Let P(x) and Q(x) be polynomials with real coefficients such that for infinitely many positive integers n, P(1)P(2)…P(n)=Q(n!).
Let d be the degree of P(x) and e the degree of Q(x).
For large n, P(1)P(2)…P(n) is a product of n terms, each of degree d, so the degree of P(1)P(2)…P(n) as a polynomial in n is dn (since each P(k) is a degree d polynomial in k, and the product over k=1 to n gives degree dn in n).
On the other hand, Q(n!) is a polynomial evaluated at n!, so as n grows, n! grows very rapidly. The degree of Q(n!) as a function of n is e times the degree of n! as a function of n, but n! is not a polynomial in n.
But for the equality to hold for infinitely many n, the growth rates must match. Consider the leading behavior:
Let P(x)=adxd+… and Q(x)=bexe+….
Then for large n:
P(1)P(2)…P(n)∼adn(1⋅2⋅⋯⋅n)d=adn(n!)d
So P(1)P(2)…P(n)∼adn(n!)d.
On the other hand,
Q(n!)∼be(n!)e
So for the equality to hold for infinitely many n, we must have e=d and adn=be for all n large enough, which is only possible if ad=1 and be=1 (since adn is exponential in n unless ad=1 or ad=0).
But if ad=1, then P(x) must be xd (since otherwise the lower degree terms will affect the product for large n), and Q(x)=xd.
Let us check this:
If P(x)=xd, then P(1)P(2)…P(n)=(1d)(2d)…(nd)=(1⋅2⋅⋯⋅n)d=(n!)d.
If Q(x)=xd, then Q(n!)=(n!)d.
So P(1)P(2)…P(n)=Q(n!) for all n.
Now, suppose P(x) is not a monomial. Then for large n, P(k)∼adkd, so P(1)P(2)…P(n)∼adn(n!)d, but the lower degree terms will affect the product by a factor that is not a polynomial in n!, so the equality cannot hold for infinitely many n.
Therefore, the only solutions are:
P(x)=xd,Q(x)=xd,for some integer d≥0.
Thus, all polynomials P(x)=xd and Q(x)=xd for d≥0 are solutions.