Solution:
It follows by the given condition that m(m+j−1)=j(j−1) and hence m2=(j−m)(j−1). If p is a prime divisor of j−m and j−1, then it divides m. Therefore p divides j and 1. This contradiction shows that j−m and j−1 are coprime. Then j−m=u2 and j−1=v2, where u and v are non-negative integers. It follows that uv=m and u2+uv=j=v2+1, and the condition 1≤j<2004 gives 0≤v≤44. So, we have to solve in non-negative integers the equation
u2+uv=v2+1
If v=0, then u=1. Assume that the pair (u0;v0) is a solution of (∗) and v0≥1. Then u0≥1. Since u0v0≥1, it follows that u0≤v0. Moreover, it is clear that v0<2u0. Set v1=v0−u0, 0≤v1<v0. We have that u02=v0(v0−u0)+1=(u0+v1)v1+1=u0v1+v12+1. Set u1=u0−v1=2u0−v0>0. Then
(u1+v1)2=(u1+v1)v1+v12+1⟺u12+u1v1=v12+1
i.e., we obtain a new solution of (∗). If v1=0, then u1=1. If v1≥1, then u1≥1. Setting v2=v1−u1<v1 and u2=u1−v2, we get in the same way a new solution. So, we get a sequence of non-negative integers v0>v1>⋯. If vk=0 for some k, then uk=1 and writing uk−1,vk−1,…,u0,v0, we get the Fibonacci sequence.
Thus solutions of (∗) are (u;v)=(1;0),(1;1),(2;3),(5;8),(13;21), and for the other solutions we have that v>44. The first solution of (∗) gives j=v2+1=02+1=1 and then m=uv=0, which does not satisfy the given condition. The other solutions give the following solutions of the problem: j=12+1=2, j=32+1=10, j=82+1=65 and j=212+1=442.