1. **Reduction to the case k=2**:
- We start by noting that it suffices to prove the result for k=2. If detA=b1b2⋯bk, we can write detA=(b1⋯bk−1)bk. Then, we can find integer matrices Bk−1′ and Bk such that A=Bk−1′Bk with detBk−1′=b1⋯bk−1 and detBk=bk. We can repeat this process iteratively.
2. **Case k=2**:
- Suppose A is an n×n matrix with integer entries and detA=bc for some integers b and c. We need to show that there exist n×n integer matrices B and C such that A=BC and detB=b, detC=c.
3. Upper Triangular Matrices:
- Multiplying a matrix A by elementary matrices representing row-addition operations does not alter the determinant of A. If A′=EA is a row-reduced form of A with E a product of elementary matrices, then detE=1 and E has integer entries. Hence, it suffices to prove the problem for upper triangular matrices.
4. **Induction on n**:
- Suppose A is an upper triangular matrix with integer entries and detA=bc. We will use induction on n.
5. **Base Case n=2**:
- Consider A=(m0xn). We need to find B and C of the form:
B=(t10yt2),C=(t1m0zt2n)
for some integers t1,t2,y,z such that t1∣m and t2∣n. We have detA=mn=bc and thus we need t1t2=b. Finally, y and z must satisfy x=t1z+t2ny. For a given x, we can guarantee a solution (y,z) exists only if gcd(t1,t2n)=1.
6. Lemma:
- If b∣mn, then we can find t1∣m and t2∣n such that b=t1t2 and gcd(t1,t2n)=1.
- Proof of Lemma: Assume mn and b have the same set of prime factors. It suffices to prove the lemma when b, m, and n are all powers of the same prime. If b∣n, choose t1=nb and t2=n. If n∣b, choose t1=1 and t2=b.
7. Inductive Step:
- Suppose A is an n×n matrix and let A′ be the lower-right (n−1)×(n−1) submatrix of A. Denote m=A1,1 and n=detA′. In block form:
A=(m0XA′)
Since A is upper triangular, detA=A1,1detA′=mn=bc. By the lemma, we can find b1,c1,b2,c2 such that m=b2c2 and n=b1c1 with gcd(b2,c1)=1. By the inductive hypothesis, we can find (n−1)×(n−1) matrices B′ and C′ such that A′=B′C′ and detB′=b1, detC′=c1.
8. **Constructing B and C**:
- Take B and C in block form:
B=(b20YB′),C=(c20ZC′)
We need row vectors Y and Z such that b2Z+YC′=X. Write Y=[y1,…,yn−1] and Z=[z1,…,zn−1] and X=[x1,…,xn−1].
9. **Solving for Y and Z**:
- Matching the first coordinate, we have b2z1+y1C1,1′=x1. There exists an integer solution (y1,z1) because gcd(b2,C1,1′)∣gcd(b2,det(C′))=gcd(b2,c1)=1.
- Matching the second coordinate, we have b2z2+y1C1,2′+y2C2,2′=x2, which becomes b2z2+y2C2,2′=x2−y1C1,2′. This admits a solution (y2,z2) for the same reason as above.
- Continue in this manner to determine Y and Z satisfying the relation. The key is that b2 is relatively prime to c1=detC′ and thus to all the diagonal entries of C′.
■