 11 Thus, by (3.6), it suffices to show that In the notation Factoring (2.19) , we dedllce that [ N(uZ)N2(u2Wc attempt WC solve 1 1+&? 2 in Z and (2.21): 3) = ïV(7 + fi) = 25. to salve *4=iv(uzZ-3) =: N A+B -A2+AB-4B2. Take A = -1 and B = 1. Then u2 = (5 + m)/2 and N(U~) calculation shows t,hat (3.8) indccd holds. Lastly, we find that By (3.5), we conclude Theorem = 2. A simple that (3.7) holds to completfe the proof. 6. (3.9) Proof. We apply Theorem R,amanujan’s papcr G49 = lhxE+ (3.10) It is easily checked 1 with n, = 49. From Weber’s [ll]. 7l’4 2 that, . trcatise [23? p. 7231, or 12 BRUCE C. BER.NDT? After a somewhat lengthy HENG HUAT calculation~ CHAN, LIANG-CHENG ZHANG WC find t,hat G*d = 9 + 48 49 AND * (9 + 3&j71’4 2 2V5 Thus , After a mild calculation, (3.12) ~+dm=9+4fi+2&/lG+Gfi =9+&?+2&(3+&-) = (2 + fi)(3&+ Using (3.10)-(3.12) in Theorem 1; we deduce 2fi) that ~~~~ =/T/F(2 + A)I,6 (3.13) where A :=&92 + 72&)(189 =Gd2OlG Using this calculation prove that in (3.13), + 72~6) + 7G2v? WC sce from (3.13) and (3.9) that it remains (3.14) ZZZ-- ; 1G + G&‘+ (3 + &‘)G1~271~4 + ; GI/?+ (3 + v’?)G’~~~‘I~. to R W A M a AC L N I U L N, c 3pw a TV AAA A CN e pi n m lt S NN F B D TR m yh SI , I 1 NA U CC 3 ET DI O N a d 1 A R ’ CS N IS U OA 8 1= 1 r * - .i i e a( z w c = 1 Sg 1 i - 3 7 = ( = C + 3 1 + 2 s n tt i2 69 7 4 2 ) I w m e e sthn. 2 ; + .f 1 1 J 2 i (+ v + f G e f( l7 q c oi a1, ’ i +/ +i v & G1 2 e i3 ) ) q. ’ ln t5 s + ’ 4 ( 2 vg , ) 4 3 4 _ ( + f 4 W n v tC T c h h : = b o d + f ( T T 7 ( m a G 7 a3 . . ( + & i ( c n g3 V i l t 2 %j f 4. y , ) 2 p1uw i n. cl 6 r ss as = + a t 2 b ) 1 1 6 l o W. nl ul p d ( (+ o ,, ’ 5 o c br ) ’ f[ c e o ) .p e j] o .2Rb r f l m ) p w 7o oo + 2 4m ) c r o 02 nN ht f J =s h& l N h I. s 1doG h o . a , 1 c l 1 i ’ ) e Te a[ t _Y10 = A f s 33 f o 1ho h = + G 4 y (2 n =r 1 p r w + (( + 8 i E 3 S i u ,990 P 3 1 + 3 C . 1G ’ ( 2r om r 3 ac 1e k a , n , -r ’ & . a I yto . c o a o1 9 = ( l 6 ,w l t t 3 2 s ) h s l e ) + o a 1 l t & a ’ BRUCE 14 C. BERNDT, HENG T~US, from (2.13) and (3.16)> (3.17) .!J90 = (2 + Comparing (3.17) and (3.15); HUAT CHAN, { dwYJs+ AND LIANG-CHENG pyz+ ZHANG /yz}1’3 d%p6 we sec t,hat, wc ml& prove that (3.18) In thc notation (2.19) and (2.21), U2(U2 - 3)2 = 18 + &. (3.19) Factoring we salve in Z[&], we find that N(U2)N2(UZ - 3) = N(18 + GV@ = 3. 62 Wc thus want t,o solvc &6 = N(U* - 3) =: N(A + I?d@ = A* - 6B2 Cho~se A = 0 and B = 1: SO that u,* = 3 + fi that (3.19) is satisficd. Then and N(U*) = 3. It is trivial to see _/q+# and thc verification Theorem (3.20) of (3.18) is complctc. 8. gl‘J8 = 43 (4&+ Proof. Set# n, = 22 in Theorem paper [ll], (3.21) An cas-y calculation &?)1’6 (p-z+p-q. 2. From Wcber’s g22 yields p = 4fi. = work [23, p. 7221, or R,amanujan’s 1+lh. Thlls, from Thcorcm 2, RAMANUtJAN’S By (3.20) CLASS and (3.22), INVARIANTS AND CUBIC CONTINUED FRACTION 15 we must show that (3.23) In t,hc not&ion of (2.19) and (2.21)> WC seek to salve UZ(U2 - 3)2 = 4(18 + 36). (3.24) T~US, , fV(U2)AqU2 - 3) = ïV(4(18 + 3m)) we attempt Let A = 1 = B, SO t,hat u* = (9 + a)/2 (3.24) holds. = 24 . 33 t,o find a solution of and N(U~) = 12. It is easily checked that Hcncc; *,+jE+/~. Thus, we cari deduce 4. Explicit Proof. (3.23), and the proof is complctc. Evaluations of Ramanujan’s It was first proved by Watson Cubic Continued Fraction [15] that (4.1) where x(q) is defined by (1.4). 34G, 3471.) For q = -C G( (4.2) -e- xfi, (F or rcferenccs t,o o&r WC havc, by (4.1) and (1.7). proofs of (4.1), sec 11, pp. 1 B 6 C B R H. u E HX J : CE RAU J L H N - 4 + & ( = -? l f ( a B f ( r( 2w u ( f n r t o4 2 t 4 S ( u W n i ( . l x 4 9 w L S [ f E a 2 P e c t 6 p x F o e . Wr tr h G ]r r e[o o f tt a G o+ G h4 et e ~ x : al 2 3 t - ( , GN , , e) oi p o w s a o R b2o h t r p ca3f l + ’ aii n l to l t m h a1 1 r p m - Ui , e f a rm ,. uf c e i e c l [ . e ~ v 9d y . h ? o e2m r f - p . ) l op = cI2 T i n b f T n. A = h= s f o NA, $ t) p o p r7o d rt mo m C I e l a G . i 2. =t 1 r a T ( ) m o. n ) 8 t . Cps o t = )? ) a2 DDT E A H N ) . a n = r1 T n m hxa f h c G P 2 , ) 7 4 4W b c n e Tc 1 o . i E . e i d hm . - 4J p 2 i NNA C I Z A G Sa o . r e 1 t1 p’ a i es n d 1 F S , = ( ’O tl: 3 od A 4 y ) w -) s ( & 4 S 4 a e l t v it , Gn v v =h ( l o G 1m h g a as ft b 3 ic i l t io , t e; lp v u o n Tfh h gar c e f gh e e iro n a 6 sd [c & pl 7 J T [ 2 w ? .uc 2 d1 l 3 t+ e eaf 2 ci l , h or 1 d’ ] e a l u a t nl o il , u ; i e s RAMANUcJAN’S E 3 x 4 P INVARIANTS . . ra i tr p u t ? ( ah ( G H ( W g E L S U f i [ f ( v l e 2 ~ ( s o i o t x 6 a3 a o t 5 , n m h a1 7r m e h hc [t [ sv ] r i c . n. = h ( .a n +9 5d l 4 l l u s e m c ) r n) o l , . ~ p. g. o l h ; d c) i e e i 1uat w r s c e . , 1 n n] y, & f c ~ l ew f 2cl g a S )G d o , + )o p m0 , ) y e c 2 2 Y 2c J 9 = f0 1. a n e l b3 TI2 o& d f mg7 d 1a ? )’ p , .= o 21 c H l 2 u 3 . 1+ iG l j ) ) t. v T owl 1 1v 7 n o on. c h . J) ) 9 - - Ym q I . e. e.p o a o r(i1 c 4t d . 2 ) J m 3v i l . y ( (m a c ( 1 po. 7 h lc dn + (4 p 2 . (d = r efu e C ( ef + ? 1 ( & e n t4 , te + 2+ J ( = , d tc i x ~ = r2 T a p J >2w0 5 P m s r ee i a (o a f f 3 ( w4 c 2 + = - i u( - e i ne . l % d _2 17 V l 4 4 P i 4 f ( ( . FRACTION p o s n ph g r W do : CONTINUED m & x A AND CUBIC a dE)_ G(-em E CLASS c e o l s/ ) e s )crh i s 2, d . ,i e n o 18 BR,UCE Example 6. Example 7. Example 8. Example C. BER,NDT, HENG HUAT CHAN. 6 appcars in Ramanujan’s AND LIANG-CHENG lest, notebook ZHANG [14, p. 3G6] and was first provcd by K.G. Ramanathan [7]. We give our thanks t,o Deug Bowman for informing us of th 3 in Carr’s book [5]. version of Lemma R,EFERENCES 1. B. C. Berndt: Nokbooks, Rumanujan’s Part III. Springer-Verlag: 2. B. C. Berndt, and H. H. Chan, Some values for New York, 1994. the Ro~ers-Ra?nan,7L~~Lr~ continued fruction (to appear). 3. B. C. Berndt, formula, H. H. Chan. and L.-C. and modular equations Pi a& 4. J. M. and P. B. Borwein, 5. G. S. Carr, Formulas G. H. H. Chan, Ramanujan’s Zhang; the AGM, and Th,eorems On Ramanujan’s Wilcy, in Pure cubic On Ramanujan’s On th,e Ro,qers-Ramanujan Sci.) 93 (1984), 9. K. G. R,amanathan, fraction, continued 7. K. G. R,amanathan. New York. Mathematics, continued 8. K. G. R,amanathan, (Math. cZass in,variants, Kronecker’s limit (to appear). 1987. 2n,d ed.; Chclsea, New York, 1970. (to appear). fraction, Acta continued Arith. fraction,. 43 (1984), 209-226. Proc. Indian Acad. Sci. J. Pure Appl. Math. 16 (1985), 67-77. continued Ramanujan’s fraction, Indian 695-724. 10. K. G. R,amanathan, (1987), Some of Kronecker’s applications limit formula, J. Indian Math. Soc. 52 71-89. 11. S. Ramanujan, Modulur (1914), 350-372. Notebooks 12. S. R,amannjan, equutions a& to x. approzimaiions (2 ,volumes), Tat,a Institnt,e Qnart. J. Math. of Fnndamcntal (Oxford) R,esearch; 45 Bombay, 1957. 13. S. R,amanujan, Collected Papers. 14. S. R,amanujan, TlAe Lest Nokbook 15. G. N. Watson, Theorems Soc. 4 (1929), 231-237. 1G. G. N. Watson, Soc. 6 (1931), TlLeorems stated Chclsca. New York. by Raman,ujan stated 19G2. anci OtlLer Unpuhlisl~,ed Papers. (IX): hy Ramanujan Narosa, two con,tinued fractions. (XIV): a sinqular modt~h~, New Delhi, Math. London Math. J. 12G-132. 17. G. N. Watson, Som,e singular moduli (I), 18. G. N. Watson; Some moduli (II), singular Q nart. J. Math. 3 (1932): 81-98. Q uart. J. Math. 3 (1932); 189-212. 19. G. N. Watson, Singular moduli (3); Proc. 20. G. N. Watson, Singular moduli (4). Acta (5). Proc. London Math. Soc. 42 (1937), 377-397. (G). Proc. London Math. Soc. 42 (1937), 398-409. 21. G. N. Watson, Sin,qulaT moduli 22. G. N. Watson. Singular moduli London Arith. Math. Soc. 40 (1936), 1988. J. London 83-142. 140 (193G), 284-323. R 2 H W AC I Le!~,rbuclt, 3 . eder Al,qebTa. . dritter b DEPARTMENT OF MATHEMATE~, URBANA,IL 61801, USA MO ML AN Band, 1409 WEST C AA C e C NV N Y h U NS F O DA 1 r e o c B U S R 1 N R 9, w GREEN STREET, UNIVERSITY OF ILLINOIS. DEPARTMENT OF MATIIEMATICS, SOCITIIWESTMISSOURI STATE UNIVERSITY,SPRINGFIELD, 65804, USA r l I 6 J 9A T I k s C 1 A CI A ? c