30 3. Intermediate Determinant Theory
Now, interchange the dummies wherever necessary in order thatp<
q<rin all sums. The result is
A=
∑
p<q<r
[
aipajqakr−aipajrakq+aiqajrakp
−aiqajpakr+airajpakq−airajqakp
]
Aijk,pqr
=
∑
p<q<r
∣ ∣ ∣ ∣ ∣ ∣
aip aiq air
ajp ajq ajr
akp akq akr
∣ ∣ ∣ ∣ ∣ ∣
Aijk,pqr
=
∑
p<q<r
Nijk,pqrAijk,pqr, i<j<k,
which proves the Laplace expansion formula from rowsi,j, andk.
3.3.3 Determinants Containing Blocks of Zero Elements
LetP,Q,R,S, andOdenote matrices of ordern, whereOis null and let
A 2 n=
∣
∣
∣
∣
PQ
RS
∣
∣
∣
∣
2 n
.
The Laplace expansion ofA 2 ntaking minors from the first or lastnrows or
the first or lastncolumns consists, in general, of the sum of
(
2 n
n
)
nonzero
products. If one of the submatrices is null, all but one of the products are
zero.
Lemma.
a.
∣
∣
∣
∣
PQ
OS
∣
∣
∣
∣
2 n
=PS,
b.
∣
∣
∣
∣
OQ
RS
∣
∣
∣
∣
2 n
=(−1)
n
QR
Proof. The only nonzero term in the Laplace expansion of the first
determinant is
N 12 ...n;12...nA 12 ...n;12...n.
The retainer minor is signless and equal toP. The sign of the cofactor is
(−1)
k
, wherekis the sum of the row and column parameters.
k=2
n
∑
r=1
r=n(n+1),
which is even. Hence, the cofactor is equal to +S. Part (a) of the lemma
follows.
The only nonzero term in the Laplace expansion of the second
determinant is
Nn+1,n+2,..., 2 n;12...nAn+1,n+2,..., 2 n;12...n.