Let m be a positive integer of 1000 digits, with the property that all of its digits are non-zero. For a positive integer n, consider ⌊nm⌋, where we define for any real number r, ⌊r⌋ to be the largest integer less than or equal to r. Determine the largest possible number of digits which are 0 for ⌊nm⌋, where 1≤n≤m.
Solution
Let M be the maximum number we seek. First we show that M≤939 must hold. Let m,n be positive integers satisfying the conditions of the problem, and let k be the number of digits of n (so, 1≤k≤1000). Let us represent nm as nm=a110b1+a210b2+⋯+aN10bN+ϵ, where 1≤ai≤9, for i=1,2,…,N, 0≤bN<bN−1<⋯<b1, and 0≤ϵ<1. Then from 10999≤m<101000 and 10k−1≤n<10k it follows that we have b1≤1000−k. We can also show that the following lemma holds:
Lemma:bi−bi+1≤k+1 holds for every i,1≤i≤N−1, and bN≤k holds also.
Proof of Lemma: Suppose for some i,1≤i≤N−1, bi−bi+1≥k+2 is satisfied. Then, if we let A=a110b1−bi+1−k−2+a210b2−bi+1−k−2+⋯+ai10bi−bi+1−k−2, B=ai+110bi+1+ai+210bi+2+⋯+aN10bN+ϵ, we see that A is a positive integer and B satisfies 0<B<10bi+1+k. We see that with A and B as above we can write nm=10bi+1+k+2A+B. But then, since we have 10bi+1+k+2An<m=10bi+1+k+2An+Bn<10bi+1+k+2An+10bi+1+k+1, we have to conclude that the 10bi+1+k+1-th digit of m is 0. But this contradicts the assumption that all the digits of m are not zero, and therefore, bi−bi+1≤k+1 must be satisfied for every i,1≤i≤N−1.
Next assume that bN≥k+1 is satisfied. If we let C=a110b1−k−1+a210b2−k−1+⋯+aN10bN−k−1, then we see that C is a positive integer and we can write nm=10k+1C+ϵ. But then we have 10k+1Cn≤m=10k+1Cn+ϵn<10k+1Cn+10k, which implies that the 10k-th digit of m is 0, contradicting the assumption of the problem. Therefore, we must have bN≤k, and this completes the proof of Lemma.
Going back to the proof of the claim that M≤939, we note that the number of 0 digits in ⌊nm⌋ equals b1+1−N. By summing the corresponding sides of the following N inequalities shown in Lemma above b1−b2≤k+1,b2−b3≤k+1,…,bN−1−bN≤k+1,bN≤k, we obtain b1≤(k+1)N−1, which yields k+1b1+1≤N. Consequently, we get b1+1−N≤b1+1−k+1b1+1=k+1kb1−1+1≤k+1k(1000−k)−1+1=1003−((k+1)+k+11002)≤1003−21002<940, where we used the fact (k+1)+k+11002≥2(k+1)⋅k+11002=21002. This shows our claim that M≤939.
Finally, we show that M≥939 by exhibiting a particular choice of m and n, satisfying the conditions of the problem, for which M=939. For this purpose, consider m=211⋯131 digits1266⋯632 digits1266⋯632 digits1266⋯632 digits1266⋯69 digits,n=211⋯131 digits Then, we see that ⌊nm⌋=100⋯0632 digits00⋯0632 digits00⋯0632 digits00⋯0632 digits00⋯09 digits and, therefore the number of its 0-digits is 939. Thus, we conclude that 939 is the desired answer to the problem.
Want a route through all this instead of an archive? The track
puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.
Source: MathNet,
licensed CC-BY-4.0.
Statement and solution reproduced as published; topic and difficulty added by this site.