C Roettger Spring 2013 Math 350 – Practice Exam 1 – Solutions Problem 1 Convert 2013 to binary and 1010001012 to decimal notation. For 2013, I use the top-down method. The nonzero multiples are always 1, no need to write these. The exponents indicate where the digits go. 2013 989 477 221 93 29 13 5 1 = = = = = = = = = 210 + 989 29 + 477 28 + 221 27 + 93 26 + 29 24 + 13 23 + 5 22 + 1 20 + 0 So 201310 = 11111011101. Easy check: 2013 + 35 = 2048 = 211 (try adding 35 = 1000112 to the answer). Now binary to decimal: 28 26 22 + 20 = 256 = 64 = 4 = 1 325 Problem 2 a) Find integers s, t such that 120s + 63t = gcd(120, 63) b) Find the inverse of 32 modulo 43. a) The extended Euclidean algorithm gives s t 120s + 63t 1 0 120 0 1 63 1 −1 57 −1 2 6 10 −19 3 (the next remainder is zero, so the gcd(120, 63) = 3 and 10 · 120 − 19 · 63 = 3. b) The very same Euclidean algorithm gives s t 43s + 32t 1 0 43 0 1 32 1 −1 11 −2 3 10 3 −4 1 so the inverse of 32 modulo 43 is 43 − 4 = 39 (we could also use −4 or any other integer congruent to −4 modulo 43, but the standardized system of representatives for congruence classes modulo 43 is of course 0, . . . , 42). Problem 3 Find an integer x < 120 such that x≡4 (mod 8) x≡2 (mod 3) x≡1 (mod 5) The Chinese Remainder Theorem (proof of it, rather) says compute M1 = 3·5 = 15, M2 = 8 · 5 = 40, and M3 = 8 · 3 = 24. Then 15 · 7 ∼ = 1 (mod 8), 40 · 1 ∼ =1 ∼ (mod 3) and 24 · 4 = 1pmod5. We put x = 4 · 15 · 7 + 2 · 40 + 1 · 24 · 4 = 420 + 80 + 96 = 596. We can change this modulo 120 and still keep all the congruences above valid. So the answer is x = 596 mod 120 = 116. Problem 4 How many solutions does the congruence 12x ≡ 20 have modulo 40? Either you remember that we had a theorem giving the number of solutions as gcd(12, 40) = 4. Or you play it safe, reproduce a bit of the proof of that theorem: First, divide by the gcd(12, 20) = 4 to get the equivalent congruence 3x ≡ 5 (mod 10). This has exactly one solution modulo 10 because gcd(3, 10) = 1 (it is x ≡ 5 (mod 10) but we don’t need that). There are 4 values of x modulo 40 that match this condition. Problem 5 Why is 21000 − 1 not a prime? Find a short proof (NOT trial division). Use the factorization a2 − 1 = (a − 1)(a + 1) to get 21000 − 1 = (2500 − 1)(2500 + 1). Similarly, you can show that 2n − 1 is not a prime unless n is a prime itself. Problem 6 a) Explain why any number with base 16-representation n = (a1 a2 . . . ak )16 satisfies n ≡ a1 + a2 + · · · + ak 2 (mod 15) b) I wrote a number in base 16, n = A 46201B16 and unfortunately smudged the second digit. But I know that n is a multiple of 15. What was the missing digit? a) For any power of 16, 16k ≡ 1k = 1 (mod 15) and therefore n= s X k=0 ak 16k ≡ s X ak 1k = k=0 s X ak (mod 15) k=0 b) Using part a), n ≡ A + a6 + 4 + 6 + 2 + 1 + B = 34 + a6 (mod 15) so n ≡ 0 (mod 15) implies a6 = 11 (written B in base 16). Problem 7 Let fn be the n-th Fibonacci number (so f0 = f1 = 1, fn+2 = fn+1 + fn ). Prove by induction that gcd(fn+1 , fn ) = 1 for all n ≥ 1, and that the Euclidean Algorithm finds this gcd in exactly n − 1 steps. Base step – n = 1 is obviously true. Induction step – assume the claim is true for one value of n. We substitute n + 1 for n, so the claim is about gcd(fn+2 , fn+1 ). The Euclidean Algorithm says fn+2 = qfn+1 + r and because of fn+1 < fn+2 < 2fn+1 (this is quite obvious, but would have to be shown by another little proof by induction, if you insist), we get q = 1 and therefore r = fn . Then the Euclidean algorithm will continue and compute in n − 1 steps gcd(fn+1 , fn ) = 1 This is part of the computation of gcd(fn+2 , fn+1 ), so it proves that the new gcd is indeed 1. The total number of steps turns out to be n, as claimed. 3