106 4. Particular Determinants
Let
An=|φi+j− 2 |n=|φm|n, 0 ≤m≤ 2 n− 2 ,
Bn=|x
i+j− 2
φi+j− 2 |n=|x
m
φm|n, 0 ≤m≤ 2 n− 2.
(4.8.9)
Lemma.
a.Bn=x
n(n−1)
An.
b.B
(n)
ij
=x
n(n−1)−(i+j−2)
A
(n)
ij
c. B
ij
n =x
−(i+j−2)
A
ij
n.
Proof of (a).Perform the following operations onBn: Remove the factor
x
i− 1
from rowi,1≤i≤n, and the factorx
j− 1
from columnj,1≤j≤n.
The effect of these operations is to remove the factorx
i+j− 2
from the
element in position (i, j).
The result is
Bn=x
2(1+2+3+···+n−1)
An,
which yields the stated result. Part (b) is proved in a similar manner, and
(c), which contains scaled cofactors, follows by division.
4.8.2 Hankelians Whose Elements are Differences
Thehdifference operator ∆his defined in Appendix A.8.
Theorem.
|φm|n=|∆
m
h
φ 0 |n;
that is, a Hankelian remains unaltered in value if eachφmis replaced by
∆
m
h
φ 0.
Proof. First Proof.Denote the determinant on the left byAand perform
the row operations
R
′
i=
i− 1
∑
r=0
(−h)
r
(
i− 1
r
)
Ri−r,i=n, n− 1 ,n− 2 ,..., 2 , (4.8.10)
onA. The result is
A=
∣
∣
∆
i− 1
h
φj− 1
∣
∣
n
. (4.8.11)
Now, restore symmetry by performing the same operations on the columns,
that is,
C
′
j
=
j− 1
∑
r=0
(−h)
r
(
j− 1
r
)
Cj−r,j=n, n− 1 ,n− 2 ,..., 2. (4.8.12)
The theorem appears. Note that the values of i and j are taken in
descending order of magnitude.