It is straightforward to show that if a1=1, then an=n for all n. Since an+1 is an increasing function in an, it follows that the set of possible a1 is of the form {1,2,…,m} for some m, which will be the answer to the problem. Consider the sequence bn=an+1−1, which has the recurrence bn+1=n⌊nbn+1⌋ It has the property that bn is divisible by n. Rearranging the recurrence, we see that n+1bn+1≤n+1bn+1<n+1bn+1+1 and as the bi are integers, we get bn+1−1≤bn<bn+1+n. For n≥2, this means that the largest possible value of bn (call this bn∗ ) is the smallest multiple of n which is at least bn+1. Also, since b1=b0+1, we find b0∗=b1∗−1, meaning that the largest value for a1 is b1∗, and thus the answer is b1∗. We have now derived a procedure for deriving b1∗ from b29∗=29. To speed up the computation, let cn=bn∗/n. Then, since bn∗=n⌈nbn+1∗⌉ we find cn=⌈nn+1cn+1⌉=cn+1+⌈ncn+1⌉ We now start from c29=1 and wish to find c1. Applying the recurrence, we find c28=2,c27=3, and so on until we reach c15=15. Then, ⌈cn+1/n⌉ becomes greater than 1 and we find c14=17,c13=19, and so on until c11=23. The rest can be done manually, with c10=26,c9=29,c8=33,c7=38,c6=45,c5=54,c4=68,c3=91,c2=137, and c1=274. The last few steps may be easier to perform by converting back into the bn∗.