Number Theory — Introduction

Number Theory — Introduction

1. Del Number theory — Introduction Version + ✏ —lastupdate:10/30/1410:32:34AM 1 Preliminary version prone to errors and subjected to changes. The version number says all! If you toss a coin you certainly would suspect (even be sure) that in the long run the number of heads and tails will be the same. In same vain, if you pick numbers randomly in the long run, rather exactly, half the numbers you pick will be even and the other half odd. So there is a two-to-one chance that you pick an even number. Analogously, there will be three-to-one chance that your number will be divisible by 3. But what is the chance that you pick a prime number? The prodigious youngster Gauss pondered over that question at the age about 15. He made long list of primes and counted the number of primes in intervals of the shape [x, x+1000],andthenobservedthatcontainedabout1/ log x primes. He thus answered your question: The answer depends on how big numbers you are allowed to pick, but if you confined your picks to be from an interval of shape [x, x + y] ,the chance is about y/ log x.Ofcoursetheanswerwasconjectural,thecomputationalcapacityislimited even for Gauss. It is common usage to let ⇡(x) denote the prime counting function;thatis⇡(x)= # p p aprime,p x .Itisstepfunctionthatincreasebyoneateachprime.For { | } example do we have ⇡(7) = ⇡(8) = ⇡(9) = ⇡(10) = 4 but ⇡(11) = 5.Inourage of computers the values of ⇡(x) has been computed for very large x,forexampleis ⇡(1022)=201467286689315906290.Oneofthemainobjectivesofanalyticnumber theory is to give approximations of ⇡(x) by functions easy to describe, like classical elementary functions; and of course one wants the approximations to be as good as possible. The cumulative function of the prime distribution is thus ⇡(x),soifthedistribution 1 Number theory — Intro MAT4250 — Høst 2014 of primes goes like 1/ log x,onesuspectsthat x dx ⇡(x) . (7) ⇠ log x Z2 PNT The integral on the right is often denoted by li x and is called the logarithmic integral, that is x dx =lix. log x Z2 The sign in 7 means that the two sides are asymptotical equal:Twofunctionsf(x) and ⇠ g(x) are said to be asymptotical equal, symbolically written f g,iflimx f(x)/g(x)= ⇠ !1 1.Forexamplearef(x)=px2 + a and g(x)=x asymptotical equal (to be precise, one should say “when x ”), and two polynomials are if they have the same dominating !1 term. It does not mean that their difference tends to zero, for example it holds true that x3 + x2 x3 + x, the difference however tends to infinity (it is asymptotic to x2!). ⇠ The statement in (7)hadformanyyearsthestatusasGauss’conjecture,butwas finally proven by Hadamar and de la Vallée Poussin, and is now called the Prime Number Theorem, or PNT for short: Theorem . (The Prime Number Theorem) x dx ⇡(x) =li(x). ⇠ log x Z2 By performing a partial integration with dv = dx and u =1/ log x,onefinds x dx x x dx = + + C log x log x log2 x Z2 Z2 where C is the constant 2/ log 2.Bysplittingtheintervalofintegrationinthetwo − intervals [2, px] and [px, x] (see the figure below) one arrives at the estimate x dx 4x px < + log2 x log2 x log2 2 Z2 which shows that li x x/ log x.Hence,therelation being transitive, one has the ⇠ ⇠ following alternative formulation of the prime number theorem, which may be more speaking as the the function x/ log x is easier accessible than li x,(butitislessprecise): Theorem . (PNT, second version) x ⇡(x) ⇠ log x Our principal goal of the course is to give a proof of the PNT, but as well to get some general understanding of the mathematical lore around the theorem. —2— Number theory — Intro MAT4250 — Høst 2014 Problem .. Show that for any n one has x dt x x (n 1)!x x dt li x = = + + + − +(n 1)! + C. log t log x log2 x ··· logn x − logn+1 t Z2 Z2 Give an explicit expression for the constant C. X Problem .. Show that x dt 2nx px + + C logn t logn x logn 2 Z2 X 2 log− 2 2 4log− x 2 log− t 2 px x t 2 Estimating the logarithmic integral 2 log− tdt R Big O and small o We use the opportunity to introduce some more notation con- stantly used in analytic number theory. If f(x) and g(x) are two functions, we say that f is ‘‘big O” of g, in writing f(x)=O(g(x)) if there is a constant C such that f(x) C g(x) .Wesaythatf is “small o” of g as x a of g—in writing | | | | ! f(x)=o(g(x))—if f(x)/g(x) 0 as x a. ! ! The Riemann ⇣-function Taking the bull by the horns, we introduce immediately the may be most renown func- tion in the whole of mathematics, the Riemann ⇣-function. Encoded in its analytic properties lie many of the secrets of the prime numbers. The ⇣-function was first stu- died by Euler in 1740, and some call it the Riemann-Euler function. Euler proved the property that is the bridge between the Riemann ⇣ and the primes, the so called Euler product. However most the fundamental properties was established by Riemann in his paper xxx where he also states the famous hypothesis about the zeros. —3— Number theory — Intro MAT4250 — Høst 2014 1 1 ⇣(s)= , ns n=1 X where s is a complex variable. It is the custom in this branch of mathematics to let σ =Res an ⌧ =Ims,sothats = σ + ⌧i. Proposition . The series defining the ⇣-function converges absolutely in the right half plane where σ =Res>1.Theconvergenceisuniforminthehalfplanesσ>σ0 > 1 and ⇣(s) is an analytic function of s for σ>0. Proof: Indeed, by the elementary theory of real infinite series one learns in school σ (e.g., the integral criterion) one knows that n 1 n− converges for any real σ>1, s σ ≥ and one has n− = n− . σ | | P N σ M σ0 As n− is a decreasing function of σ, it follows that n=M n < n=N n− whe- never σ>σ0 > 1, and this shows that the convergence is uniform in half planes P P σ>σ0 > 1. The ⇣-function therefore is an analytic function. In the same vain, it follows that ⇣(s) < ⇣(σ0) when Re s>σ0. o | | | | The ⇣-function may be extended to a meromorphic function in the whole complex plain with simple pole and residue 1 at s =1as the sole singularity. An argument using partial integration extends it to the right half plane σ>0.Belowweshallgivethis extension which also as a nice introduction to the commonly used technic of partial integration. The version we need is for Riemann-Stieltjes integrals, since step-functions are involved, and is more subtle than the usual calculus version. But first we introduce the Euler product: The Euler product As the name indicates, the Euler product was discovered by Euler. It connects the Riemann ⇣-function with the prime numbers, and is in that respect really the hug of the whole theory. One of Euler’s applications was to show there are infinitely many primes, which might be shooting sparrows with canons, but as we shall later on shall see, the idée has nice consequences. EulerProduct Proposition . (Euler product) One has for such s that Re s = σ>1 the following equality s s 1 ⇣(s)= n− = (1 p− )− . − n p X Y s 1 ks Proof: Summing the geometric series, one finds (1 p− )− = k 0 p− ,hence − ≥ P s 1 ks s (1 p− )− = p− = n− − p x p x k 0 p n p x Y Y X≥ | X) —4— Number theory — Intro MAT4250 — Høst 2014 where we in the second equality uses Mertens theorem saying that the product of finitely many absolutely convergent series equals their Cauchy-product. The Cauchy product is just the series whose terms are products of one term from each of the series involved. Clearly one has s s σ n− n− n− , − n p n p x n x X | X) X≥ σ and the right side of the inequality can be made as small as we please the series n n− being absolute convergent. o P Problem .. Use the Euler product to show there are infinitely many primes. Hint: 1 The harmonic series n n− diverges. X From the proof one obtainsP the following estimate that will be useful later on: Useful Lemma . 1 (1 p− ) < 1/ log x. − p x Y Proof: Putting s =1we arrive at 1 1 1 1 (1 p− )− = n− n− log x. − ≥ ≥ p x p n p x n x Y | X) X o The primes are of density zero As a teaser, we give the following result, certainly very weak compared to the PNT. Anyhow, it tells us that the density of the primes is zero; that is, the relative portion of primes less than x tends to zero when x grows. Proposition . lim ⇡(x)/x =0. x !1 Proof: Let q be a natural number to be chosen later. We divide [0,x] into intervals of length q;thatis,intervalsofshape[(a 1)q, aq] where the a’s are the natural numbers − with a [x/q].Ineachofthosesubintervalsthereareφ(q) integers relatively prime to q,henceatmostφ(q) primes not dividing q.Ifk is the number of different prime factors in q,thisgives ⇡(x) φ(q)x/q + k, and hence ⇡(x)/x φ(q)/q + k/x. —5— Number theory — Intro MAT4250 — Høst 2014 Now, if ✏>0 is given, let y be such that 1/ log y<✏,thenlettingq = p y p,wehave after lemma .

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    14 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us