NUMBER THEORY ARTHUR BARAGAR Ansrnacr. The following ares number of topics in mumber the- ‘oxy which might help with some Olympiad type problems. 1 have Included eeveral past Olympiad problens at appropiate places ‘This isnot te aay that the methods described bel are all you. ned, Quilt ota, core cleverness required Cowrents 1, The Euclidean Algorithm 2. Modular Arithmetic 2.1, Multiplicative Inverses Grovps 2.3, Fermat's Little Theorem, Bulet’s Theorem, and Gauss" Lemma 24, The Euler phi function 2.5. Polyzomiala in modular arithmetic |. The Gaussian integers 3 4. Quadratic integer rings 5, Independence over Q 6. Cyclotomic polynomials and DeMoivre's Theorem 7. Stereographie projection and Pythagorean triples 8. The prime number theorem 1. Tae EUCLIDEAN ALGORITHM ‘You are probably familiar with the division algorithm. This says that ‘given a and b, there exists aq (the quotient) and an r (the remainder) such that 0 0, d divides both a and b, TARA Fine 25,1908.2 ARTHUR BARAGAR and ife also divides both a and b, then e divides d. We can also define a follows: dm min(or +by > 0:2, €Z): Check to make sure that these define the same thing, In particular (and this iseften useful), there exist integers + and y euch that ged(o,8) = a2 + by. Given a ard b, we can solve for and y via the Bucidesn algorithm. “This is just «repeated use of the division algorithm, and is esist to dlemonstrate using a mimerical example. So let us find = and y given a= 23 and b= 15: 23 = 1-13410 13 = 11043 10 = 3-341 ‘We can turn this around 1 = 0-3-3 19-3-(13-10) 4-10-3.13, 4-(@3-13)-3-19 = 408-718. ‘Thus, = dand y= -7. “The Euelidean algorithm also plays a role in some very famous puz- zles. For example, explain how to measure exactly 4L of water, given 4 jug which holds exactly 3L and another which holds exactly #L (and of course, neither of which has any markings.) Do you see how the ‘Buclidean algorithm plays a role in the solution? ae ints . Exercise. ('59, #1) Prove that the fraction 73 ‘every natural number n. jrreduetble for 2, MopuLAR ARITHMETIC ‘You again are probably familiar with modular arithmetic, but let ws review some definitions, We say two numbers a and b are equivalent (Gnod m) ifm divides a —b. We sometimes write [a] for the class of ‘elements equivalent to @ modulo m, so [a] = {b: m divides (b—a)}NuM@er THEORY 3 ‘Then, for any integer a there exists an integer b such that [a] = [i] and 0 < hc m. Tis, the set {{0),..,[m— I]} is a complete set of representatives modalo m, We write Bm. = {(0, m= A}. ‘Wo often dcop the brackets and just wite Emi. = {0,...m— 1}. In modular arithmetic, adlition and multiplication are exactly 05 pected. That is, we define f+B = +4 (alle) = fab} (Check that those definitions are well defined. ‘That is, that it doesn't matter which representative we choose. We will no longer use the ‘square bracket notation, and assume that it'll be clear when we mean, fan integer and when we mean the class represented by that number. 2.1. Multiplicative Inverses. We will call a number b the inverse of ‘a modulo m if, ab=1 (mod m). Note that a is invertible if and only if ged(a,m) = 1. Suppose that d= ged(a,m). Then there exist = and y such that daartmy 0 we have az=d (mod m). In pacticuar, if d = 1, then ais invertible. Now, if is invertible then. there exists euch that ab=1 (mod m). ‘That is, m divides ab — 1. That is, there exsts y such that at raaee 1 = ob my. Hence, 1= ged(e,m)“ [ARTHUR BARAGAR 22. Groups. The notion of a group is something you'll learn about in the future, but there's really no reason to not introduce them here: Definition. A group G is a set of elements together with a relation (which well wile a8 multiplistion) euch that for any a, b, and ¢ in G, we have 1, ab G (closure) 2. (abje = albe) (associativity) 3. There exists an identity 1 such that for all a € G, al=la=a. 4. For every a there exists an clement a~* such that Aneta. ‘Ewe further have ab = ba, then we call G an Abelian or commutative poup. In the ebove definition, we represented the relation as a product. ‘We could also write a relation as an addition, but we usually only do 50 if the relation is commutative. If we use the notation +, then well ‘also call the icentity 0 instead of 1. ‘Note that 2/mZ, forms a group where the relation is addition, We write (Brn) = {a € B/mZ,: goa(m) = 1} ‘Note that (Z/m2)* is a group under multiplication. “More familar examples of groups are the integers under addition; the rationals not including zero under multiplication; the rationals under addition, etc. 'A related concept is that of a fed: Definition. A field F isa set of elements together with two relations ‘band « with the properties that F is a commutative group under + (0) is commutative group under multiplicton; and for all a, bane eink, o(b-+ 6) = ab+ ac ‘This last property is called the distributive law. ‘Familiar examples of fields are the rational numbers; the real num- ‘bers; and the complex numbers. ‘Note that {'p isa prime, then Z/pZ is a field. If m is not a prime, then Z/mZ.is not afield.‘NUMBER THEORY 5 23, Fermat's Little Theorem, Euler’s Theorem, and Gauss’ Lemma. ‘Theorem 2.1 (Fermat's Little Theorem). Suppose p is a prime. Then for all a not divisible by p, we have otal (mod »), Proof, Consider the set {1,.-.p~ 1} and the set {0,22 ...(P~ 1a} [Note that these two sets are equal, modulo p (why?). Thus, (1) 2%p— 1)! (aod 7). Now, just ‘eanod! the (p ~ 1)! (that i, multiply both sides by its inverse) a Definition. The Euler phi function is lon) = #{0-€ Z/rnZ: ged(o,m) = uler generalized Ferma result: ‘Thoorem 2.2 (Euler's Theorem). Suppose goi(a,m) = 1. Then aX") =1 (mod m). And Gauss generalized tht: ‘Theorem 2.3. Suppose C is a finite group with n elements, and a € G. Then aL This often usefl to know that if G is a fnite group with m elements, @€ G, and m is tho smalest positive integer such that 0 = 2, then ‘m divides n. (Prove this) 24. The Euler phi function. The Buler phi funetion has some in- teresting propertios: 1. IEpis a prime, then p(t) = 77"4(p— 1). 2. em and are relatively prime, then (ran) = olm)e(n) ‘One usually uses the Chinese remainder theorem to prove result ‘Theorem 2-4 (Chinese Remainder Theorem). Let m ond m be relax tively prime, and let @ and bbe ang integers. There exists a unique = such that 0 <2-( 32] ) 2s ot vise by 6 for any ntgern 2 0 6 Cycuorowte POLYNOMIALS AND DeMorvne's TusonEx ‘Theorem 6.1 (DeMoives Theorem). e” = cosd + ising. ‘Tis thearem canbe used to prove th angle sum formas: cox(a'+A) + tsn(a-+8) ood eine? (cosa + isina)(cos 6 + isin 8) (cosacos9 ~sinasind) + isinacosP + cosasin 0)0 ARTHUR BARAGAR Note that this proof also weed independence of 1 and i. ‘The same argument can be used to find double angle formulas, or triple angle forrmulas, et ‘The polynomial Pe) =2"— is called a cyclotomic polynomial, It’s roots are aon, ifn =pis primo, then P(a)=(e— Nett +. t22 4241), and does not factor further, Let's look more closely at this root el = cos(2n/p) + isin( n/p). Its inverse is w= con(2e/p) — tsi). “Tus, w= 2-4 2° is ral (in fact #* +2 is eal), ‘This mumberw always satisfies polynomial of degree For example let p= 8, nd w= 2-427, Then wea (teat 42te and hence wew-1 = ttedltetee? = 2% ‘Thus, we have just shown that shpat px ti)=0. 1+ v5 ~. Exercise. ('62, # 4) Solve the equation cod? e+ cas? 22 + 00s" 3 = 1. Exercise. ('63, # 5) Prove that .08(n/7) ~c0s(2n/7) + c0s(3n/7) = 1/2 2ece(2n/8) =NUMBER THEORY ” 17. STEREOGRAPHIC PROJBCTION AND PYTHAGOREAN TRIPLES ‘You are probably familiar with Pythagorean triples ~ integer triples (a,b,¢) such that +P ac, such os (9,46) and (6, 12,12). You may oven be familiar with 2 general formula for all these triples, and saw this formula derived via a num- ber of algebraic manipulations and repeated observations that certain numbers are squares. In the following, let us take a more geometzic approach, ‘Note that if (a,8,¢) i on the circle Pythagorean triple, then (a/c,8/¢) is a point Payal (4) few) () Let us consider the line through (0,1) and a point (f,0) on the 2- axis, This line intersects the circle at two points ~ at (0,1) and (sey) ‘at (ey). Comparing slopes, we get yor Heir » z=t1-1)- Since (2,) is a point on the circle, this gives ey-ey-1 w-Yw-D+ H+)2 ARTHUR BARAGAR, ‘The solution y = 1 gives thenorth pole, so we're interested in the other solutions, which are w@+y+t-#) = git #41 ‘Thus in terms of, the points on the circle are (#4, 234) . Noto that if tis rational, then so are z and y, and if r and y are rational, then 80 it t. Now, let us write t = p/q. Then 2pq a) n= (Ste Ree p> 920 arrelatively prime, positive, and of diferent parity, then (be) = par Pr +9). “his gives us all Pythagorean triples with even, Those with ood are sven by the pis (yg) bth ot a (60) = (9,7 —o)/2 + )/2). One can also get those with a odd by finding the solutions with Seven and svieing and ne advantage of this method in eat st works for all qudratic carves, For example, ty eon = 2 = con a + = 20 and on +P = 32. Tn genera i wil ot wok on bis, o Bight degre curves, (Tis tas fundamental conept ia algae german) AD exception isthe cure vedi2 Use stereographie projection on this curve using the point (0,0) and the line 2 = 1 Exercise, ('75, #5) Determine, with proof, whether or not one can find 1975 points on the cireunafarence of a circle with unit radius such that the distance between any two of them is a rational number. 8, Tie FRIME NUMBER THEOREM ‘Theorem 8.1 (The prime number theorem). Let n(2) be the number of primes less than x, That is, a(e)= Hp prime: p<2).Nuumen THEORY " ‘Then (2) grows asymptotically like me ‘That is, ig Oe eS aeec gear ag venan Saari Fa ae ee ‘Theorem 8.2. There exists an infinite number of primes. Peed eal ely eau stas o eale e P= {Ps Pa, Pa;—-sPa} = (2,3; Pah ee Napa) Note that NV and p; are relatively prime for all i, 50 p; does not divide rare sy peepee etre et ekesnyr pete RES mie ui auey Sov agieh nay Poor Se ne eave cee meas ce ore ta te Cee ee ee eee Te Wat itwcnie on Ganenwancm mie Taare acne eee ean Redan Pte Wesomsetaisnte Eetel on pores nmenan uals Bre et ec ntatcay coe are relatively prime. See elite dt eater anna pega tran ae mes