GEORGIAN MATHEMATICAL JOURNAL: Vol. 3, No. 5, 1996, 485-500 ON SOME ENTIRE MODULAR FORMS OF WEIGHTS AND 29 FOR THE CONGRUENCE GROUP Γ0 (4N ) 7 2 G. LOMADZE Abstract. Entire modular forms of weights 27 and 92 for the congruence group Γ0 (4N ) are constructed, which will be useful for revealing the arithmetical sense of additional terms in formulas for the number of representations of positive integers by quadratic forms in 7 and 9 variables. 1. In this paper N, a, k, q, r, t denote positive integers; u, s are odd positive integers; H, c, g, h, j, m, n, α, β, γ, δ, ξ are integers; A, B, C, D, G are complex numbers; z, τ (Im τ > 0) are complex variables,and H = {τ ∈ C| Im τ > 0}. Further, ( uh ) is the generalized Jacobi symbol; nt is a binomial coefficient; ϕ(k) is Euler’s function; e(z) = exp 2πiz; η(γ) = 1 if γ ≥ 0 and η(γ) = −1 if γ < 0. Let X ϑgh (z|τ ; c, N ) = (−1)h(m−c)/N × m≡c (mod N ) ×e hence 1 g g 2 τ e m+ z ; m+ 2N 2 2 (1.1) X ∂n n ; c, N ) = (πi) (−1)h(m−c)/N (2m + g)n × ϑ (z|τ gh ∂z n m≡c (mod N ) 1 g 2 g ×e m+ τ e m + z (n ≥ 0). (1.2) 2N 2 2 1991 Mathematics Subject Classification. 11F37, 11F27, 11F11. Key words and phrases. Entire modular form, congruence subgroup, multiplier system, theta-function, half integral weight. 485 c 1997 Plenum Publishing Corporation 1072-947X/96/0900-0485$12.50/0 486 G. LOMADZE Put (n) ϑgh (τ ; c, N ) = ∂n ϑgh (z|τ ; c, N )z=0 (n ≥ 0). ∂z n (1.3) It is known (see, for e.g., [1], formulas (11) and (12)) that (n) (n) ϑg,h+2j (τ ; c, N ) = ϑgh (τ ; c, N ) (n ≥ 0), β g 2 (n) (n) ϑg,h+βg+βN (τ ; c, N ) (n ≥ 0), c+ ϑgh (τ + β; c, N ) = e 2N 2 β g 2 (n) ϑgh (τ − β; c, N ) = (−1)n e − c+ × 2N 2 (n) × ϑ−g,−h+βg−βN (τ ; −c, N ) (n ≥ 0). (1.4) (1.5) From (1.1) and (1.2), according to notation (1.3) it follows, in particular, that (n) ϑgh (τ ; 0, N ) = (πi)n ∞ X m=−∞ (−1)hm (2N m + g)n × 1 g 2 τ (n ≥ 0). Nm + ×e 2N 2 (1.6) For ξ2 6= 0, ξ1 , g, h, N with ξ1 g + ξ2 h + ξ1 ξ2 N ≡ 0 (mod 2), put Sgh ξ1 ; c, N = ξ2 X (−1)h(m−c)/N e mmodN |ξ2 | m≡c (mod N ) ξ g 2 1 m+ . 2N ξ2 2 Finally, let n ατ + β γτ + δ n ατ + β Γ0 (4N ) = γτ + δ Γ= For τ ∈ H put o αδ − βγ = 1 , o ∈ Γ γ ≡ 0 (mod 4N ) . s (γτ + δ)s/2 = (γτ + δ)1/2 , − π π < arg(γτ + δ)1/2 ≤ . 2 2 Definition. Let M be a matrix of an arbitrary substitution from Γ0 (4N ), and let v(M ) be a multiplier system on Γ0 (4N ) and of weight 2s . We shall say that a function F defined on H is an entire modular form of weight 2s and of multiplier system v(M ) on Γ0 (4N ), if (1) F is regular on H; ON SOME ENTIRE MODULAR FORMS OF WEIGHTS 487 α β of substitutions from Γ0 (4N ) and all (2) for all matrices M = γ δ τ ∈ H, ατ + β = v(M )(γτ + δ)s/2 F (τ ); F γτ + δ (3) in the neighborhood of the point τ = i∞, F (τ ) = ∞ X Am e(mτ ); m=0 (4) for all substitutions from Γ in the neighborhood of each rational point τ = − γδ (γ 6= 0, (γ, δ) = 1), (γτ + δ)s/2 F (τ ) = ∞ X A0m e m=0 m ατ + β . 4N γτ + δ Lemma 1 ([1], p. 58, Lemma 3). If g is even, then for n ≥ 0 and all substitutions from Γ0 (4N ) we have (n) ατ + β ϑgh ; 0, 2N = γτ + δ 2βN sgn δ = (sgn δ)n i(2n+1)η(γ)(sgn δ−1)/2 i(1−|δ|)/2 (γτ + δ)(2n+1)/2 × |δ| αγδ 2 h2 βδg 2 δ 2ϕ(2N )−2 (n) e ϑαg,h (τ ; 0, 2N ). ×e − 16N 4 4N Lemma 2 ([1], p. 61, Lemma 4). If γ 6= 0, then for n ≥ 0 (n) (γτ + δ)(2n+1)/2 ϑgh (τ ; 0, 2N ) = −1/2 n = e (2n + 1) sgn γ/8 2N |γ| − i sgn γ × n ατ + β X (n) × ; H, 2N + ϕg0 gh (0, H; 2N ) ϑg0 h0 γτ + δ H mod2N t n X o X Atk n (n−t) ατ + β ; H, 2N , · ϑg0 h0 + t k! z=0 γτ + δ t=1 k=1 where g 0 = δg − γh − 2γδN, h0 = −βg + αh − 2αβN, αβ g 0 2 βg g 0 e − × H+ H+ ϕg0 gh (0, H; 2N ) = e 4N 2 4N 2 δ × Sg−αg0 ,h−βg0 ; −αH, 2N , −γ (1.7) 488 G. LOMADZE Atk z=0 = ( (2k)!(−2N γπi(γτ + δ))k 0 if t = 2k if t = 6 2k (1.8) (t = 1, 2, . . . , n; k = 1, 2, . . . , t). From (1.8) it follows that t X Atk = k! z=0 k=1 (A t,t/2 (t/2)! z=0 0 if 2|t, if 2 - t. (1.9) 2. Lemma 3. For given N let where Ω(τ ; gl , hl , cl , Nl ) = = Ω(τ ; g1 , g2 , g3 ; h1 , h2 , h3 ; c1 , c2 , c3 ; N1 , N2 , N3 ) = n 1 = ϑ00 (τ ; c1 , 2N1 )ϑg2 h2 (τ ; c2 , 2N2 ) + N1 g1 h1 o 1 00 − ϑg2 h2 (τ ; c2 , 2N2 )ϑg1 h1 (τ ; c1 , 2N1 ) ϑg3 h3 (τ ; c3 , 2N3 ), N2 3 X hk . 2|gk , Nk |N (k = 1, 2, 3), 4|N Nk (2.1) (2.2) k=1 Then for all substitutions from Γ in the neighborhood of each rational point τ = − γδ (γ = 6 0, (γ, δ) = 1), we have (γτ + δ)7/2 Ω(τ ; gl , hl , 0, Nl ) = ∞ X n=0 Gn e n ατ + β . 4N γτ + δ (2.3) Proof. 1. From Lemma 2 for n = 2 and n = 0 (respectively with g1 , h1 , N1 , g10 , h10 , H1 and g2 , h2 , N2 , g20 , h02 , H2 instead of g, h, N, g 0 , h0 , H), according to (1.8)–(1.9), it follows that 1 (γτ + δ)3 ϑ00g1 h1 (τ ; 0, 2N1 )ϑg2 h2 (τ ; 0, 2N2 ) = N1 3 = −e sgn γ (4N1 N2 γ 2 )−1/2 × 4 X ϕg10 g1 h1 (0, H1 ; 2N1 )ϕg20 g2 h2 (0, H2 ; 2N2 ) × × H1 mod2N1 H2 mod2N2 ON SOME ENTIRE MODULAR FORMS OF WEIGHTS × ατ + β n 1 ατ + β ; H1 , 2N1 ϑg20 h20 ; H2 , 2N2 + ϑ00g0 h0 N1 1 1 γτ + δ γτ + δ ατ + β −4γπi(γτ + δ)ϑg10 h01 ; H1 , 2N1 × γτ + δ o ατ + β ; H2 , 2N2 . ×ϑg20 h02 γτ + δ 489 (2.4) Replacing N1 , g1 , h1 , H1 , g10 , h10 by N2 , g2 , h2 , H2 , g20 , h02 in (2.4), and vice versa, we obtain 1 (γτ + δ)3 ϑ00g2 h2 (τ ; 0, 2N2 )ϑg1 h1 (τ ; 0, 2N1 ) = N2 3 = −e sgn γ (4N1 N2 γ 2 )−1/2 × 4 X 0 ϕg2 g2 h2 (0, H2 ; 2N2 )ϕg10 g1 h1 (0, H1 ; 2N1 ) × × H1 mod2N1 H2 mod2N2 n 1 ατ + β ατ + β × ϑ00g0 h0 ; H2 , 2N2 ϑg10 h10 ; H1 , 2N1 + N2 2 2 γτ + δ γτ + δ ατ + β −4γπi(γτ + δ)ϑg20 h02 ; H2 , 2N2 × γτ + δ o ατ + β ; H1 , 2N1 . ×ϑg10 h01 γτ + δ (2.5) Subtracting (2.5) from (2.4), we obtain n 1 (γτ + δ)3 ϑ00 (τ ; 0, 2N1 )ϑg2 h2 (τ ; 0, 2N2 ) + N1 g1 h1 o 1 − ϑ00g2 h2 (τ ; 0, 2N2 )ϑg1 h1 (τ ; 0, 2N1 ) = N2 3 = −e sgn γ (4N1 N2 γ 2 )−1/2 × 4 X ϕg10 g1 h1 (0, H1 ; 2N1 )ϕg20 g2 h2 (0, H2 ; 2N2 ) × × H1 mod2N1 H2 mod2N2 n 1 ατ + β ατ + β × ϑ00g0 h0 ; H1 , 2N1 ϑg20 h20 ; H2 , 2N2 + N1 1 1 γτ + δ γτ + δ o 1 00 ατ + β ατ + β ϑg0 h0 − ; H2 , 2N2 ϑg10 h01 ; H1 , 2N1 . N2 2 2 γτ + δ γτ + δ From Lemma 2 for n = 0, we have (γτ + δ)1/2 ϑg3 h3 (τ ; 0, 2N3 ) = e sgn γ/8 (2N3 |γ|)−1/2 × (2.6) 490 G. LOMADZE × X ϕg30 g3 h3 (0, H3 ; 2N3 )ϑg30 h30 H3 mod2N3 ατ + β ; H3 , 2N3 . γτ + δ (2.7) Multiplying (2.6) by (2.7), according to (2.1), we obtain (γτ + δ)7/2 Ω(τ ; gl , hl , 0, Nl ) = × = −e(7 sgn γ/8)(8N1 N2 N3 |γ|3 )−1/2 × 3 Y X ϕgk0 gk hk (0, Hk ; 2Nk )Ω H1 mod2N1 k=1 H2 mod2N2 H3 mod2N3 ατ + β ; gl0 , h0l , Hl , Nl . γτ + δ (2.8) 2. In (2.8) let γ be even. Then, by (1.7), gk0 (k = 1, 2, 3) are also even. Now in (1.2) and (1.3) instead of m let us introduce new letters of summations mk defined by the equalities m − Hk = 2Nk mk (k = r, t, 3), respectively, and for brevity we put g 0 2 Tk = Hk + 2Nk mk + k (k = r, t, 3). 2 Then, by (2.1), in (2.8) for r = 1, t = 2 and r = 2, t = 1, we get ϑg00r0 hr0 ατ + β ατ + β Y ; Hr , 2Nr ; Hk , 2Nk = ϑgk0 hk0 γτ + δ γτ + δ k=t,3 = (πi)2 2 N/Nr 0 ατ + β (−1)hr mr 2(Hr + 2Nr mr ) + gr0 e Tr × 4N γτ + δ =−∞ ∞ X mr × × = × 0 (−1)ht mt e mt =−∞ ∞ X 0 (−1)h3 m3 e m3 =−∞ N/N t Tt 3 T3 4N N/N 4N ατ + β × γτ + δ ατ + β = γτ + δ ∞ n ατ + β X n ατ + β r t Bn r e × Bnt e δ 4N γτ + 4N γτ + δ n =0 =0 ∞ X nr ∞ X n3 ∞ X t ∞ n ατ + β X n ατ + β 3 Bn 3 e = Cn(r,t) e , 4N γτ + δ 4N γτ + δ =0 n=0 (2.9) since by (2.2) nk = NNk Tk are non-negative integers. Thus, for even γ, (2.3) follows from (2.1),(2.8), and (2.9). In (2.8) let now γ be odd. If h1 , h2 , h3 are even, then by (1.7), g10 , g20 , 0 g3 are also even, and we obtain the same result. But if hr is odd, then by ON SOME ENTIRE MODULAR FORMS OF WEIGHTS 491 (1.7), hr0 and gr0 will also be odd, and in (1.2) we shall have 2 2 m + gr0 /2 = m + (gr0 − 1)/2 + m + (gr0 − 1)/2 + 1/4. Hence ατ + β h0 ατ + β r × ; Hr , 2Nr = (πi)2 e γτ + δ 16Nr γτ + δ ∞ X 2 N/Nr 0 ατ + β (−1)hr mr 2(Hr + 2Nr mr )+gr0 e × Wr (r = 1, 2). 4N γτ + δ m =−∞ ϑ00gr0 h0r r Here by (1.4) we can imply that h0r = 1, since by (1.7), h0r and hr have the same parity, and 2 Wr = Hr + 2Nr mr + (gr0 − 1)/2 + + Hr + 2Nr mr + (gr0 − 1)/2 (r = 1, 2). (2.10) Analogously, h0 ατ + β ατ + β t ; Ht , 2Nt = e × γτ + δ 16Nt γτ + δ ∞ N/N X 0 ατ + β t (−1)ht mt e × Wt (t = 2, 1). 4N γτ + δ m =−∞ ϑgt0 h0t t The same formula is valid for t = 3 with the same remarks concerning h30 as for ht0 , t = 2, 1. Further, for k = t, 3 we have Tk if 2|gk0 , 0 2 (2.11) Wk = (Hk + 2Nk mk + (gk − 1)/2) + 0 0 if 2 - gk . (Hk + 2Nk mk + (gk − 1)/2) Thus, if among h10 , h02 , h03 at least one is odd, then, as in (2.9), for r = 1, t = 2 and r = 2, t = 1 we have Y ατ + β ατ + β ; Hr , 2Nr ϑgk0 hk0 ; Hk , 2Nk = ϑ00gr0 hr0 γτ + δ γτ + δ k=t,3 3 ∞ 1 1 X hk0 ατ + β X 0 nr ατ + β =e N × B e 4N 4 Nk γτ + δ n =0 nr 4N γτ + δ k=1 r ∞ n ατ + β n ατ + β X 3 t = × Bn0 3 e Bn0 t e 4N γτ + 4N γτ + δ δ n =0 n =0 ∞ X t 3 3 ∞ 1 1 X h0k ατ + β X (r,t) n0 ατ + β =e N = Dn0 e 4N 4 Nk γτ + δ 4N γτ + δ 0 k=1 n =0 492 G. LOMADZE = ∞ X Dn(r,t) e n=0 n ατ + β , 4N γτ + δ since, by (2.2), (2.10), and (2.11), nk = N Nk Wk and n = 14 N (2.12) 3 P k=1 h0k Nk + n0 are non-negative integers. Thus, for odd γ, (2.3) follows from (2.1), (2.8), and (2.12). Theorem 1. For given N the function Ω(τ ; gl , hl , 0, Nl ) is an entire modular form of weight 7/2 and of the multiplier system β∆ sgn δ 2 v(M ) = i3η(γ)(sgn δ−1)/2 i3(|δ|−1) /4 (2.13) |δ| α β is a matrix of the substitution from Γ0 (4N ) and ∆ is the (M = γ δ determinant of an arbitrary positive quadratic form with integer coefficients in 7 variables) on Γ0 (4N ) if the following conditions hold: (1) (2) 2|gk , Nk |N (k = 1, 2, 3), 4|N 3 X k=1 h2k Nk 4| s X k=1 (2.14) gk2 , 4Nk (2.15) 3) for all α and δ with αδ ≡ 1 (mod 4N ) ∆ N N N 1 2 3 Ω(τ ; αgl , hl , 0, Nl ) = Ω(τ ; gl , hl , 0, Nl ). |δ| |δ| (2.16) Proof. 1. It is well known that theta-series (1.1)–(1.2) are regular on H; hence the function Ω(τ ; gl , hl , 0, Nl ) satisfies condition (1) of the Definition. 2. It is easily verified that (2.15) implies 4|N δ 2 3 X k=1 h2k /Nk , 4| 3 X δ 2ϕ(2Nk )−2 gk2 /4Nk , (2.17) k=1 since 2 - δ, because αδ ≡ 1 (mod 4N ). From (2.14) it follows that Γ0 (4N ) ⊂ Γ0 (4Nk ) (k = 1, 2, 3). (2.18) By Lemma 1 for n = 2 and n = 0, according to (2.17) and (2.18), for all substitutions from Γ0 (4N ), we have ατ + β ϑg00r hr ; 0, 2Nr = i5η(γ)(sgn δ−1)/2 i(1−|δ|)/2 × γτ + δ 2βN sgn δ r (2.19) (γτ + δ)5/2 ϑ00αgr ,hr (τ ; 0, 2Nr ), × |δ| ON SOME ENTIRE MODULAR FORMS OF WEIGHTS ατ + β ; 0, 2Nt = iη(γ)(sgn δ−1)/2 i(1−|δ|)/2 × γτ + δ 2βN sgn δ t (γτ + δ)1/2 ϑαgt ,ht (τ ; 0, 2Nt ), × |δ| ατ + β ϑg3 h3 ; 0, 2N3 = iη(γ)(sgn δ−1)/2 i(1−|δ|)/2 × γτ + δ 2βN sgn δ 3 (γτ + δ)1/2 ϑαg3 ,h3 (τ ; 0, 2N3 ). × |δ| 493 ϑgt ht (2.20) (2.21) Thus for all substitutions from Γ0 (4N ), if r = 1, t = 2 and r = 2, t = 1, we have ατ + β Y ατ + β ϑgk hk ; 0, 2Nr ; 0, 2Nk = ϑg00r hr γτ + δ γτ + δ k=t,3 2β sgn δ N N N r t 3 = i7η(γ)(sgn δ−1)/2 i3(1−|δ|)/2 (γτ + δ)7/2 × |δ| |δ| ατ + β Y ατ + β 00 ; 0, 2N ; 0, 2N ϑ . (2.22) ×ϑαg r k αg ,h ,h k k r r γτ + δ γτ + δ k=t,3 It is not difficult to verify that i3(1−|δ|)/2 2 2 = i3(|δ|−1) /4 . |δ| (2.23) Hence, by (2.1), (2.22), (2.23), (2.13), and (2.16) we get ατ + β 2 ; gl , hl , 0, Nl = i3η(γ)(sgn δ−1)/2 i3(|δ|−1) /4 × Ω γτ + δ β sgn δ N N N 1 2 3 (γτ + δ)7/2 Ω(τ ; αgl , hl , 0, Nl ) = × |δ| |δ| = v(M )(γτ + δ)7/2 Ω(τ ; gl , hl , 0, Nl ). Thus, the function Ω(τ ; gl , hl , 0, Nl ) satisfies condition (2) of the Definition. 3. From (1.6) it follows for r = 1, t = 2 and r = 2, t = 1 that ϑ00gr hr (τ ; 0, 2Nr )ϑgt ht (τ ; 0, 2Nt )ϑg3 h3 (τ ; 0, 2N3 ) = = (πi)2 ∞ X (−1)hr mr +ht mt +h3 m3 (4Nr mr + gr )2 × mr ,mt ,m3 =−∞ ×e X k=r,t,3 ∞ X 1 (2Nk mk + gk /2)2 τ = Bn(r,t) e(nτ ), 4Nk n=0 (2.24) 494 G. LOMADZE where n= 3 X 1 (2Nk mk + gk /2)2 = 4Nk k=1 = 3 X 3 (Nk m2k + mk gk /2) + k=1 1X 2 gk /4Nk , 4 k=1 by (2.14) and (2.15), is a non-negative integer. Thus, by (2.1) and (2.24), the function Ω(τ ; gl , hl , 0, Nl ) satisfies condition (3) of the Definition. 4. By Lemma 3, the function Ω(τ ; gl , hl , 0, Nl ) satisfies condition (4) of the Definition. 3. Lemma 4. For given N let Ψ1 (τ ; gl , hl , cl , Nl ) = Ψ1 (τ ; g1 , g2 , g3 ; h1 , h2 , h3 ; c1 , c2 , c3 ; N1 , N2 , N3 ) = n 1 = ϑ00 (τ ; c1 , 2N1 )ϑg2 h2 (τ ; c2 , 2N2 ) + N1 g1 h1 o 1 00 ϑg2 h2 (τ ; c2 , 2N2 )ϑg1 h1 (τ ; c1 , 2N1 ) ϑ0g3 h3 (τ ; c3 , 2N3 ) (3.1) − N2 and Ψ2 (τ ; gl , hl , cl , Nl ) = Ψ2 (τ ; g1 , g2 , g3 ; h1 , h2 , h3 ; c1 , c2 , c3 ; N1 , N2 , N3 ) = n 1 ϑ000 (τ ; c1 , 2N1 )ϑg2 h2 (τ ; c2 , 2N2 ) + = N1 g1 h1 o 3 00 (3.2) ϑg2 h2 (τ ; c2 , 2N2 )ϑ0g1 h1 (τ ; c1 , 2N1 ) ϑg3 h3 (τ ; c3 , 2N3 ), − N2 where 2|gk , Nk |N (k = 1, 2, 3), 4|N 3 X hk . Nk (3.3) k=1 Then for all substitutions from Γ, in the neighborhood of each rational point 6 0, (γ, δ) = 1), we have τ = − γδ (γ = (γτ + δ)9/2 Ψj (τ ; gl , hl , 0, Nl ) = ∞ X n=0 G(j) n e n ατ + β (j = 1, 2). 4N γτ + δ (3.4) ON SOME ENTIRE MODULAR FORMS OF WEIGHTS 495 Proof. 1. Formula (2.6) is obtained in Lemma 3. From Lemma 2, for n = 1, we have (γτ + δ)3/2 ϑ0g3 h3 (τ ; 0, 2N3 ) = −e(3 sgn γ/8)(2N3 |γ|)−1/2 (i sgn γ) × ατ + β X (3.5) ϕg30 g3 h3 (0, H3 ; 2N3 )ϑg0 0 h0 ; H3 , 2N3 . × 3 3 γτ + δ H3 mod2N3 Multiplying (2.6) by (3.5), according to (3.1), we obtain (γτ + δ)9/2 Ψ1 (τ ; gl , hl , 0, Nl ) = × X = e(9 sgn γ/8)(8N1 N2 N3 |γ|3 )−1/2 (i sgn γ) × 3 Y ϕgk0 gk hk (0, Hk ; 2Nk )Ψ1 H1 mod2N1 k=1 H2 mod2N2 H3 mod2N3 ατ + β ; gl0 , hl0 , Hl , Nl . (3.6) γτ + δ Reasoning further just as in Lemma 3.2, but taking everywhere ατ +β ατ +β instead of ϑg3 h3 γτ +δ ; H3 , 2N3 , from (3.1) and γτ +δ ; H3 , 2N3 (3.6) we obtain (3.4) if j = 1. 2. From Lemma 2 for n = 3, n = 0, and respectively for n = 2, n = 1, it follows that ϑg0 3 h3 × 1 (γτ + δ)4 ϑ000 g1 h1 (τ ; 0, 2N1 )ϑg2 h2 (τ ; 0, 2N2 ) = N1 1 e(sgn γ)(4N1 N2 γ 2 )−1/2 (i sgn γ) × = N1 n ατ + β X ϕg10 g1 h1 (0, H1 ; 2N1 ) ϑg0000 h0 ; H1 , 2N1 + 1 1 γτ + δ H1 mod2N1 3 X t o X Atk 3 (3−t) ατ + β ; H1 , 2N1 × ϑg0 h0 k! z=0 1 1 γτ + δ t t=1 k=1 ατ + β X ; H2 , 2N2 = × ϕg20 g2 h2 (0, H2 ; 2N2 )ϑg20 h02 γτ + δ H2 mod2N2 X = e(sgn γ)(i sgn γ)(4N1 N2 γ 2 )−1/2 ϕg10 g1 h1 (0, H1 ; 2N1 ) × + H1 mod2N1 H2 mod2N2 ατ + β n 1 ϑg0000 h0 ; H1 , 2N1 ϑg20 h20 (τ ; 0, 2N2 ) + N1 1 1 γτ + δ o ατ + β (3.7) −12γπi(γτ + δ)ϑg0 0 h0 ; H1 , 2N1 ϑg20 h02 (τ ; 0, 2N2 ) 1 1 γτ + δ ×ϕg20 g2 h2 (0, H2 ; 2N2 ) 496 G. LOMADZE and 3 (γτ + δ)4 ϑ00g2 h2 (τ ; 0, 2N2 )ϑ0g1 h1 (τ ; 0, 2N1 ) = N2 X = e(sgn γ)(i sgn γ)(4N1 N2 γ 2 )−1/2 ϕg10 g1 h1 (0, H1 ; 2N1 ) × H1 mod2N1 H2 mod2N2 n 3 ατ + β ϑ00g0 h0 ; H2 , 2N2 + ×ϕg20 g2 h2 (0, H2 ; 2N2 ) N2 2 2 γτ + δ o ατ + β ; H2 , 2N2 ϑ0g0 h0 (τ ; 0, 2N1 ). −12γπi(γτ + δ)ϑ0g0 h0 1 1 2 2 γτ + δ (3.8) Subtracting (3.8) from (3.7) and multiplying the obtained result by (2.7), according to (3.2), we get (γτ + δ)9/2 Ψ2 (τ ; gl , hl , 0, Nl ) = × X = e(9 sgn γ/8)(8N1 N2 N3 |γ|3 )−1/2 (i sgn γ) × 3 Y ϕgk0 gk hk (0, Hk ; 2Nk )Ψ2 H1 mod2N1 k=1 H2 mod2N2 H3 mod2N3 ατ + β ; gl0 , hl0 , Hl , Nl . (3.9) γτ + δ In (3.9) let γ be even. Then, reasoning as in Lemma 3.2, by (3.2), from (3.9) we obtain ατ + β n 1 ατ + β ϑg0000 h0 ; H1 , 2N1 ϑg20 h02 ; H2 , 2N2 + N1 1 1 γτ + δ γτ + δ ατ + β o 3 00 ατ + β − ϑg0 h0 ; H2 , 2N2 ϑ0g0 h0 ; H1 , 2N1 × 1 1 N2 2 2 γτ + δ γτ + δ n (πi)3 ατ + β ; H3 , 2N3 = ×ϑg30 h30 × γτ + δ N1 ∞ X 3 N/N1 0 ατ + β × × (−1)h1 m1 2(H1 + 2N1 m1 ) + g10 e T1 4N γτ + δ m =−∞ 1 × − 3(πi)3 N1 × N/N 4N 2 T2 ατ + β + γτ + δ N/N1 0 ατ + β o (−1)h1 m1 2(H1 + 2N1 m1 ) + g10 e × T1 4N γτ + δ =−∞ ∞ X m1 m2 =−∞ 0 (−1)h2 m2 e 2 N/N2 0 ατ + β (−1)h2 m2 2(H2 + 2N2 m2 ) + g20 e × T2 4N γτ + δ =−∞ ∞ X m2 ∞ X ON SOME ENTIRE MODULAR FORMS OF WEIGHTS × = − n × N/N 4N 3 T3 ατ + β = γτ + δ ∞ n ατ + β X n ατ + β 1 2 Bn2 e Bn1 e + + 4N γτ δ 4N γτ + δ =0 n =0 2 ∞ n ατ + β o n ατ + β X 1 2 Bn0 1 e × Bn0 2 e + δ 4N γτ 4N γτ + δ n =0 =0 ∞ X n2 0 (−1)h3 m3 e m3 =−∞ ∞ X n1 ∞ X 497 1 ∞ n ατ + β n ατ + β X 3 Cn e = . Bn 3 e 4N γτ + δ 4N γτ + δ n=0 =0 ∞ X n3 (3.10) Here Tk and nk have the same meaning as in Lemma 3.2. Thus, for even γ, (3.4) follows from (3.2), (3.9) and (3.10) if j = 2. Further, reasoning as in Lemma 3.2, we obtain (3.4) for odd γ if j = 2. Theorem 2. For given N the functions Ψ1 (τ ; gl , hl , 0, Nl ) and Ψ2 (τ ; gl , hl , 0, Nl ) are entire modular forms of weight 9/2 and of the multiplier system β∆ sgn δ 2 v(M ) = iη(γ)(sgn δ−1)/2 i(|δ|−1) /4 (3.11) |δ| α β (M = is the matrix of the substitution from Γ0 (4N ) and ∆ is the γ δ determinant of an arbitrary positive quadratic form with integer coefficients in 9 variables) on Γ0 (4N ) if the following conditions hold: (1) 2|gk , Nk |N (k = 1, 2, 3), (2) 4|N 3 X hk2 Nk k=1 4| s X gk2 , 4Nk (3.12) (3.13) k=1 (3) for all α and δ with αδ ≡ 1 (mod 4N ) N N N 1 2 3 sgn δ Ψj (τ ; αgl , hl , 0, Nl ) = |δ| −∆ = Ψj (τ ; gl , hl , 0, Nl ) (j = 1, 2). |δ| (3.14) Proof. 1. As in the case of Theorem 1, the functions Ψ1 (τ ; gl , hl , 0, Nl ) and Ψ2 (τ ; gl , hl , 0, Nl ) satisfy condition (1) and, by Lemma 4, also condition (4) of the Definition. 2. By Lemma 1 for n = 3 and n = 1, according to (2.17) and (2.18), for all substitutions from Γ0 (4N ), we have ατ + β ϑ000 ; 0, 2N1 = sgn δi7η(γ)(sgn δ−1)/2 i(1−|δ|)/2 × g1 h1 γτ + δ 498 G. LOMADZE 2βN sgn δ 1 (γτ + δ)7/2 ϑ000 αg1 ,h1 (τ ; 0, 2N1 ), |δ| ατ + β ϑg0 1 h1 ; 0, 2N1 = sgn δi3η(γ)(sgn δ−1)/2 i(1−|δ|)/2 × γτ + δ 2βN sgn δ 1 × (γτ + δ)3/2 ϑ0αg1 ,h1 (τ ; 0, 2N1 ), |δ| ατ + β ; 0, 2N3 = sgn δi3η(γ)(sgn δ−1)/2 i(1−|δ|)/2 × ϑg0 3 h3 γτ + δ 2βN sgn δ 3 0 (γτ + δ)3/2 ϑαg (τ ; 0, 2N3 ). × 3 ,h3 |δ| × (3.15) (3.16) (3.17) Thus, for all substitutions from Γ0 (4N ) we respectively get: (1) by (2.19), (2.20), and (3.17) ατ + β ατ + β ατ + β ; 0, 2Nr ϑgt ht ; 0, 2Nt ϑg0 3 h3 ; 0, 2N3 = ϑ00gr hr γτ + δ γτ + δ γτ + δ 2β sgn δ N N N 1 2 3 = sgn δiη(γ)(sgn δ−1)/2 i3(1−|δ|)/2 × |δ| |δ| ×(γτ + δ)9/2 ϑ00αgr ,hr (τ ; 0, 2Nr )ϑαgt ,ht (τ ; 0, 2Nt )ϑ0αg3 ,h3 (τ ; 0, 2N3 ) (3.18) for r = 1, t = 2 and r = 2, t = 1; (2) by (3.15), (2.20) for t = 2 and (2.21) ατ + β ατ + β ατ + β ϑg0001 h1 ; 0, 2N1 ϑg2 h2 ; 0, 2N2 ϑg3 h3 ; 0, 2N3 = γτ + δ γτ + δ γτ + δ 2β sgn δ N N N 1 2 3 × = sgn δiη(γ)(sgn δ−1)/2 i3(1−|δ|)/2 |δ| |δ| 000 (τ ; 0, 2N1 )ϑαg2 ,h2 (τ ; 0, 2N2 )ϑαg3 ,h3 (τ ; 0, 2N3 ); (3.19) ×(γτ + δ)9/2 ϑαg 1 ,h1 (3) by (2.19) for r = 2 , (3.16), and (2.21) ατ + β ατ + β ατ + β ; 0, 2N2 ϑ0g1 h1 ; 0, 2N1 ϑg3 h3 ; 0, 2N3 = ϑ00g2 h2 γτ + δ γτ + δ γτ + δ N N N 1 2 3 η(γ)(sgn δ−1)/2 3(1−|δ|)/2 2β sgn δ = sgn δi i × |δ| |δ| 00 (τ ; 0, 2N2 )ϑ0αg1 ,h1 (τ ; 0, 2N1 )ϑαg3 ,h3 (τ ; 0, 2N3 ). (3.20) ×(γτ + δ)9/2 ϑαg 2 ,h2 It is not difficult to verify that −2 2 i3(1−|δ|)/2 = i(|δ|−1) /4 . |δ| (3.21) ON SOME ENTIRE MODULAR FORMS OF WEIGHTS 499 Hence, by (3.1), (3.18), (3.21), (3.11), (3.14) (if j = 1) and (3.2), (3.19)– (3.21), (3.11), (3.14) (if j = 2) we get ατ + β 2 Ψj ; gl , hl , 0, Nl = sgn δiη(γ)(sgn δ−1)/2 i(|δ|−1) /4 × γτ + δ −β sgn δ N N N 1 2 3 (γτ + δ)9/2 Ψj (τ ; αgl , hl , 0, Nl ) = × |δ| |δ| = v(M )(γτ + δ)9/2 Ψj (τ ; gl , hl , 0, Nl ) (j = 1, 2). Thus the functions Ψj (τ ; gl , hl , 0, Nl ) (j = 1, 2) satisfy condition (2) of the Definition. From (1.6) it follows that ϑ00gr hr (τ ; 0, 2Nr )ϑgt ht (τ ; 0, 2Nt )ϑ0g3 h3 (τ ; 0, 2N3 ) = P ∞ X h m (−1) k=r,t,3 k k (4Nr mr + gr )2 (4N3 m3 + g3 ) × = (πi)3 mr ,mt ,m3 =−∞ X ×e k=r,t,3 ∞ X 1 Cn(r,t) e(nτ ) (2Nk mk + gk /2)2 τ = 4Nk n=0 for r = 1, t = 2 and r = 2, t = 1, ϑ000 g1 h1 (τ ; 0, 2N1 )ϑg2 h2 (τ ; 0, 2N2 )ϑg3 h3 (τ ; 0, 2N3 ) = ∞ P3 X = (πi)3 (−1) k=1 hk mk (4N1 m1 + g1 )3 × m1 ,m2 ,m3 =−∞ 3 ∞ X X 1 ×e Cn e(nτ ) (2Nk mk + gk /2)2 τ = 4Nk n=0 k=1 and = (πi)3 ϑ00g2 h2 (τ ; 0, 2N2 )ϑ0g1 h1 (τ ; 0, 2N1 )ϑg3 h3 (τ ; 0, 2N3 ) = ∞ P3 X (−1) k=1 hk mk (4N2 m2 + g2 )2 (4N1 m1 + g1 ) × m1 ,m2 ,m3 =−∞ ×e ∞ 3 X X 1 (2Nk mk + gk /2)2 τ = Dn e(nτ ), 4Nk n=0 k=1 since in all these expansions n= 3 X 1 (2Nk mk + gk /2)2 , 4Nk k=1 500 G. LOMADZE as has been shown by Theorem 1.3, is a non-negative integer. Therefore, in view of (3.1) and (3.2), the functions Ψj (τ ; gl , hl , 0, Nl ) (j = 1, 2) satisfy condition (3) of the Definition. References 1. G. Lomadze, On some entire modular forms of weights 5 and 6 for the congruence group Γ0 (4N ). Georgian Math. J. 1(1994), No. 1, 53–76. (Received 24.11.1994) Author’s address: Faculty of Mechanics and Mathematics I. Javakhishvili Tbilisi State University 2, University St., Tbilisi 380043 Republic of Georgia