Você está na página 1de 8

Cryptarithms By a cryptarithm we mean an arithmetic sum in which some or all of the digits have been substituted by letters, like

in a code, with the agreement that different letters represent different digits. The task is to decode the code and figure out which letter represents which digit. For example, from the multiplication 12 2 = 24, we can create the cryptarithm AB 2 = BC, with the solution A = 1,B = 2,C = 4. But this cryptarithm has other solutions, e.g., A = 2,B = 4,C = 8. Some cryptarithms can be very challenging indeed. One of the most famous such problems is: SEND + MORE = MONEY, or, written as an ordinary addition:

Try solving this on your own! 1. Notation When we write a number using letters, as in SEND, we will always use an overline. This is to avoid the possibility of an ambiguous meaning: if we write ab it means a times b (that is, a b), whereas when we write ab, it means the two digit number with tens digit a and units digit b. 2. A solved example Lets start with a much simpler cryptarithm and solve it completely: Problem 1: ON 4 = GO. There are three digits to be found here: O,N,G.

Since ON 4 is a two digit number, we must have ON 24. This means that O is either 1 or 2. Since ON 4 is an even number, and its units digitis O, this digit must be even. Hence, O = 2. Since N 4 has units digit 2, N must be either 3 or 8. If N = 8 then we would get ON = 28, which exceeds 24 (and in any case does not fit). So the only possibility is N = 3. And this fits, since 23 4 = 92. So the code is: O = 2, N = 3, G = 9.

That was simple, wasnt it? But cryptarithms arent always so simple! We will see more challenging problems shortly. Problem 2: ABCD 4 = DCBA. Here we are looking for a four digit number whose digits get reversed on multiplication by 4. There are four digits to be found: A,B,C,D. Moreover, these represent different digits.

Since ABCD 4 is a four digit number, ABCD < 2500, hence A 2. Since DCBA is a multiple of 4, its units digit is even; hence, A = 2. Since the units digit of DCBA is 2, the units digit of D 4 is 2, hence D = 3 or D = 8. The first possibility does not fit since DCBA > 20004 = 8000. Hence, D = 8. Since ABCD 4 is at most 8999, the digit B is at most 2. Since 8CB2 is a multiple of 4, the number formed by its last two digits, namely B2, is a multiple of 4. This implies that B cannot be 0 or 2; hence, B = 1. So ABCD = 21C8, and the equation reads: 21C8 4 = 8C12. Since 4 8 = 32, there is a carry of 3 into the tens column, so 4C + 3 has units digit 1. This means that the units digit of 4C is 8, so C is either 2 or 7.

Since 21C8 4 > 8400, we must have 8C12 > 8400, and so C 4. Thus, the possibility C = 2 does not fit. Hence, C = 7, and ABCD = 2178. We have now uncovered all the unknowns: A = 2, B = 1, C = 7, D = 8.

But we must check whether what we have got fits; it does: (1) That too was fairly easy. Writing the answer the way we have done is quite tedious. Sometimes the reasons are easier to explain orally to someone sitting with you, than to write them out. 3. Problems Here are some more cryptarithms you could tackle. 1. 2. 3. 4. 5. ABCD 9 = DCBA ABCDE 4 = EDCBA ABCDEF 5 = FABCDE TWO TWO = THREE Try to create a few cryptarithms by yourself. You will find that it is not quite as easy as it may have first seemed! And it is particularly challenging if we want the cryptarithm to have precisely one solution. (You would have found that the above four cryptarithms have just one solution each.)

4. Reversible Numbers The fact that 2178 4 = 8712 prompts us to ask the following question. For any positive integer n, let r(n) denote the reversal of that integer. (It is taken for granted that we are working in base ten.) Example: r(231) = 132, r(1947) = 7491, r(230) = 32. Now we ask:

Find the set S of positive integers n such that r(n) > n, and r(n) is a multiple of n. The phrase r(n) > n has been inserted to keep out palindromic integers like 11, 121, 131, etc, for which we have r(n) = n. There are obviously infinitely many of these, and they do not add any value or interest to the question. Because of this restriction, there are no single digit numbers in S. 4.1. Two digit numbers. What about two digit numbers? Suppose that n = ab is such a number, where a and b are digits; then b > a, and 10b + a is a multiple of 10a + b, say 10b + a = k(10a + b) where k > 1 is a positive integer. Transposing terms we get (2) and so (3) If k takes the values 2, 3, 4, 5, 6, 7, 8, 9, the fraction (10k - 1)(10 -k) takes the following values: (4) In every one of these cases, the numerator of the reduced fraction exceeds 10, and this is impossible if a and b are digits. Hence, there are no two digit integers with the reversibility property. 4.2. Three digit numbers. How about three digit numbers? Suppose

that n = abc is such a number; here a,b,c are digits, and c > a. Then we have cba = k abc for some integer k > 1. We consider the different possibilities.

k = 2: In this case, a is even, and a < 5, hence a = 2 or a = 4. If a = 2 then (a,c) {(2, 1), (2, 6)}, and if a = 4 then (a,c) {(4, 2), (4, 7)}. It is easy to check that these do not fit the problem.

k = 3: We have: a {1, 2, 3}, and (a,c) {(1, 7), (2, 4), (3, 1)}. But none of these fits. k = 4: Here a = 2, and (a,c) = (2, 3), which does not fit. k = 5: No possible value for a. k = 6: No possible value for a. k = 7: Here a = 1, and (a,c) = (1, 3), which does not fit. k = 8: No possible value for a. k = 9: Here a = 1, and (a,c) = (1, 9), hence b = 109 or n = 119; but neither of these fits. {0, 1}, and so n

Hence there is no three digit number with the given property. Here are the Mathematica command by which we can verify the above answer: ClearAll[r, f]; r[n_] := FromDigits[Reverse[IntegerDigits[n]]]; f[n_] := r[n] > n && IntegerQ[r[n]/n]; SetAttributes[f, Listable]; s = 3; Select[Range[10^(s - 1), 10^s - 1], f];} The answer computed is {}; i.e., the empty set. Note that this is only a computer assisted verification. It can be regarded as a proof to the extent that the software one uses is correct. (Not that anyone doubts the correctness of this software! we are only being politically correct.)

This particular instance was easy, so our comment may not carry any punch. But there are theorems for which the only available proof involves heavy usage of computer software, and in such cases this comment assumes significance. 4.3. Four digit numbers. We move on to four digit numbers. From our earlier analysis we know that there do exist four digit reversible numbers, for example, 2178. Suppose that n = abcd is such a number; here a,b,c,d are digits, and d > a. Then we have dcba = k abcd for some integer k > 1. If we consider the different possibilities for k (namely, k {2, 3, 4, 5, 6, 7, 8, 9}, we find the analysis going along practically the same lines as for the three digit case. We find that there are no solutions if k = 2, 3, 5, 6, 7, 8; but if k = 4 or k = 9, we do get one solution in each case: if k = 4 we get n = 2178, and if k = 9, we get n = 1089. (Curiously, 2178 = 2 1089.) This means that the two smallest members of the set S are 1089 and 2178. If we run the above Mathematica commands using s = 4, we get this answer: (5) This is what we had anticipated, but is always nice to get a confirmation that ones answer is correct! 4.4. Numbers with five or more digits. If we run the above program using s = 5, here is what we get: (6)

Now, that is curious, isnt it? We get numbers that look almost the same as the ones we got using s = 4, but with a 9 inserted in the middle. If we put s = 6, the answer is just as curious: (7) This time we get numbers that look the same as the ones we got using s = 4, but with a 99 inserted in the middle! And if we put s = 7, here is what Mathematica gives us: (8) Well, well, well. Will this trend continue for higher values of s? We seem to have hit upon an interesting result, havent we? Conjecture. The only integer values of r(n)n are 1, 4 and 9. Can we make this into a theorem? Note that we have included the value 1 because r(n) = 1 for every palindromic integer n. We shall leave you the task of exploring this conjecture. 5. Extensions We can broaden this investigation by asking the following question: For positive integers n, let f(n) = r(n)n where r(n) denotes the integer whose digits are in the revese order as compared with n. We conjectured above that the only integral values taken by f(n) are 1, 4, 9. What are nice rational values taken by f(n)? By nice we mean that both the numerator and denominator of f(n) are small integers; for

example, fractions like 12, 13, 23, and so on.

Você também pode gostar