22. Let a and b be positive integers for which 2ab2−b3+1a2=k is a positive integer. Since k>0, it follows that 2ab2≥b3, so 2a≥b. If 2a>b, then from 2ab2−b3+1>0 we see that a2>b2(2a−b)+1>b2, i.e., a>b. Therefore, if a≤b, then a=b/2. We can rewrite the given equation as a quadratic equation in a, a2−2kb2a+k(b3−1)=0, which has two solutions, say a1 and a2, one of which is in N0. From a1+a2=2kb2 and a1a2=k(b3−1) it follows that the other solution is also in N0. Suppose w.l.o.g. that a1≥a2. Then a1≥kb2 and
0≤a2=a1k(b3−1)≤kb2k(b3−1)<b.
By the above considerations we have either a2=0 or a2=b/2. If a2=0, then b3−1=0 and hence a1=2k,b=1. If a2=b/2, then b=2t for some t, and k=b2/4,a1=b4/2−b/2. Therefore the only solutions are
(a,b)∈{(2t,1),(t,2t),(8t4−t,2t)∣t∈N}
It is easy to show that all of these pairs satisfy the given condition.