information
Article Numerical Calculations to Grasp a Mathematical Issue Such as the Riemann Hypothesis
Michel Riguidel Department of Computer Science and Networks, Télécom ParisTech, 75015 Paris, France; [email protected]; Tel.: +33-67687-5199
Received: 3 April 2020; Accepted: 23 April 2020; Published: 26 April 2020
Abstract: This article presents the use of data processing to apprehend mathematical questions such as the Riemann Hypothesis (RH) by numerical calculation. Calculations are performed alongside graphs of the argument of the complex numbers ζ(x + iy) = a + ib and ξ(x + iy) = p + iq, in the 1 critical strip. On the one hand, the two-dimensional surface angle tan− (b/a) of the Riemann Zeta y y function ζ is related to the semi-angle of the fractional part of 2π ln 2πe and, on the other hand, the Ksi function ξ of the Riemann functional equation is analyzed with respect to the coordinates (x, 1 x; y). − The computation of the power series expansion of the ξ function with its symmetry analysis highlights the RH by the underlying ratio of Gamma functions inside the ξ formula. The ξ power series beside the angle of both surfaces of the ζ function enables to exhibit a Bézout identity au + bv c between ≡ the components (a, b) of the ζ function, which illustrates the RH. The geometric transformations in complex space of the Zeta and Ksi functions, illustrated graphically, as well as series expansions, calculated by computer, make it possible to elucidate this mathematical problem numerically. A final theoretical outlook gives deeper insights on the functional equation’s mechanisms, by adopting a computer–scientific perspective.
Keywords: Riemann Hypothesis; numerical calculation; functional equation; congruence
1. Introduction This article presents the use of data processing to apprehend mathematical questions such as the Riemann Hypothesis (RH) by numerical calculation. The geometric transformations in complex space of transcendental functions, illustrated graphically, as well as series expansions, calculated by computer, make it possible to elucidate this mathematical problem numerically. The article sets out its reasoning using descriptive graphics. The mathematical calculations have been developed and checked by computer, and the graphics have been performed through extensive specific Python programs. The RH [1] (“non-trivial zeros are located exclusively on the critical line x = 1/2”) is illustrated here with mathematical notions and numerical calculations: power series expansions of the Gamma function and Ksi function of the Riemann’s functional equation, and calculations on complex numbers and angles between two surfaces or two curves. The whole study is focused in the critical strip . S The study contribution follows from the examination of the two-angle conjunction of the Zeta function and the Ksi function. On the one hand, the Zeta function has a remarkable property on the critical line: The real/imaginary ratio of the Zeta function is interpreted as the tangent of an angle. On the other hand, the Ksi function is congruent, within a homothety and a rotation, and to a remainder series, whose sum function angle is not symmetrical with respect to the critical line. On the contrary, the residual congruent series intrinsically contains this symmetry on the critical line. It is therefore essentially an angular functional constraint of the holomorphic Zeta function and of the associated Ksi function, which is at the origin of the property of the RH. This constraint enigma is contained in the Gamma functions’ ratio of the Ksi function.
Information 2020, 11, 237; doi:10.3390/info11050237 www.mdpi.com/journal/information Information 2020, 11, x FOR PEER REVIEW 2 of 32
Information 2020, 11, 237 2 of 31 Ksi function, which is at the origin of the property of the RH. This constraint enigma is contained in the Gamma functions’ ratio of the Ksi function. The nine-pagenine-page article,article, in in German, German, by by B. B. Riemann Riemann [1 ],[1], is essentialis essential in thein the history history of mathematics of mathematics and isand instructive is instructive in understanding in understanding the conjecture’s the conjecture’s context. context. Historical Historical articles articles by Hadamard by Hadamard [2], Vall [2],ée PoussinVallée Poussin [3], Bohr [3], and Bohr Landau and Landau [4], Hardy [4], [ 5Hardy], and Hardy[5], and and Hardy Littlewood and Littlewood [6] allow us[6] toallow grasp us the to extentgrasp ofthe the extent question’s of the question’s difficulty. difficulty. Deligne’s demonstrationDeligne’s demonstration in 1973 [7 ,in8], 1973 via algebraic[7,8], via geometry,algebraic geometry, gives the visiongives the and vision the victorious and the pathvictorious in the path case ofin finitethe case sets. of A finite thoughtful sets. A introduction thoughtful introduction to the properties to the of theproperties Zeta function of the Zeta is the function book by is EC the Titchmarsh: book by EC 1951 Titchmarsh: edition [ 91951], 1986 edition edition [9], [ 101986]. Less edition fundamental [10]. Less booksfundamental retrace books the various retrace works the various up to theworks years up 1974to the [11 years] and 1974 1985 [11] [12 and]. A 1985 deep [12]. insight A deep of analytic insight numberof analytic theory number concepts theory and concepts methods and can method be seens incan the be 2004 seen book in the by 2004 Iwaniec book and by Kowalski Iwaniec [and13]. TheKowalski search [13]. for non-trivial The search zeros for non-trivial using the properties zeros usin ofg complexthe properties analysis of orcomplex number analysis theory is or presented number intheory historical is presented and recent in historical publications and [recent14–20]. publications [14–20]. After this introduction, Section2 2 presents presents thethe methodology,methodology, i.e.,i.e., thethe graphicgraphic andand computationalcomputational illustration strategy. Section Section 33 presentspresents preparatorypreparatory calculationscalculations thatthat areare essentialessential forfor thethe results.results. Section4 4 presents presents thethe resultsresults ofof thethe calculationscalculations byby graphicalgraphical representationsrepresentations andand powerpower expansions.expansions. Section 55 presents presents thethe discussiondiscussion onon proofproof strategies,strategies, beforebefore thethe conclusion,conclusion, wherewhere thethe resultsresults areare summarized and research roadmapsroadmaps areare presentedpresented moremore abstractly.abstractly.
2. Methods This articlearticle presentspresents aa relation relationϕ of of the the function functionζ(s ) =( )ζ=(x + (iy + ) = a) += + ib in a congruencein a congruence form au + bv c, with a and b being the real and imaginary parts of ζ(s), respectively,( ) and, u and v being form ≡ + ≡ , with and being the real and imaginary parts of , respectively, and, independentand being functionsindependent of Zeta, functions which of are Ze dependentta, which are only dependent on x and yonly. This on congruence and . This has itscongruence source in the Riemann’s functional equation ζ(s) = ξ(s)ζ(1 s)( and) the Gamma( ) function Γ whose variations has its source in the Riemann’s functional equation− = ( ) 1− and the Gamma function Γ andwhose dependencies variations and are investigated.dependencies The are study investigated is based. onThe the study specificities is based of on the the Zeta specificities function, the of Ksithe functionZeta function, and the the Gamma Ksi function function, and by the analyzing Gamma them function, with respectby analyzing to the aforementionedthem with respect relations. to the Theaforementioned components arelations.and b, formed The ofcomponents ripples, cancel and each other, formed out periodically of ripples, and cancel alternately, each other and thusout causeperiodically on the criticaland alternately, line, in the and neighborhood thus cause ofon these the undulations,critical line, thein the emergence neighborhood of an additional of these zeroundulations, but, this the time, emergence common of to an both additional curves. Thezero article but, this thus time, provides common a numerical to both curves. illustration The article of the Riemannthus provides Hypothesis a numerical (RH). illustration of the Riemann Hypothesis (RH). The methodology flowflow chart is summarized inin FigureFigure1 1::
Figure 1. Methodology Flow chart of numerical calculationscalculations to grasp the RH issue.
2.1. Strategy for Discovering Fertile Domains and Sterile Domains to Zeros We try to identify, in a domain s = x + iy, s C , a function ϕ of the complex plane which We try to identify, in a domainD ≔{ = + , ∈ℂ∈ }, a function of the complex plane ( ) = 2 obeys:which obeys:s ∀ ∈ ⊂ℂ∶C : ϕ ζ ∘ s ( )0=0. Expressed. Expressed in inR ℝ, the, the requirement requirement is is as as follows: follows: wewe trytry toto ∀ ∈ D ⊂ n ◦ 2 o 2 identify, in a domain M = (x, y) R a pair (u, v) of real functions in R which obeys: M identify, in a domainD ≔{ =(∈ , ) ∈ℝ } a pair ( , ) of real functions in ℝ which∀ ∈obeys: D ⊂ R2 : (ζ(s)) u(x, y) + (ζ(s)) v(x, y) = 0. ∀ ∈< ⊂ ℝ ×∶ ℜ ( ) × = ( , )×+ ℑ ( ) × ( , ) =0. This domain is a fertile domain where it is possible to find non-trivial zeros, since (ζ(s)) = 0 This domain D is a fertile domain where it is possible to find non-trivial zeros, since< ℜ ( ) = then entails (ζ(s)) = 0, and vice-versa. This function pair (u, v) must obey the following requirements: 0 then entails= ℑ ( ) =0, and vice-versa. This function pair ( , ) must obey the following Therequirements: pair (u, v) mustThe notpair result ( , ) from must the Zetanot function,result from nor fromthe anZeta auxiliary function, expression nor from of Zeta, an auxiliary but must onlyexpression depend of on Zeta, the but point must coordinates only depend of the on domain; the point this coordinates pair must alsoof the not domain; cancel outthis simultaneously pair must also
Information 2020, 11, 237 3 of 31 for any given point, or more precisely, can only cancel out simultaneously for a zero measure set of points in this domain. To solve the RH, it is necessary to show both that the favorable domain is the critical line and also that there are no such pairs in the other domains, in particular in the critical strip S, L apart from the line , (zone ). We thus propose to find pairs of functions in the critical strip, L S\L such as au + bv = c by favoring the pairs (u, v) that cancel out c. It is clear that the Riemann functional equation offers tracks for mapping out the research roadmap, but it is not sufficient. Indeed, the functional equation expresses a powerful necessary constraint: if the point M s or (x, y) is a non n o trivial zero, then Mˆ sˆ = 1 s or (1 x, y) , M s = x iy or (x, y) and Mˆ sˆ = 1 x + iy or (1 x, y) − − − − − − − are also zeros. The pair (u, v) that we are looking for must therefore meet this requirement and exhibit this symmetry. It is hence advisable to decompose the critical strip into longitudinal sections of double lines x = xp and x = 1 xp, given the functional equation symmetry. Conceived with this constraint, − the problem then becomes finding pairs up, vp for all points P with coordinates sp = xp + iy from the lines x = xp such that: y : ζ sp up(y) + ζ sp vp(y) = 0. ∀ < × = × y : ζ sˆp uˆp(y) + ζ sˆp vˆp(y) = 0. ∀ < × = × uˆp, vˆp is the symmetric pair, associated with the pair up, vp . If this search for lines perpendicular to the real axis proves unfruitful, it is advisable to entrench oneself towards a search for purely isolated points xp, yp in the critical strip which satisfies this identity. On the other hand, to denounce the RH, it suffices to find at least one point Q such that: Q xq, yq : ζ sq uq + ζ sq vq = 0; uq , 0; vq , 0 ∃ < × = × 2.2. Qualification of Domains by Filtering Modulo a Reference Series This strategy can only succeed by the determination of an intrinsic property of the Zeta function, which can generate an equivalence relation in order to fold the imaginary axis on itself in a quotient space, so as to avoid calculations to infinity. It is through the search for angles and congruent series that this simplification is achieved. In our case, we decompose the Ksi function into irreducible series and we calculate the congruent function associated modulo this reference series. This method finally has the advantage of taking into account the Zeta function constraints through the Ksi function. The process allows us to move away from the Zeta function, by emphasizing the Ksi function instead, which is easier to analyze, as a two Gamma function ratio. The whole process takes place thanks to Information Technology. In the critical strip, the theoretical approach therefore consists in decomposing the Ksi function of the functional equation into amplitude and phase elements. The phase is broken down into a power series. This power series is filtered modulo a power series in (x 1/2) so as to dissolve the axis − of symmetry.
2.3. Illustration Strategy The illustration strategy of the article here consists in modifying the Zeta function by successive transformations that are linear combinations (function of s) of the real part a and the imaginary part b so as to reach a zero-function target au + bv = 0. Theoretical approaches use universal and powerful mathematical instruments and sometimes neglect the intrinsic numerical properties of the Zeta function, the Ksi function of the Riemann functional equation or the Gamma function. Indeed, certain elements of these functions are not easily elucidated by mathematical tools, such as for example the fractional part of variables or the angle oriented in complex space. Numerical computation, whilst incapable of demonstrating anything formally, opens theoretical horizons and bridges an important gap in mathematical analysis, because it Information 2020, 11, 237 4 of 31 Information 2020, 11, x FOR PEER REVIEW 4 of 32
2.4.readily Extensive crowbars Use of itself Power into Series even Expansions the most stubborn of functions, unveiling their mysteries. It is the subsidiarityThere are insu manyfficiency works of theon Zeta,numerical Ksi and methods Gamma for functions Zeta function that this zeros: article The attempts computer to rectify.made it possible to calculate specific values with high precision [21]. In this article, we place ourselves at the 2.4. Extensive Use of Power Series Expansions confluence of numerical approaches and theoretical formulations: We calculate and visualize curves and surfaces,There are which many we works illuminate on numerical with power methods series. for ZetaIndeed, function the polynomial zeros: The form computer of a function made it possibleremains tomore calculate intuitive specific than values its conventional with high precisionmathematical [21]. Inexpression. this article, For we example, place ourselves the Gamma at the function’sconfluence local of numerical behavior approachesis less easily and understood theoretical in formulations:the integral theoretical We calculate form and visualize curves than and surfaces, 2 which⁄ ( we⁄) ½ illuminate 1 + with+ power− series.⁄ + Indeed, the polynomial form of a function in the form ( ) ( ) . Moreover, our approach is above all remains more intuitive than its conventional ! mathematical ! expression. For example, the Gamma a question of evaluating the influence and the symmetry of the abscissa under theR configuration ∞ tz 1e tdt function’s local behavior is less easily understood in the integral theoretical form 0 − − than of several variables ( = , ½ = −½, =1− ) compared to its ordinate . The state of fusion, z 1/2 1 1 139/5 1 in the form √2π/e(z/e) − 1 + + + . Moreover, our approach is above all or on the contrary of separability,12 zof the2!( 12variablesz)2 − 3!(12 inz)3 a formulaO z4 is also more practical to apprehend a question of evaluating the influence and the symmetry of the abscissa x under the configuration of when this appears under the configuration of power series expansions. The originality of this article 1 stemsseveral from variables this reconciliationx = x, x1 2 between= x /2 ,numericalx = 1 xcalculationcompared and to itsthe ordinate formula yreadability.. The state of fusion, 0 / − 1 − or on the contrary of separability, of the variables in a formula is also more practical to apprehend when2.5. Mathematical this appears Notation under the configuration of power series expansions. The originality of this article stems from this reconciliation between numerical calculation and the formula readability. Where possible, classical mathematical notations have been selected. To lighten the writing, it is necessary2.5. Mathematical to define Notation some original notations.
2.5.1.Where Notations possible, for the classical Zeta Function mathematical notations have been selected. To lighten the writing, it is necessary to define some original notations. Simplification of writing a ratio: = 1⁄(2 ) = 12 ; =1 ⁄(24 ) = 1⁄ 24 2.5.1. Notations for the Zeta Function Integer part and fractional part of1 a real: =[ ] + { } ;1 ∈ ℝ; [ ] ∈ℕ;0≤{ } <1 Simplification of writing a ratio: 2π = 1/(2π) = 1/2π; 24y = 1/(24y) = 1/24y IntegerThe domains part and and fractional associated part points of a real: are apresented= [a] + a in; Tablea R; 1[ aand] NFigure; 0 a2. < 1 { } ∈ ∈ ≤ { } TheReal domains part, imaginary and associated part of points a complex: are presented = + = in Table ℜ(1 )and+ Figureℑ( ); ∈ℂ; , ∈ℝ2. Real part, imaginary part of a complex: s = x + iy = (s) + i (s); s C; x, y R Conjugate of a complex number: ̅ = − ; ̅ ∈ℂ < = ∈ ∈ Conjugate of a complex number: s = x iy; s C − ∈ SymmetricSymmetric of of a a point point s: sˆ: = ̂ 1=1−s=1− − s = 1 x iy − − − OppositeOpposite of of a a point point s (conjugates (conjugate of theof the symmetric symmetric point): point):sˇ = 1̌ =1− s = sˆ = = 1̂̅ =1− + x + iy − −
FigureFigure 2. 2. CriticalCritical strip strip and and critical critical line line on on the the Cℂ plane, with the four associated points.points. Table 1. The four associated points: M, Mˆ , M, Mˇ . Table 1. The four associated points: , , , . M := s = (x, y) M := s = (x,y) ∶= =( ,− )− ∶= =( , ) Mˆ∶=:= 11 − s = =( (11 −x, ,y − )) ∶=M ˇ :∶== M( ˆ 1− : = ( 1 ) =(1− , )s) = (1 x, y) − − − − − Homothety of the imaginary axis : = ⁄ 2 HomothetyCritical strip: of the ∶=0≤ ≤1; ≠1 imaginary axis y: t = y/2π Critical strip: := 0 x 1; s , 1 A point in theS critical≤ strip:≤ = + ; ∈ ℂ, ≠ 1; , ∈ ℝ; 0 < < 1 ; > 0 A point in the critical strip: s = x + iy ; s C, s , 1; x, y R; 0 < x < 1 ; y > 0 Critical line: ℒ∶= =½ ∈ ∈ Critical line: := x = 1/2 A point on theL critical line: = ½ + ; ∈ ℝ Critical strip, except the critical line: \ ℒ∶=0≤ ≤1; ≠½; ≠1
Information 2020, 11, 237 5 of 31
A point on the critical line: 𝓈 = 1/2 + iy ; y R ∈ Critical strip, except the critical line: := 0 x 1; x , 1/2; s , 1 P s S\ L ≤ ≤ Zeta function: ζ(s) = ∞ n . n=1 − Dirichlet eta function: η(s) = ζ(s) 1 2 s+1 . − − In the critical strip, the series ζ(s) is divergent; the analytical extension by η(s) is used. The function Ksi is written: ξ(s) = 2sπs 1 sin(πs/2)Γ(1 s). − − The Zeta function satisfies the functional equation: ζ(s) = ξ(s)ζ(1 s) = ξ(s)ζ(sˆ). − 2.5.2. Anamorphosis of the y Axis It produces a stronger elongation of the y axis as y grows: y y Anamorphosis of y: ey = log = t(log t 1) 2π 2πe − 2.5.3. Digressive Series u˘ We define by the adjective “digressive” a series that exhibits a beginning similar to a known series, but which varies noticeably with the subsequent terms. For example, the cosine series is written: 2 4 6 2n cos x = 1 x + x x + ... + ( 1)n x + ℴ x2n+1 . We will then note, in order to simplify the writing, − 2! 4! − 6! − (2n)! a digressive cosine:
2 4 6 2n a2x a4x a6x n a2nx + cos˘ (x) = 1 + + ... + ( 1) + x2n 1 − 2! 4! − 6! − (2n)!
We obviously require the sequence an to have “nice” properties so as not to degrade the meaning h i of proximity with the cosine function. The notation cos˘ only recalls the behavior analogous to a cosine, but it is clear that it is not the cosine function. In the same form, we will note, in a mnemonic way, the digressive exponential with its argument:
a x a x2 a x3 a xn e˘x = 1 + 1 + 2 + 4 + ... + n + (xn); arg(e˘x) = x˘ 1! 2! 3! n!
e˘iθ = cos˘ θ + i sin˘ θ
Pay attention throughout the remainder of this article to the modulus and argument (angle) in question. Indeed: ix ix iε2 ix If e˘ = e + ε1e ; then arg e˘ = x˘ = x + ε3; but ε3 , ε2
→ → If e˘ix = eix + ε eiε2 = 1 then ε 2 + 2ε eix.eiε2 = 0 and e˘ix = ei(x+δx) k k k 1 k 1 1 Again, these mnemonic notations are simply a notational convention and only refer to a specific power series and have no computational meaning.
2.5.4. Symmetric Polynomial, Symmetric Series
In this article, we call a symmetric polynomial P around x = 1/2 in x and (1 x), a polynomial − such that x : P(x) = P(1 x); for example, polynomials of the type: ∀ − X n n n P(x) = an x + (1 x) + bn(x(1 x)) − −
A sequence an , symmetrical around a C, obeys a similar property: h i ∈ X n X n z C : f (a + z) = ∞ an(a + z) = f (a z) = ∞ an(a z) ∀ ∈ n=0 − n=0 −
1
1
Information 2020, 11, 237 6 of 31
3. Materials and Preparatory Calculations It is assumed that the reader is already familiar with the fundamentals of mathematics. There are many books that the reader can consult: a classic or standard book, which can be recommended to the reader interested in the specific subjects, is proposed in each following paragraph. It is crucial to understand the behavior of the Zeta, Ksi and Gamma functions in the critical strip, in particular the values for which the real and imaginary parts cross zero, and the angles between the two parts, because they play an essential role in the RH. In this article, we use basic theorems to dissipate the illustration’s complexity. We therefore use the following theorems, results and formulas: Bézout’s theorem, Stirling’s formula, Taylor’s formula, and operations on complex numbers. In order to facilitate the reading of the following sections, preparatory calculations by computer are grouped in this section. The reader can skip the development of the calculations at first reading, although they bring together the formation of the tools necessary for reasoning and understanding latter sections. In addition, this section represents a more substantial part of the computer work of this article. The study of the Ksi function’s numerical behavior requires power series expansions as a function of y, assuming that x is small compared to y. These computational developments require rigor and significant efforts; the calculations were checked by computer. We use the Python language [22] and the SciPy ecosystem [23] for all the computations and we visualize them using the matplotlib library [24]; we use mpmath [25] as an arbitrary-precision substitute for Python’s float/complex types and for functions (Gamma, Zeta), with arbitrary precision.
3.1. Bézout’s Identity The book [26] is intended as an introduction to the subject of Commutative Algebra, not a substitute for the treatise such as Bourbaki. A geometrical viewpoint of the Bézout’s theorem and convergent series rings is provided in the book of Chenciner [27]. Bézout’s identity is the result of elementary arithmetic, which proves the existence of solutions to the linear Diophantine equation: au + bv = gcd(a, b) of unknown integers u and v, where a and b are relative integer coefficients and where gcd(a, b) is the greatest common divisor of a and b. Let a and b be two relative integers ( Z) and their gcd(a, b) = d; then there are two integers u and v such ∈ that au + bv = d. Étienne Bézout generalized this result, notably to polynomials, and more generally the result can be stated on any principal ring. This theorem echoes Hilbert’s zero theorem (Hilbert’s Nullstellensatz). Let K be an algebraically closed field, let a , ... , am K[X , ... , Xn] be polynomials 0 ∈ 1 without common zeros. Then there exists u , ... , um K[X , ... , Xn] verifying a u + ... + amum = 1. 0 ∈ 1 0 0 Our approach, adjacent to this identity, consists in trying to write the functional equation ζ(s) = a + ib = ξ(s) ζ(sˆ) = (p + iq) aˆ + ibˆ = paˆ qbˆ + i pbˆ + qaˆ in the form of two equations − au + bv = c and i(au0 + bv0) = ic0. ζ and ξ are then series, and we study the divisibility of these series and their congruence with reference to a pivotal series.
3.2. Taylor Function Standard text books on Mathematical Analysis are those by Rudin [28] and especially by Whittaker and Watson [29], which contain a solid mathematical discussion of the transcendental functions. If the function f with complex values is differentiable in z up to the order n 1, then the ≥ Taylor–Young function is written as:
(2) (n) f 0(z) f (z) f (z) f (z + h) = f (z) + h + h2 + ... + hn + ℴ( hn) 1! 2! n!
The main part of the power series expansion from f to z to order n is the polynomial Pn:
X (k) n f (z) k Pn(z + h) = h k=0 k!
1
Information 2020, 11, 237 7 of 31
The Taylor series from f to z is defined as the power series whose nth partial sum is equal to Pn, for all integers n. The Taylor series of f at this point z is the series of functions:
X f (n)(z) f (z + h) = ∞ hn n=0 n! This series is used for theoretical reasoning. However, the nth-order expansions are numerically used to reach numerical accuracy. In this article, the functions are holomorphic and indefinitely differentiable. The power series expansions makes it easier to find the angles between real and imaginary surfaces. The following power series expansions are used:
a(a 1) a(a 1)..(a n + 1) (1 + z)a = 1 + az + − z2 + ... + − − zn + ℴ( zn) 2! n!
z2 zn ez = 1 + z + + ... + + (zn) 2! n! z2 z4 z2n cos z = 1 + ... + ( 1)n + z2n+1 − 2! 4! − (2n)! z3 z5 z2n+1 sin z = z + ... + ( 1)n + z2n+2 − 3! 5! − − (2n + 1)! During the calculations, compound series appear. These series are specifically the seeds of divisions of power series. Some are listed below:
2 3 4 5 6 7 8 z 7z 3z 11z 13z z 17z 19z e− ( 3 + 2z) = 3 5z + + + + ... − − − 2 − 2 24 − 120 48 − 5040 40320
2 3 4 5 6 7 8 z 2 15z 35z 21z 33z 143z 13z 17z e− 1 4z + 4z = 1 + 3z + + + ... − − − − 2 6 − 8 40 − 720 336 − 2688 3 4 5 6 7 8 9 1 z 3z 2 5z z 7z z z z 11z e− ( 2 + z) = 1 + z + + + ... − 2 − − 2 − 12 8 − 240 180 − 1120 8064 − 725760 3 4 5 6 7 1 z 2 2 10z 11z 49z 17z z e− 2 8z + 3z = 1 + 3z 5z + + + ... − 2 − − − 3 − 8 120 − 180 56 In this article, we use power series expansions extensively, because they are simple to interpret. On the other hand, they are long and complicated to calculate for the Gamma function and the Ksi function, especially when we try to highlight and understand their behavior according to the abscissa x under the configuration of several variables x = x, x1 = x 1/2, x = 1 x with respect to its 0 /2 − 1 − ordinate y.
3.3. Power Series The reader can obtain more information about series and Complex Analysis in References [30,31], where a comprehensive treatise on holomorphic functions is given. P n A power series a (series of positive integer powers) is a series of functions of the form: anz h i where an is a complex sequence. Series is associated with its radius of convergence R( a ) = h i P n + h i sup z z C, anz converge > R + . The indefinitely derivable and expandable functions {| | ∈ } ∈ ∪ { ∞} P n in power series at point z0 are written in the neighborhood of point z0 as sum f (z) = n∞=0 anz of the power series of the variable z z , and this is then their Taylor series. A function of the complex − 0 variable, defined on an open set U (subset of a topological space which contains no point of its border), is said to be analytical on U when it admits a power series expansion in the neighborhood of every point of U. Such a function is indefinitely differentiable on U and called holomorphic. The product of two power series is defined, using the Cauchy product of series with complex terms:
1
Information 2020, 11, 237 8 of 31
X X X Xn X ∞ n ∞ n ∞ n ∞ n anz bnz = akbn k z = cnz n=0 n=0 n=0 k=0 − n=0 R( c ) in f (R( a ), R( b )), h i ≥ h i h i P n P n P n In this case, the division of two power series is: n∞=0 anz = n∞=0 cnz / n∞=0 bnz P n P n P n The series are then congruent: ∞ anz ∞ cnz mod ∞ bnz n=0 ≡ n=0 n=0 The main parts, to the nth order, are also congruent: An(z) Cn(z) mod Bn(z) ≡ 3.4. Bernoulli’s Numbers and Bernoulli’s Polynomials
Bernoulli’s numbers n are involved in the Gamma function and in the Ksi function. These are B n t = P t the coefficients of the power series of et 1 n∞=0 n n! . − B 1 1 1 1 = 1; = 1/2; i f k > 0 : + = 0 ; = ; = ; = ; = ; B0 B1 − B2k 1 B2 6 B4 − 30 B6 42 B8 − 30 B10 = 5 ; = 691 ; = 7 ; = 3617 ; = 43867 . 66 B12 − 2730 B14 6 B16 − 510 B18 798
n 1 n 1 2(2n)! 2n = ( 1) − 2n ; 2n ( 1) − B − |B | B ∼ − (2π)2n
n(x) are the Bernoulli polynomials: B X cos(2kπx nπ/2) n(x) = 2n! ∞ − B − k=1 (2kπ)n
2 3 2 (x) = 1; (x) = x 1/2; (x) = x x + 1/6; (x) = x 3x /2 + x/2; B0 B1 − B2 − B3 − (x) = x4 2x3 + x2 1/30; (x) = x5 5x4/2 + 5x3/3 x/6. B4 − − B5 − − 3.5. Group of Similarities: Homothety and Rotation of a Complex Function Let z be a complex number. In this article, we consider homotheties, with center O and ratio k , k R : (z) = kz. We also consider the rotations of a holomorphic function. These are rotations ∈ H R with center O and angle θ: (z) = zeiθ. Homotheties and rotations in the complex plane constitute R a subgroup of the group of similarities. y (x 1/2) − − In this article, we consider a homothety of ratio k = 2π We also use four rotations of center O and respective angles:
2 π 1 (x 1/2) θ1 = ; θ2 = 2π ey ; θ3 = ; θ4 = − . 4 − −24y − 2y
The rotations of the Zeta function have the property of preserving the position of non-trivial zeros: ζ(s) = 0 ζ(s) eiθ = 0. ζ(s) , 0 ζ(s) eiθ , 0. In the case of non-trivial zeros, L’Hôpital’s ⇒ × ⇒ × rule applies: 0 = ξ(s) 0 ζ (s) = ξ(s) ζ (1 s) . × ⇒ 0 × 0 − 3.6. The Gamma Function For further details on the Gamma function, the reader is referred to the historical book of Artin [32]. The Bernoulli–Euler–Legendre Gamma function is: Z ∞ z 1 t Γ(z) = t − e− dt 0
The Gamma function can be written in a power series form: z 1/2 z+µ(z) Stirling’s formula: z C Z∗ : Γ(z) = √2π z − e− ∀ ∈ \ − Information 2020, 11, 237 9 of 31
P 2k Binet’s function: µ(z) = ∞ B k=1 2k(2k 1)z2k 1 − − A property of the Gamma function is in particular: Γ(z) = Γ(z). The Gamma function quickly takes very low values, as y increases in the critical strip: Information 2020, 11, x FOR PEER REVIEW 68 9 of137 32 Γ(1/2 + i100) = ( 1.09179 + 1.04964i) 10 ; Γ(1/2 + i200) = (3.88183 8.28654i) 10 . − × − − × − In order to analyze the local behavior of the Gamma function, it is preferable to consider the Γ( )⁄|Γ( )| since it is the angle between both surfaces or curves that comes into play (Figure 3). The ( ) ( ) functionGamma Γfunctionz / Γ z intervenes since it is by the the angle values between 2⁄ bothand surfaceŝ⁄2 with or the curves values that for comes in intothe interval play (Figure [0,1]3).. The Gamma function intervenes by the values s/2 and sˆ/2 with the values for x in the interval [0, 1].
Figure 3. The Gamma standardized surface (real part and imaginary part), in the critical strip. Figure 3. The Gamma standardized surface (real part and imaginary part), in the critical strip. The The surfaces cancel each other out in turn and almost periodically. The functional equation uses the surfaces cancel each other out in turn and almost periodically. The functional equation uses the variable s/2 for Gamma. variable 2⁄ for Gamma. Undulations of the Gamma Function in the Critical Strip Undulations of the Gamma Function in the Critical Strip It is important to grasp the behavior of the Gamma function with x in the interval [0, 1] to identify the constraintsIt is important of the to Zeta grasp function. the behavior The Gamma of the function Gamma behaves function similarly with toin thethe complex interval function[0,1] to identify= ( )the= constraintsz = ( + of)( xthe+iy )Zeta function. The Gamma function behaves similarly to the complex f : f z z x iy and( plays ) a major role in the RH. Indeed, in Stirling’s approximate function ∶= ( ) = = ( + z 1 ) and plays a major role in the RH. Indeed, in Stirling’s formula Γ(z) √2π/z (z/e) e 12z , the element √z plays an essential role, because it disrupts the ≈ Γ(z) ≈ 2 ⁄ ( ⁄) symmetryapproximate in theformula function ξ between s and sˆ = 1, thes. element √ plays an essential role, because it disrupts the symmetry in the function between− and ̂ =1− . In the critical strip x , 1, y , 0, the Gamma function is never canceled out. At constant y, ≠ 1, ≠ 0 the argumentIn the critical (angle) strip of Γ varies along, xthe. If Gamma the surfaces functionΓ(x, yis) arenever displayed, canceled the out. angle At constant of the Gamma, the Γ Γ( , ) functionargument ripples (angle) out of between varies alongπ and π. alongIf the thesurfacesy axis, but it is are important displayed, to the note angle that theof the ripples Gamma are −− obliquefunction to ripples the axis outx. between The Gamma functionand along is therefore the a axis, function but it that is important locally nests to note the variablesthat the ripplesx and yareinextricably. oblique to the axis . The Gamma function is therefore a function that locally nests the variables and inextricably. 3.7. The Ksi Function of the Functional Equation 3.7. The Ksi Function of the Functional Equation The Ksi function (Figure4) has remarkable properties that should be elucidated if we want to understandThe Ksi the function RH. The (Figure Ksi function 4) has remarkable manages the properties symmetry that in sshouldand sˆ =be1 elucidateds of the Zetaif we function: want to − ζunderstand(s) = ξ(s)ζ (thesˆ). TheRH. functional The Ksi relationship,function manages presented the insymmetry various forms,in isand as follows: ̂ =1− of the Zeta function: ( ) = ( ) ( ̂). The functional relationship, presented in various forms, is as follows: (s) (1 2 s ) ζ ( ) s Γ /( /2⁄) ξ(s) = = π Γ ½− − 2 ( ) =ζ(sˆ) = Γ(1/2)Γ(s/2) ( ̂) Γ(½)Γ( 2⁄)
Information 2020, 11, 237 10 of 31 Information 2020, 11, x FOR PEER REVIEW 10 of 32
Figure 4.4. Above:Above: KsiKsi functionfunction in in the the critical critical strip: strip: the the real real part part and and the imaginarythe imaginary part cancelpart cancel each othereach outother periodically out periodically and alternately. and alternately. Below: Below: Ksi modulus Ksi modulus and argument. and argument. The Ksi The modulus Ksi modulus is equal is to equal 1 for xto= 1 for1/2. =½ The surface. The surface obliquity obliquity (angle (anglearg(ξ) ) is small:( )) is Thesmall: undulations The undulations are not are strictly not strictly parallel parallel to the realto the axis. real The axis. drawing The drawing might might make make the reader the re thinkader thatthink the that angle the angle does notdoes depend not depend on x: thison is: this not true.is not Thetrue.x ,Thearg( ξ ,) linkage( ) linkageis very weak,is very but weak, is nevertheless but is nevertheless responsible responsible for the RH. for the RH.
However, toto analyze analyze the the synchronization synchronization of theof zeroingthe zeroing of the of real the and real imaginary and imaginary parts, the parts, relation the canrelation be written can be symmetrically:written symmetrically: Z Z ∞ ⁄ ∞ X 2 ⁄ F( s() =) = θ ( ) (t)ts/2 1dt == ∞ e πn tts /2 1dt − − − 0 0 1
s/ 2⁄ sˆ/ 2 ̂⁄ F (s() )==ζ( s() Γ)(Γs(/ 22⁄))π− == F(sˆ)( =̂) = ζ(sˆ)(Γ ̂)(Γsˆ/( 2̂⁄))2π− It isis essentialessential toto focus focus on on the the various various components components of of this this function functionξ to observeto observe the formationthe formation and gestationand gestation of this of symmetry.this symmetry. The The complex complex formation formation of symmetry of symmetry is in is fact in fact located located at the at corethe core of the of Gammathe Gamma function, function, which which intermingles intermingles the variables the variablesx and y andso vigorously so vigorously that it is that naive it is to naive hope to to hope split allto split the components all the components of this function of this function into both into independent both independent parts, one parts, function one offunctionx, and theof other, and of they. s 1/2 Γ(sˆ/2) otherThe of set-up. used in this article is: ξ(s) = π − . The function ξ, which is within a factor, Γ( (̂⁄)s/ 2) a GammaThe set-up function used ratio, in this intermingles article is: the ( variables) = ½x and. yThe, but function the local angle, which between is within both a factor, surfaces a ( ⁄) (realGamma and function imaginary) ratio, shows intermingles a lesser degreethe variables of dependence and compared, but the local to x angle. between both surfaces (real Onand the imaginary) critical line, shows the a ratiolesser becomes degree of the dependence ratio of two compared conjugated to Gamma. functions, and the modulus of the function ξ is 1. The function ξ then behaves like undulations which vary between 1 On the critical line, the ratio becomes the ratio of two conjugated Gamma functions, and the− andmodulus1. It isof not, the however,function a perfectis 1. The classic function trigonometric then behaves function. like The undula functionaltions which relation vary distorts between the relationship−1 and 1. It betweenis not, however, the Gamma a perfect functions, classic and trigonom therefore,etric thefunction. result The between functiζonal(s) and relationζ(sˆ), distorts outside the criticalrelationship line. between In the Ksi the formula, Gamma the functi ratioons, of and Gamma therefore, functions the result severely between destroys, ( ) along and the ( ̂)x, axis,outside the the obliquity critical ofline. the In undulations the Ksi formula, of the the Gamma ratio of function, Gamma but functions this does severely not disappear destroys, completely. along the This axis, isa the primary obliquity result of of the this analysisundulations of the of Ksi the function. Gamma In function, addition, but the inversethis does of thisnot functiondisappear is: 1 ξcompletely.(s)− = ξ(sˆ This). In is the a primary decomposition result of of this Ksi intoanalysis elements, of the weKsi must function. be able In addition, to find the the inverse inverse elements of this andfunction better is: understand ( ) = ( hoŵ). In symmetry the decomposition works. of Ksi into elements, we must be able to find the inverse elements and better understand how symmetry works.
Information 2020, 11, 237 11 of 31
3.7.1. Serial Expansion of Ksi One of the major elements of this article is the power series expansion of the Ksi function (if x is small compared to y).
s 1/2 √s µ(sˆ/2) µ(s/2) ξ(s) = (2πe) − P(s)Q(s); P(s) = s ; Q(s) = e − √ssˆ
Indeed, it is with the symmetry study in x and 1 x around x = 1/2 that the RH can be grasped. − The power series expansion is composed of several pieces: a main part P, that of an expression in s, multiplied by an exponential part Q of lower importance, but more sophisticated, with the intervention of the Binet formula of Bernoulli terms. The calculation of the power series expansion of this formula is very laborious, but this calculation makes it possible to exhibit the angle between the real and imaginary parts. This angle depends on x and y, and this connection occurs differently between s and sˆ. It is this dissimilarity that prevents finding a Bézout relationship in the critical strip, except on the critical line. The power series expansion consists of the product of two factors P and Q. s A first expression corresponds to the polynomial extracted from the expression √s/ √ssˆ .
√iy √1 ix/y P(s) = − s ys √1 (x(x 1)/y+2i(x 1/2))/y − − − s 1/2 1 /2 1 2 ix x(x 1) 2i(x /2) − = eiπ/4 y / s 1 1 − + − − − y − y2 y
x = x; x1 = x 1/2; x = x 1 0 /2 − 1 − p p !1/2 !! s/2 iy 1 ix0/y ix x x 2ix1/2 − P(s) = − = eiπ/4 y1/2 s 1 0 1 0 1 + q s − 2 s − y − y y y 1 x x /y + 2ix1 /y − 0 1 /2 The second expression Q corresponds to the exponential of Binet’s formula:
Q(s) = exp(∆µ)
X 2k 1 2 − 2k 1 2k 1 2k ∆µ = µ(sˆ/2) µ(s/2) = ∞ B sˆ − s − − k=1 2k(2k 1) − − We can extract the expression for the power series expansion when x is small compared to y: X ( )k+1( )2k 1 !1 2k !1 2k 1 2/y − 2k ix − i(1 x) − ∆µ = i ∞ − B 1 + 1 + − k=1 2k(2k 1) − y y − The expression ∆µ is symmetrical in x and x 1: The inverse of Ksi comes under the expression − ∆µ. The formula of Ksi then becomes: −
1/2 s iπ y − iq(x,y) ξ(s) = e 4 p(x, y)e 2πe
3.7.2. Expansion of the Main Part p
s !1/2 ! ix x(x 1) 2i(x 1/2) − 2 p(x, y) = 1 1 − − − y − y2 − y Information 2020, 11, 237 12 of 31
This is where a delicate but essential issue comes in. The expansion p(x, y) actually contains the (x 1/2) (x 1/2) series e− − which should be extracted. By dividing the series p(x, y) by the series e− − , we obtain the series p(s). A laborious calculation of p(s) gives the following fourth order expansion:
(x 1/2) p(x, y) = e− − p(s)
ip1 p2 ip3 p4 p(s) = p0 + + + + + ... y y2 y3 y4
p0 = 1 1 1 p = x(1 x) 1 −4 − 2 − 1 1 p = 5 + 3x + 6x2 + 2x3 3x4 2 24 4 − − ! 1 1 57x x2 9x4 p = 8 + + 8x3 + x5 x6 3 48 8 − 4 − 4 2 − − ! 1 29 49 25 13x 55x2 7 17x3 79x4 77x5 2x6 x8 p = × + + × + x7 + 4 96 2 − 960 − 8 2 12 − 8 − 10 − 3 4
y 1/2 s − Thus, when multiplying by this factor, the expression 2πe turns out to be:
1/2 x iy 1/2 x iy y − − (x 1/2) y − y − e− − = 2πe 2π 2πe
y iy iπ y iy iπ 1 y y − 4 − 4 i2π( 8 2π log 2πe ) The multiplication of 2πe by e also is simplified: 2πe e = e − y y We find the expression of an anamorphosed y: ey = log 2π 2π e Considering the fractional part, it follows that: ey = [ey] + ey . Let α be the angle α = 2π( ey ). y 1/2 x iy− y 1/2 x − − (x 1/2) − iα The expression 2πe e− − therefore becomes: 2π e .
3.7.3. Expansion of Exponential Part q
X ( )k+1( )2k 1 !1 2k !1 2k 1 2/y − B2k ix0 − ix1 − q(x, y) = ∞ − 1 + 1 k=1 2k(2k 1) − y − y − It is essential to calculate the polynomial q(x, y) = eiq(x,y) using the series expansion in series z z z2 z3 q e = 1 + 1! + 2! + 3! + ... for each of the terms of the sum. The laborious calculation of (s) using the exponential of Binet’s formula gives the order 6:
iq1 q2 iq3 q4 iq5 q6 q(s) = 1 + + + + + + + ... y y2 y3 y4 y5 y6
1 q = 1 3 1 1 q = (x + x ) 2 −18 − 6 0 1 31 1 q = x + 3x2 + x + 3x2 3 810 − 18 0 0 1 1 139 1 q = + 62x + 75x2 + 180x3 62x + 30x x + 75x2 + 180x3 4 −81 120 9 120 − 0 0 0 − 1 0 1 1 1 × × Information 2020, 11, 237 13 of 31
43x2 x3 x2x 43x2 x x2 x3 9871 1 139x0 0 0 4 139x1 x0x1 0 1 1 0 1 1 4 q5 = + + + x + + + + + x 36280 6− 810 − 60 2 0 − 810 18 6 − 60 6 2 1
1189x2 263x3 7x4 5x2x q = 2251 + 1 9871x0 + 0 + 0 0 x5 9871x1 + 67x0x1 0 1 6 − 36527 6 − 23357 23345 216 − 12 − 0 − 23357 540 − 72 x3x 1189x2 5x x2 x2x2 263x3 x x3 7x4 0 1 + 1 0 1 0 1 + 1 0 1 1 x5 − 6 23345 − 72 − 6 216 − 6 − 12 − 1 i 1 1 q(s) 1 + + x0 + x1 3y − 6y2 3 The polynomial q(s) is a symmetric polynomial at x = x and x = x 1. Its inverse is obtained 0 1 − by changing x to 1 x and y to y, (modulo a monomial in x1 = x 1/2)). − − /2 − 3.7.4. Final Expression of Ksi
y 1/2 x The expression of Ksi eventually becomes: ξ(s) = − eiα (s) 2π A (s) is a polynomial with complex coefficients, which we calculate: A ia a ia (s) = p(s)q(s) = 1 + 1 + 2 + 3 + ... A y y2 y3
1 x x2 1 1 a = + = (x(1 x)) 1 12 − 2 2 12 − 2 − 2 1 x x2 x3 x4 1 1 + x + 3x a2 = + + = + x(1 x) − −288 − 24 12 12 − 8 −2 122 − 24 × 139 x 19x2 x3 5x4 x5 x6 a3 = + + 5 3!123 − 192 − 576 36 96 − 48 − 48 × 571 139x 67x2 101x3 7x4 91x5 x7 x8 a4 = + + + + −5 4!124 − 20736 12960 3456 − 768 − 2880 96 384 × Finally, taking into account all the elements, we can present the final formula, by separating the influence of x as follows:
2i 2 4 139i 2 571 (s) = 1 + + × × A 24y − (24y)2 15(24y)3 − 15(24y)4
2 1 11x 2 3 4 x(1 x) i 1+x+3x 2 3 x +4x +2x + − + + i − − − y 2 − 24y 4 24y2 − × 2 3 ! 139 53x + 319x + 107x 5x4 5x5 x6 + − 54 − 90 30 15 − − − 16 24y3 ×
With the expression of this polynomial , we discover that it is possible to perform two digressive A 1 rotations. The first is the rotation 3 of angle θ3 = 24y , and a fourth rotation 4 (which will not be R 2 − R (x 1/2) sensitive on the critical line of angle θ = − ). These rotations are digressive, because there is 4 − 2y a series as a remainder. With these rotations, all computations carried out, and the polynomial A finally becomes the polynomial : K Information 2020, 11, 237 14 of 31
(s) K 2 i i(x 1/2) 2 2 − i(x 1/2) (x 1/2)( 1+10x 8x ) = e− 24y − 2y + − + − − − y 48y2 7 5 23x2 3 4 5 i +(x 1/2) x+ x +x 2x + − 60 − − 12 − 6 − − 48y3 7 133 2753x 707x2 181x3 87x4 5 6 +(x 1/2) + + 5x +12x + − 120 − 720 360 60 − 10 − 10 − 242 y4
1 2 1 2 1 2 i i(x /2) y / x y / x − − The final Ksi formula is therefore: ξ(s) = − eiα (s) = − eiα e˘ 24y e˘− 2y 2π K 2π
3.7.5. Special Case of Ksi: x = 1/2 We can resume the calculations of p, q, r = pq taking into account the angle of rotation θ = 1 , 3 − 24y in order to reveal the digressive rotations:
3i 32 139i 3 52 19 32851i 592153 3 6090313i 473920289 p = 1 + + × × + × + + ... − 8y − 2!(8y)2 3!(8y)3 4!(8y)4 − 5!(8y)5 − 6!(8y)6 7!(8y)7 8!(8y)8
2 (247+33)i (263 7+34) (2 32 132+35)i p = 1 3i 3 + + × × × − 8y − 2!(8y)2 3!(8y)3 4!(8y)4 − 5!(8y)5 6 4 7 (64 9241+3 ) (2 3 107 3557+3 )i 27379 9769+38 × + × × + × − 6!(8y)6 7!(8y)7 8!(8y)8 3i 3i 7 2i 7 12 1019i 9241 3 107 3557i 379 9769 p(y) = e˘ −8y = e −8y + × + × + × × + × 3!y3 4!y4 − 5!y5 − 6!y6 8 7!y7 2 8!y8 × × i 1 (73/10)i 131/5 (2921/14)i 97423/70 (481601/20)i 214939 q = 1 + + + + + ... 3y − 2!(3y)2 − 3!(3y)3 4!(3y)4 5!(3y)5 − 6!(3y)6 − 7!(3y)7 8!(3y)8 i i 7 9i 18 7 9 323i 97283 481561i 214181 q(y) = e˘ 3y = e 3y × + × + × + − 10 3!y3 5 4!y4 14 5!y5 − 70 6!y6 − 20 7!y7 8!y8 × × × × × 4 2 6 2 i 1 (2 3 7 5)i 26327 5 (2 3 29 10079+7)i r = pq = 1 − − × 24y 2!(24y)2 5 3!(24y)3 5 4!(24y)4 7 5!(24y)5 − − − × − × − × 253217807+35 (243243 11009087 5)i × − 35 6!(24y)6 5 7!(24y)7 − × − × 2732132131 23027 5 × − 5 8!(24y)8 − × The final solution therefore provides: θ 2 17807θ 2i 29 10079θ 3 43 11009087θ 4i r iθ3 3 3 3 3 = e 5y 7i 7θ3 + 7 2! ×35 3! + × 35 4! − − × − × × 2 5 13 131 23027θ3 7× 5! ... − ×
1 1 1 iπ /2 /2 /2 y − iq(1/2,y) y − iα iα iθ ξ(𝓈) = e 4 p(1/2, y)e = e r(y) = e e˘ 3 2πe 2π × 2 7iθ 2 1/(7 24 ) In total: ξ( ) = eiαe˘iθ3 = ei(α+θ3) 1 3 1 + iθ + 17807/49 θ 2 × iθ 3 + ... − 5y 3 2! 3 − 3! 3 2 More approximately: ξ( ) = eiαe˘iθ3 ei(α+θ3) 1 7iε eiθ3 − 5y
1
Information 2020, 11, 237 15 of 31
1 1 As ξ (s) = ξ(1 s), we can verify that if x = 1/2, ξ(𝓈) − = ξ( ) or ξ( )ξ( ) = 1. We verify − − numerically that r r = 1. × ( )i ( )i r r = (1 i 1 1003/5 4027/5 5128423/7 168359651/35 × − 24y − 2!(24y)2 − 3!(24y)3 − 4!(24y)4 − 5!(24y)5 − 6!(24y)6 (68168266699/5)i 587283555451/5 ) − 7!(24y)7 − 8!(24y)8 ( )i ( )i (1 + i 1 + 1003/5 4027/5 + 5128423/7 × 24y − 2!(24y)2 3!(24y)3 − 4!(24y)4 5!(24y)5 168359651/35 + (68168266699/5)i 587283555451/5 ) − 6!(24y)6 7!(24y)7 − 8!(24y)8
18 312603562112249 r r = 1 + × + ... × 7 10!(24y)10 × 4. Results This section first presents the computer results and the formula on the critical line and continues with a graphical presentation and a numerical calculation. Then, the section apprehends the calculations in the critical strip in order to prove that the points s of the critical line are located on the limit of a property of the functional relation. Finally, the section presents a geometric interpretation of the functional equation, which clarifies the RH. To understand the arrangements of both surfaces which contribute to the RH, the mechanisms of the joint undulations of the real and imaginary surfaces of the Zeta function and the Ksi function must be demystified. The local behavior of the Zeta and Ksi functions is mainly constructed in relation to the Gamma function. The Ksi function, which brings two symmetric values of the Zeta function into constant y, is essentially a ratio of two Gamma functions. It is therefore essential to analyze and visualize the behavior of the Gamma function and the Ksi function in the critical strip, and to focus attention on the specific case of the critical line. The essential constraint is the functional equation which imposes a strong requirement to constant y, between x and 1 x. − 4.1. Concordant Undulations of the Two Curves ζ( ) and Bézout’s Identity on the Critical Line
In this paragraph, we show that we can find a function ϕ(y) ζ(1/2, y) that is canceled out over ◦ the entire longitudinal section of the critical line. This function depends on y only.
4.1.1. Result Statement On the critical line , there is a congruence, in a Bézout identity form, between the real part L (ζ(s)) = a and the imaginary part (ζ(s)) = b of the Zeta function. < =
y R ; x = 1/2; = 1/2 + i y; ζ( ) = a + ib; u(y), v(y) : a u(y) + b v(y) = 0 ∀ ∈ ∃ 4.1.2. Graphical Approach for the Function ϕ On the critical line, we first try to find a sequence of operations that allows us to graphically compare both curves (real part and imaginary part) of the Zeta function. Then, we deduce a linear combination of the real and imaginary parts that cancel out. We want to preserve the status of non-trivial zeros, which invites us to utilize linear combinations, in particular rotations. First rotation: approximation of the curves • The first rotation (Figure5) is obviously the rotation (θ ) = eiπ/4 that restores the symmetry R1 1 between the values of the real part and the imaginary part, by removing the initial bias of the two 1 1 cos(y log(1))+i sin y log(1) parts due to the first term of the Riemann sum ζ(s) = 1s + 2s + ... = 1x + ... = 1 1 (1 + 0 i) + s ... = 1 + 0 + s ... since cos(0) = 1 and sin(0) = 0. The real part and the imaginary × 2 2
1
Information 2020, 11, x FOR PEER REVIEW 16 of 32