The sequence a0,a1,… is defined by the initial conditions a0=1, a1=6 and the recursion an+1=4an−an−1+2 for n>1. Prove that a2k−1 has at least three prime factors for every positive integer k>3.
Solution
Consider the sequence b0,b1,…, defined by the initial conditions b0=1, b1=2, and the recursion bn+1=4bn−bn−1 for n≥1. We have an=bn+1−1, for all n≥0. In particular, a2k−1=b2k−1, k≥0. It is not hard to see that the general formula for bn is bn=2(2+3)n+(2−3)n,n≥2.(1) Let k≥3 be fixed. It follows from (1) that b2k=∑j=02k−1(2j2k)22k−2j3j and thus b2k≡1(mod3) and b2k≡1(mod4).(2) It is easy to see that the terms of the sequence, defined by cn=23(2+3)n−(2−3)n, are positive integers, which satisfy bn2−3cn2=1 for all n∈N. In particular, we have b2k2−1=3c2k2.(3) Applying the identity (x2s−y2s)(x2s+y2s)=(x2s+1−y2s+1) with x=2+3 and y=2−3 for s=0,1,…,k−1, we get 23j=0∏k−1(2b2j)=(2+3)2k−(2−3)2k. Therefore, c2k=2k+1b2b22…b2k−1.(4) It follows from (2) and (3) that 2∣b2k+1; and gcd(b2j,6)=1 for every j=1,2,…,k−1.(5) Since b2k+1≡2(mod3), by (3), (4) and (5) we have 3∣b2k−1 and 22k+1∣b2k−1.(6) Now, suppose that there exists an m≥3 such that b2m−1 has at most two prime factors. Then it follows from the relations in (2) that these prime factors must be 2 and (or) 3. Furthermore, (6) implies that we must have b2m=22m+1⋅3+1 (7). Therefore, by (3) we get c2m2=3b2m2−1=4m+1(3.4m+1).(8) On the other hand, by (4) we have c2m2=4m+1∏j=1m−1b2j2>4m+1(3c2m−1+1), and thus c2m2>4m+1(3.4m+1) as c2m−1≥2m by (4). This is a contradiction to (8) and therefore a2k−1 has at least three prime factors for every positive integer k≥3.
Alternative solution:
This approach is based on the fact that all linear sequences are periodic modulo arbitrary positive integer. We show that a2k−1 is divisible by 2, 3, 7. Observe that if an is even for some n, then an+2=4an+1−an+2 is also even. Since a1=6 is even, we conclude that a2t−1 is even for all positive integers t. Therefore, a2k−1 is even. Let bn=an(mod3). Since b0=1 and b1=0, it follows by induction that b2t=1 and b2t+1=0. Hence, b2k−1=0, i.e. a2k−1 is divisible by 3. Let cn=an(mod7). We have c0=1,c1=−1,c2=−3,c3=−2,c4=−3,c5=−1,c6=1,c7=0,c8=1,c9=−1. Since c0=1, c1=−1 and c8=1, c9=−1, we conclude that the sequence c1,c2,… is periodic with period 8. It follows from c7=0 that c7+8t=0 for all positive integers t. It remains to notice that 2k−1=7+8(2k−3−1), implying that c2k−1=0. Therefore, a2k−1 is divisible by 7.
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.