You are on page 1of 9

Problem sets on decimal expansions of fractions from a group theory perspective

Sybilla Beckmann February 7, 2012


The four problem sets below are explorations on decimal expansions of fractions in light of group theory. They provide students who are studying group theory with opportunities to make conjectures in familiar settings, to work with concrete examples in order to develop intuition, and ultimately to work out general proofs using newly learned mathematics.

Overview of the problem sets

The rst, second and third problem sets focus mainly on the relationship between the period of the decimal representation of a fraction and its denominator. There are some fun applications of this, such as nding the period and full repetend of the decimal representation of a fraction with a long period. (I will use the terminology that in a repeating decimal, the block of digits that repeats is called the repetend, and the number of digits in the repetend is called the period.) Naturally occurring examples of cosets are given. These are related to the fact that fractions with the same denominator can be divided into subsets according to their decimal representations: some have the same decimal representation except shifted over. The fourth problem set concerns the sum of the digits in the repetend of the decimal representation of 1 for prime numbers p. For example, p 1 = .142857142857... 7 6 9. 2 In fact, the rst and fourth digits, the second and fth digits and the third and sixth digits in the repetend 142857 each add up to nine. Is this a coincidence? Wonderfully enough, these types of problems provide more than a natural, motivating link between school arithmetic and abstract algebra; they go back to the very roots of modern algebra and number theory. According to Klein [3], reprinted also in [2], in the year 1795 Gauss computed the full repetends of 1+4+2+8+5+7 = 1 has period 6 and

1 the decimal representations of p for all prime numbers p up to 1000 in an effort to determine the dependence of the period on p. In 1801, Gauss published Disquisitiones Arithmeticae [1], introducing congruent numbers and modular arithmetic. Articles 312 through 318 of Disquisitiones Arithmeticae give applications of modular arithmetic to decimal expansions of fractions. Except for the exposition, some of the applications given here are essentially the same as some of Gausss.

Notation and terminology: If n 2 is a natural number, Zn denotes the ring of congruence classes modulo n, namely {0, 1, . . . , n 1}, under addition and multiplication modulo n. The units of Zn (the elements that have a multiplicative inverse) is denoted by Z . It is a group under multiplication modulo n n and consists of the elements of Zn represented by integers that are relatively prime to n.

First problem set: the period and the denominator

In this problem set, students rst discover in examples that for a prime number p 1 (other than 2 or 5), the period of the decimal representation of p divides p 1. They then explore this relationship and ultimately explain it (either directly or with Lagranges theorem). Along the way are hands on computations with subgroups and cosets. If these problems are assigned as homework, it might make sense for the students to work through all the problems but to hand in only a subset, such as problems 5 through 7. Perhaps a student who is unable to give the general proof in problem 5 could hand in problems 1 through 4 for partial credit. 1. Find the decimal representations of 1 1 1 1 1 , , , , , 7 11 13 37 101 1 1 1 , , , 41 271 239 1 1 1 , , . 4649 73 137 Make sure that you nd enough decimal places so that you see the full repeating pattern. In each case, compare the denominator minus 1 and the period of the decimal. What relationship do you notice? The goal of the next several problems is to help you nd an explanation for why this relationship holds. Notice that in each example above, the denominator is a prime number.

1 1 2. Find the decimal representations of 1 , 11 , and 37 by longhand division. 7 In each case, you get a sequence of remainders. For example, the rst 1 remainder you get in nding the decimal representation of 7 is 3 because 10 7 = 3. The next one is 2 because 30 28 = 2. How can you determine 1 the period of 7 from the remainders? Explain why.

3. Find the powers of 10 (namely 100 , 101 , 102 , . . .) in Z , in Z , and in 7 11 Z . Compare these to the remainders you found in problem 2. Can you 37 explain why this occurs? If you can, skip to problem 5, if not, continue on to problem 4.
1 1 4. For each of 1 , 11 , and 37 , write the sequence of equations that correspond 7 to the steps of the longhand division process. For example, in dividing 1 by 7 longhand, the equations corresponding to the rst three steps are

10 = 1 7 + 3, 3 10 = 4 7 + 2, 2 10 = 2 7 + 6. Now think about problem 3 again. Use these ideas for problem 5. 5. Prove that if n is a natural number greater than 1 that is divisible neither by 2 nor by 5, then the remainders that appear when one nds the decimal 1 representation of n by longhand division is the subgroup of Z generated n 1 by 10. Prove that therefore, the period of n is the order of 10 in Z . n You could begin by letting a1 , a2 , a3 , . . . be the digits in the decimal 1 representation of n , and letting r1 , r2 , r3 , . . . be the remainders when one 1 nds the decimal representation of n by longhand division. Then, as in problem 4, nd equations that correspond to the steps of the longhand division process. 6. More generally, let n be a natural number greater than 1 and let a be a natural number that is less than n and relatively prime to n. How is the set of remainders that appears when one nds the decimal representation a of n by longhand division related to Z ? Answer this after you work out n the following examples.
2 3 Find the set of remainders for 11 and for 11 . How are these related to the 2 subgroup of Z11 generated by 10? Find the set of remainders for 13 , for 3 5 13 , and for 13 . How are these related to the subgroup of Z13 generated by 10?

7. Use problem 5 to prove that if p is a prime greater than 5, then the relationship between p 1 and the period of the decimal representation of 1 p that you noticed in problem 1 above always holds. One can also prove problem 7 by bypassing problem 5 (as well as 2, 3, and 4) and working directly with the order of 10 in Z rather than the subgroup genern ated by 10. But this provides less of an opportunity for hands-on computations in Z . n 3

Second problem set: applications of the rst problem set

In this problem set, students rst use the ideas of the previous problem set to nd the period of a repeating decimal without actually doing the longhand division. Then students discover that for fractions with a xed denominator, many have the same repetend except shifted over. These correspond to the cosets seen in problem 6 of the rst problem set. Finally, students see how to use the ideas developed in this problem set to nd the full repetends of the fractions with long periods in the rst problem. In the rst problem, both 1/23 and 1/29 have the largest possible period, namely 22 and 28, respectively. A curious student might wonder how often this happens. A still unproven conjecture due to Gauss (which was generalized by E. Artin, see for example [2], page 127), is that there are innitely many primes 1 p such that the decimal representation of p has period p 1.
1 1 1. Use a calculator to nd the decimal representations of 23 and 29 . These are necessarily repeating decimals, but can you tell from the calculator what their periods are? Use problems 5 and 7 of problem set 1 to help 1 1 you nd the periods of 23 and 29 in an ecient way.

2. Compare the repetends of decimal expansions of the following pairs of fractions. What do you notice? 1 10 , 11 11 9 2 , 11 11 3 8 , 11 11 4 7 , 11 11 6 5 , 11 11 Now look at all the decimal expansions of fractions with denominator 13 and numerator a natural number less than 13. Compare the repetends of these numbers. What relationships do you notice? You should see two dierent types of repetends in this case. Compare what you saw in these examples to what you discovered in problem 6 of the rst problem set. 3. Continuing with the ideas of problem 2, consider all fractions of the forms a 37 , where a is a natural number less than 37. Without actually working out the decimal expansions of these numbers, use what you discovered in the previous problem and problem 6 of the rst problem set to predict 4

which of these fractions will have the same repetend, except shifted over. Check your answers by nding the decimal expansions.
a Now do the same with all fractions of the form 31 , where a is a natural number less than 31. Notice that in this case, most calculators wont show all the digits in the repetends.

4. Returning back to the fractions of problem 1, use the ideas of the last two 1 1 problems to nd the full repetends of 23 and 29 in an ecient way with an ordinary calculator. Give a complete and thorough explanation for why your answers must be correct. Hints: Remember that when you multiply a number by a power of ten it shifts the decimal point over. Also, think about how to write power of ten 23 as a mixed fraction (with a whole number part and a fractional part). How is the fractional part related to the remainder of the power of ten when it is divided by 23? Finally, consider using repeated squaring in Z in order to nd the remainder of a large power of 10 when it is divided 23 by 23.

Third problem set: the period and composite denominators

This problem set builds on the work of the rst problem set and considers frac1 tions of the form pq where p and q are distinct primes which are both greater than 5. As in the rst problem set, students start by working out examples and formulating a conjecture. Before proving their conjecture, students work with group theoretical examples that help point the way to a general proof. It would be worthwhile to point out in class that in general, the period of 1 divides (n), the number of natural numbers less than n that are relatively n prime to n (where is the Euler phi function). 1. Find the decimal representations of 1 1 1 1 , , , . 11 37 11 101 11 41 37 101 Based on these examples, and your work in problem 1 of the rst problem set, make a conjecture relating the period of 1 , pq and the periods of 1 1 , , p q 5

when p and q are distinct prime numbers, both of which are greater than 5. Now look back at problem 5 of the rst problem set and start to think about how to prove the conjecture you just made. 2. Find the subgroup generated by 10 in each of the following groups: Z , Z , Z Z , Z . 11 37 11 37 407 (Notice that 407 = 11 37.) How are the orders of these subgroups (generated by 10) related to each other? Similarly, nd the subgroup generated by 10 in each of the following groups: Z , Z , Z Z , Z . 1111 101 11 101 11 Again, how are the orders of these subgroups (generated by 10) related to each other? Do the same with the following: Z , Z , Z Z , Z . 707 101 7 101 7 3. Assume that p and q are distinct primes, both bigger than 5. Prove that the order of 10 in Z Z is the same as the order of 10 in Z . p q pq You might use the following to do this. Show that the map F : Z Zp Zq dened by F (a) = (a mod p, a mod q) is a ring homomorphism. Show that F induces a ring isomorphism f : Zpq Zp Zq . Then show that f induces a group isomorphism : Z Z Z . pq p q 4. Use the previous problem to explain how the order of 10 in Z is related pq to the order of 10 in Z and in Z , and to explain why this is so. Then q p prove your conjecture in problem 1.

Fourth problem set: the sum of the digits in the repetend

In this problem set students rst discover that if n is a natural number greater than 1 that is not divisible by 2, 3, or 5, then the sum of the digits in the 1 repetend of n is always a multiple of 9 (this is actually true as long as n is not 6

divisible by 3). They go on to see that for a prime number p greater than 5, if 1 N is the period of p , then if N is even, the sum of the digits in the repetend of 1 p is N 9 . 2 In order to prove this, students will need to know that Z contains exactly one p element of order two. This follows from the fact that p is prime (either by direct proof or by appealing to the fact that Z is cyclic). p The rst problem could even be preceded by a problem asking students to 1 nd all fractions n of period six or less, where n is not divisible by 2, 3, or 5. Some students might like to follow up on this problem set by investigating what happens when the period N is odd. 1. Find the decimal representations of 1 1 1 1 , , , , 11 37 101 1111 1 1 1 , , , 41 271 11111 1 1 1 1 1 , , , , , 7 13 91 259 77 1 1 1 , , . 143 481 407 To enough decimal places so that you see the full repetend in each case. In each example above, add all the digits of the repetend. What do you notice? (The fractions in this problem are listed in increasing order of period and 1 includes all fractions n of period six or less, where n is a natural number greater than 1 that is not divisible by 2, 3, or 5.) 2. Prove that if n is a natural number greater than 1 that is not divisible by 2, 3, or 5, then the sum of the digits in the repetend of the decimal 1 expansion of n is a multiple of 9. (The result is still true even when 2 or 5 divide n, but the proof is simpler this way.) One approach to proving this is to go back to the equations suggested in problem 5 of the rst problem set: 10 = a1 n + r1 10r1 = a2 n + r2 . . . 10rN 1 = aN n + rN ,

1 where N is the period of the decimal representation of n , where a1 , . . . , aN are the digits of its repetend, and where r1 , r2 , . . . , rN are the remainders from the longhand division process. Notice that this uses the fact that the repetend begins right after the decimal point (how? And what hypothesis guarantees this?). Then add the equations above and show that

(a1 + . . . + aN )n = 9(r1 + . . . + rN ).
1 3. Now consider the sums of the digits of the repetends of fractions p , where p is a prime number. For each of the following fractions, divide the sum 1 of the digits of the repetend of the decimal expansion of p by 9. How does 1 that quotient relate to the period of p ? Formulate a conjecture.

1 1 1 , , , 11 37 101 1 , 41 1 , 7 1 , 239 1 , 73 1 , 271 1 , 13 1 , 4649 1 . 137


1 p

(The above are all fractions of the form is a prime that is greater than 5.)

of period eight or less where p

Start to think about how you could prove your conjecture. Could the suggested setup of problem 2 be helpful? 4. For the three cases, p = 7, p = 73, p = 101, nd the subgroup generated by 10 in Z . (Remember what you learned in problem 5 of the rst p problem set: that such a subgroup generated by 10 is the set of remainders r1 , . . . , rN , notation as in problem 2 above, and that the order N is the 1 period of p .) Observe that in each case, the elements of the subgroup come in pairs that add up to p. Also observe that in each case, 10N/2 = p 1 = 1 in Z . What is the relationship between these two facts? p 5. Prove your conjecture in problem 3 when the period of use the idea of problem 4 to prove that ai + aN/2+i = 9,
1 p

is even. In fact,

(notation as in problem 2). In other words, prove that the ith digit and 1 the (N/2 + i)th digit in the repetend of p add up to 9, for i between 1 and N/2 1. (Check that this is true in your examples with even periods!) You will need to use the hypothesis that p is prime in your proof, since the result is not true in general. For example, check out p = 1111 = 11 101.

References
[1] C.F. Gauss, Disquisitiones Arithmeticae, English Edition, Springer-Verlag, New York, 1986 [2] J.R. Goldman, The Queen of Mathmatics, A K Peters, Wellesley, Massachusetts, 1998 [3] F. Klein, The Development of Mathematics in the 19th Century, Math Sci Press, Brookline, Massachusetts, 1979

You might also like