JACOB’S LADDERS, HETEROGENEOUS QUADRATURE FORMULAE, BIG ASYMMETRY AND RELATED FORMULAE FOR THE RIEMANN ZETA-FUNCTION 4 1 0 JANMOSER 2 n Abstract. In this paper weobtain as our mainresultnew class of formulae a expressing correlation integrals of the third-order in Z on disconnected sets J G˚1(x),G˚2(y) by means of an autocorrelative sum of the second order in Z. 3 Moreover, the distance of the sets G˚1(x),G˚2(y) from the set of arguments of 1 autocorrelative sum isextremely big, namely ∼Aπ(T), T →∞,where π(T) istheprime-countingfunction. ] A C . h 1. Introduction t a 1.1. Let us remind the following formulae m 2T 2π [ Z4(t)dt Z4(t ), T , 1 ZT ∼ lnT T≤tνX≤T+2T ν →∞ (1.1) v T+U 2π 6 Z2(t)dt Z2(t ), U =√T lnT, T Z ∼ lnT ν →∞ 5 T T≤tXν≤T+U 8 (see [5], (4.4), (4.7)), where 2 . 1 1 Z(t)=eiϑ(t)ζ +it , 0 Å2 ã 4 (1.2) t 1 t 1 ϑ(t)= lnπ+ImlnΓ +i : −2 Å4 2ã v (see [9], pp. 79, 329), and t is the Gram sequence (comp. [9], p. 99). The i ν X { } formulae (1.1) were proved by us in connection with the Kotelnikov-Whittaker- r Nyquist theorem. Namely, by these formulae we have expressed the biquadratic a and quadratic effects for the continuous signals Z(t), t [T,2T]; t [T,T +U] ∈ ∈ from the point of view of information theory. 1.2. The formulae (1.1) are: (a) asymptotic quadrature formulae (from the left to the right), (b) asymptotic summation formulae (from the right to the left). Next, for the second formula in (1.1) (for example) we have t [T,T +U] t [T,T +U], ν (1.3) ∈ → ∈ Z2(t) Z2(t ), ν → Key words and phrases. Riemannzeta-function. 1 Jan Moser Jacob’s ladder ... i. e. the segment [T,T +U] and the exponent 2 are conserved. Remark 1. By(1.3)itisnaturaltocallthe formulaeofthekind(1.1)homogeneous formulae. On the contrary, we obtain in this paper some heterogeneous asymptotic quad- rature formulae – such formulae that the properties (1.3) are not fulfilled. 2. Heteregeneous quadrature formulae 2.1. Let (see [3], (3)) G (x)=G (x;T,H)= 1 1 π = t: t ( x)<t<t (x) , 0<x , 2ν 2ν { − } ≤ 2 [ T≤t2ν≤T+H (2.1) G2(y)=G2(y;T,H)= π = t: t ( y)<t<t (y) , 0<y , 2ν+1 2ν+1 { − } ≤ 2 [ T≤t2ν+1≤T+H H =T1/6+2ǫ, where the collection of sequences t (τ) , τ [ π,π], ν =1,2,... ν { } ∈ − is defined by the equation (see [3], (1)) ϑ[t (τ)]=πν+τ; t (0)=t , ν ν ν and (see [3], (8)) x x y y (2.2) m G (x) = H + , m G (y) = H + , 1 2 { } π O(cid:16)lnT(cid:17) { } π O(cid:16)lnT(cid:17) where m G (x) , ...is the measure of the set G , .... 1 1 { } Let ϕ G˚ (x) =G (x), ϕ G˚ (y) =G (y). 1 1 1 1 2 2 { } { } The following Theorem holds true. Theorem 1. ω(t)Z[ϕ (t)]Z2(t)dt= Z 1 G˚1(x) m G (x) x H 1 = { } Z(t )Z t + + , Q lnP ν Å ν lnP ã OÅlnTã 1 0 X 0 T≤tν≤T+U1 (2.3) ω(t)Z[ϕ (t)]Z2(t)dt= Z 1 G˚2(y) m G (y) y H 2 = { } Z(t )Z t + + , − Q lnP ν Å ν lnP ã OÅlnTã 1 0 X 0 T≤tν≤T+U1 π x,y 0, , T , ∈(cid:16) 2 i →∞ where 1 lnlnt (2.4) ω(t)= 1+ , lntß OÅ lnt ã™ Page 2 of 9 Jan Moser Jacob’s ladder ... and (see [1], (38); H U ) 1 → 1 U2 Q = 1= U lnP + 1 , 1 π 1 0 OÅ T ã X (2.5) T≤tν≤T+U1 T U1 =√T lnP0, P0 =… . 2π Furthermore we have (2.6) m G˚ (x) , m G˚ (y) <T1/3+ǫ. 1 2 { } { } 2.2. Remark 2. Every of the correlation integrals in (2.3) contains the product (2.7) Z[ϕ (t)]Z2(t) 1 and, as usually, we have (see [6], (6.2); [7], (8.4)) t (2.8) t ϕ (t) (1 c) (1 c)π(t), t , 1 − ∼ − lnt ∼ − →∞ i. e. wehavebigdifferenceofargumentsin(2.7),wherecistheEulerconstantand π(t) is the prime-counting function. Next, in the case t=T˚; ϕ (T˚)=T T T˚ 1 ⇒ →∞ ⇔ →∞ we obtain from (2.8) T˚ T 1 T˚ T (1 c) 1 (1 c) − ∼ − lnT˚ ⇒ − T˚∼ − lnT˚ ⇒ T˚ T lnT˚ lnT, ⇒ ∼ ⇒ ∼ i. e. T T˚ T (1 c) , − ∼ − lnT and (see (2.5)) T T T˚ (T +U ) (1 c) U (1 c) . 1 1 − ∼ − lnT − ∼ − lnT Consequently we have ˚ ρ [T˚,T +H];[T,T +U ] (1 c)π(T); 1 (2.9) { }∼ − ˙ ˚ [T,T +U ] [T˚,T +H], 1 ≺ where ρ stands for the distance of the˙corresponding segments. We may, of course, put (2.10) G (x) [T,T +H]=G¯ (x) G (x),... 1 1 1 ∩ → if necessary. Remark 3. We have the following properties Page 3 of 9 Jan Moser Jacob’s ladder ... (a) ˚ G˚ (x),G˚ (y) [T˚,T +H] [T,T +U], 1 2 ∈ → (comp. (1.3), (2.10)), where G˚ (˙x),G˚ (y) are disconnected sets, 1 2 (b) if =G˚ (x),G˚ (y) 1 2 G then extremely big distance occurs, namely (comp. (2.9), (2.10)) ρ ;[T,T +U ] (1 c)π(T), T , 1 {G }∼ − →∞ (c) for the corresponding orders of Z (comp. (1.3)) 1+2 1+1 → then the formulae (2.3) are strongly heterogeneous (comp. Remark 1). Remark 4. Moreover we explicitly notice the following: (a) the formulae (2.3) are not accessible by the current methods in the theory of the Riemann zeta-function, (b) small improvements of the exponents 1 1 , , ... 6 2 are irrelevant for main direction of this paper (comp. our paper [2], Ap- pendex A: On I.M. Vinogradov’ scepticism on possibilities of the method of trigonometric sums). 3. Big asymmetry and related formulae 3.1. Let (see [4], p. 29) G (x)=G (x;T,U )= 3 3 2 π = t: g ( x)<t<g (x) , 0<x , 2ν 2ν { − } ≤ 2 [ T≤g2ν≤T+U2 (3.1) G4(y)=G4(y;T,U2)= π = t: g ( y)<t<g (y) , 0<y , 2ν+1 2ν+1 { − } ≤ 2 [ T≤g2ν+1≤T+U2 U =T5/12+2ǫ, 2 where the collection of sequences g (τ) , τ [ π,π], ν =1,2,... ν { } ∈ − is defined by the equation (see [4], (6)) π τ ϑ [g (τ)]= ν+ ; g (0)=g , 1 ν ν ν 2 2 where (comp. (1.2)) 1 t t t π ϑ(t)=ϑ (t)+ , ϑ (t)= ln , 1 OÅtã 1 2 2π − 2 − 8 Page 4 of 9 Jan Moser Jacob’s ladder ... and (see [4], (13)) x x m G = U + , 3 2 (3.2) { } π O(cid:16)lnT(cid:17) y y m G = U + . 4 2 { } π O(cid:16)lnT(cid:17) Let ϕ G˚ (x) =G (x), ϕ G˚ (y) =G (y). 1 3 3 1 4 4 { } { } The following Theorem holds true. Theorem 2. ω(t)Z2[ϕ (t)]Z2(t)dt ZG˚3(x) 1 ∼ 4 (3.3) U sinx+ ω(t)Z2[ϕ (t)]Z2(t)dt, ∼ π 2 ZG˚4(x) 1 π x 0, , T . ∈(cid:16) 2i →∞ Remark 5. By the asymptotic formula (3.3) is expressedthe property ofbig asym- metry in the distribution of the values of Z on disconnected sets G˚ (x),G˚ (x). 3 4 Namely, the correlation integral (comp. (2.8)) on the disconnected set G˚ (x) es- 3 sentially exceeds of that on the set G˚ (x). For example 4 4 U , T , ZG˚3(π/2)−ZG˚4(π/2) ∼ π 2 →∞ where m G˚ (π/2) +m G˚ (π/2) =U 3 4 2 { } { } if (comp. (2.10)) G (x) [T,T +U ] G (x),... 3 2 3 ∩ → 3.2. Finally, the following Theorem holds true. Theorem 3. ω(t)Z[ϕ (t)]Z2(t)dt= Z 1 G˚3(x) m G (x) = { 1 } ω(t)Z2[ϕ (t)]Z2(t)dt 2m{G3(x)} ZG˚3(x) 1 − m G (x) { 1 } ω(t)Z2[ϕ (t)]Z2(t)dt+ (HT−ǫ), − 2m{G4(x)}ZG˚4(x) 1 O (3.4) ω(t)Z[ϕ (t)]Z2(t)dt= Z 1 G˚2(y) m G (y) = { 2 } ω(t)Z2[ϕ (t)]Z2(t)dt 2m{G4(y)}ZG˚4(y) 1 − m G (y) { 2 } ω(t)Z2[ϕ (t)]Z2(t)dt+ (HT−ǫ), − 2m{G3(y)}ZG˚3(y) 1 O π x,y 0, , T . ∈(cid:16) 2i →∞ Page 5 of 9 Jan Moser Jacob’s ladder ... Remark 6. For the disconnected sets G˚ (x),G˚ (y),G˚ (x),G˚ (x),G˚ (y),G˚ (y);G˚ G˚ = ,G˚ G˚ = 1 2 3 4 3 4 1 2 3 4 ∩ ∅ ∩ ∅ we have the following property: the correlation integrals of the order 1 + 2 on G˚ (x),G˚ (y) are expressed as the linear combinations of the correlation integrals 1 2 of the order 2+2 on G˚ (x),G˚ (x);G˚ (y),G˚ (y) correspondingly. 3 4 3 4 4. Proof of Theorem 1 4.1. In the paper [1] (see (10)) we have proved the following autocorrelative for- mula Z(t )Z(t +β)= ν ν X (4.1) T≤tν≤T+U1 2sin(βlnP ) = 0 U lnP + (√T ln2T), 1 0 π βlnP O 0 where 1 (4.2) β = , U =√TlnP . OÅlnTã 1 0 In the case x π βlnP =x β = , x 0, 0 ⇒ lnP0 ∈(cid:16) 2 i we obtain from (4.1), (4.2) x Z(t )Z t + = ν Å ν lnP ã X 0 (4.3) T≤tν≤T+U1 2sinx = U ln2P + (√T ln2T). 1 0 π x O Consequently we have (see (2.5), (4.2)) 1 x Z(t )Z t + = Q lnP ν Å ν lnP ã 1 0 X 0 (4.4) T≤tν≤T+U1 sinx 1 π sinx 2 =2 + ; x 0, ,1 . x OÅlnTã ∈(cid:16) 2i ⇒ x ∈ïπ ã 4.2. Next, in the paper [3], (5),(9) we have obtained the following mean-value formula on the disconnected sets G (x),G (y) (see (2.1)) 1 2 2 Z(t)dt= Hsinx+ (xT1/6+ǫ), Z π O G1(x) (4.5) Z(t)dt= 2Hsiny+ (yT1/6+ǫ), Z −π O G2(y) H =T1/6+2ǫ; T1/6ψ2ln5T T1/6+ǫ. → Hence, from (4.4) by (2.3) we obtain 1 sinx Z(t)dt=2 + (T−ǫ), m G (x) Z x O (4.6) { 1 } G1(x) 1 siny Z(t)dt= 2 + (T−ǫ). m G (y) Z − y O { 2 } G2(y) Page 6 of 9 Jan Moser Jacob’s ladder ... 4.3. In the paper [7], (9.2), (9.5) we have proved the following Lemma: if ˚ ϕ [T˚,T +U] =[T,T +U], 1 { } ˙ then for every Lebesgue-integrable function f(x), x [T,T +U] ∈ we have ˚ T+U T+U f[ϕ (t)]Z˜2(t)dt= f(x)dx, (4.7) Z ¯ 1 Z T˚ T T T T [ϕ ], U 0, , ≥ 0 1 ∈Å lnTò where Z2(t) Z˜2(t)= =ω(t)Z2(t); 1+ lnlnt lnt (4.8) O lnt (cid:8) (cid:0)1 (cid:1)(cid:9) 1 lnlnt ω(t)= = 1+ . 1+ lnlnt lnt lntß OÅ lnt ã™ O lnt (cid:8) (cid:0) (cid:1)(cid:9) Consequently, we have (see (4.7), (4.8)) ˚ T+U T+U ω(t)f[ϕ (t)]Z2(t)dt= f(x)dx, (4.9) Z ¯ 1 Z T˚ T T U 0, . ∈Å lnTò 4.4. Hence, by (4.6), (4.9) we obtain following formulae 1 ω(t)Z[ϕ (t)]Z2(t)dt= m{G1(x)}ZG˚1(x) 1 sinx =2 + (T−ǫ), x O (4.10) 1 ω(t)Z[ϕ (t)]Z2(t)dt= m{G2(y)}ZG˚2(y) 1 siny = 2 + (T−ǫ). − y O Finally, simple elimination of the values sinx siny 2 , 2 x y from (4.4), (4.9) gives (2.3). Page 7 of 9 Jan Moser Jacob’s ladder ... 4.5. In the case f(x)=1 in (4.7) we obtain (see (2.1), (2.10), (4.8)) ˚ 1 2 T+H 1 2 ζ +it dt< ζ +it dt ZG˚1(x)(cid:12)(cid:12) Å2 ã(cid:12)(cid:12) ZT˚¯ (cid:12)(cid:12) Å2 ã(cid:12)(cid:12) ∼ (cid:12) (cid:12) (cid:12) (cid:12) (cid:12) T+H (cid:12) (cid:12) (cid:12) lnT 1 dt, ∼ Z · T i. e. ˚ (4.11) T+H 1 2 ζ +it dt<AHlnT, T . ZT˚¯ (cid:12)(cid:12) Å2 ã(cid:12)(cid:12) →∞ (cid:12) (cid:12) Let (cid:12) (cid:12) ˚ (4.12) T +H T˚ T1/3+ǫ, − ≥ (comp. [6], (2.5); where 1 is th˙e Balasubramanianexponent). Then 3 ˚ T+H 1 2 ˚ ˚ (4.13) ζ +it dt (T +H T˚)lnT >B(T +H T˚)lnT, ZT˚¯ (cid:12)(cid:12) Å2 ã(cid:12)(cid:12) ∼ − − (cid:12) (cid:12) ˙ ˙ and (see (4.11), (4.(cid:12)13)) (cid:12) ˚ (4.14) T +H T˚<CH =CT1/6+ǫ. − ˙ Now, (4.14) contradicts (4.12). Consequently we have that ˚ m G˚ (x) <T +H T˚<T1/3+ǫ 1 { } − ˙ and we obtain the second inequality in (2.6) by the similar way. 5. Proofs of Theorem 2 and Theorem 3 5.1. Inthepaper[4],(14),(15)wehaveprovedthefollowingmean-valueformulae Z2(t)dt= Z G3(x) x T 2 = U ln + (cx+sinx)U + (T5/12ln2T), 2 2 π 2π π O (5.1) Z2(t)dt= Z G4(y) y T 2 = U ln + (cy siny)U + (T5/12ln2T) 2 2 π 2π π − O Page 8 of 9 Jan Moser Jacob’s ladder ... on the disconnected sets G (x),G (y), (comp. (3.1)). From (5.1) we have in the 3 4 case x=y Z2(t)dt Z2(t)dt= Z −Z G3(x) G4(x) 4 = U sinx+ (xT5/12ln2T). 2 π O Consequently,we obtainfrom (5.1) by (4.7) – (4.9) with f(x)=Z2(x) the formula (3.3). 5.2. Next, from the formula (see [4], (16)) 1 1 Z2(t)dt Z2(t)dt m G (x) Z − m G (x) Z ∼ { 3 } G3(x) { 4 } G4(x) sinx 4 , T ∼ x →∞ we obtain by (4.7) – (4.9) (f(x)=Z2(x)) the following formula 1 ω(t)Z2[ϕ (t)]Z2(t)dt (5.2) m{G3(x)} ZG˚3(x) 1 − 1 sinx ω(t)Z2[ϕ (t)]Z2(t)dt 4 , T − m{G4(x)}ZG˚4(x) 1 ∼ x →∞ and, by the similar way, we obtain the formula for y (0,π/2]. Hence, the elimi- ∈ nation of sinx siny 2 , 2 x y from (4.10), (5.2) implies the formula (3.4). I would like to thank Michal Demetrian for his help with electronic version of this paper. References [1] J.Moser,‘Onautocorrelative sum inthetheory of theRiemannzeta-function‘, ActaMath. Univ.Comen.,37(1980), 121-134, (inRussian). [2] J.Moser,‘OntherootsoftheequationZ′(t)=0‘,ActaArith.40(1981),97-107,(inRussian), arXiv: 1303.0967. [3] J. Moser, ‘New consequences of the Riemann-Siegel formula‘, Acta Arith., 42 (1982), 1-10, (inRussian),arXiv: 1312.4767. [4] J. Moser, ‘New mean-value theorems for the function |ζ 1 +it |2‘, Acta Math. Univ. 2 Comen.,46-47(1985), 21-40. (cid:0) (cid:1) [5] J.Moser,‘OntheorderofTitchmarshsuminthetheoryoftheRiemann’szeta-functionand onthebiquadraticeffectintheinformationtheory‘,CzechoslovakMath.J.,41(116),(1991), 663-684,(inRussian),arXiv: 1112.9548. [6] J. Moser, ‘Jacob’s ladders and the almost exact asymptotic representation of the Hardy- Littlewoodintegral’,Math.Notes88,(2010) 414-422,arXiv: 0901.3937. [7] J.Moser,‘Jacob’sladders,thestructureoftheHardy-Littlewoodintegralandsomenewclass ofnonlinearintegralequations‘,Proc.Stek.Inst. 276,(2011), 208-221,arXiv: 1103.0359. [8] E.C. Titchmarsh, ‘On van der Corput ’s method and the zeta-function of Riemann (IV)‘, Quart.J.Math.,(Oxford),5,(1934), 98-105. [9] E.C.Titchmarsh,‘ThetheoryoftheRiemannzeta-function‘ClarendonPress,Oxford,1951. Department of Mathematical Analysis and Numerical Mathematics, Comenius Uni- versity,MlynskaDolina M105,84248Bratislava,SLOVAKIA E-mail address: [email protected] Page 9 of 9