3.3 The Laplace Expansion 31
The retainer minor is signless and equal toR. The sign of the cofactor is
(−1)
k
, where
k=
n
∑
r=1
(n+2r)=2n
2
+n.
Hence, the cofactor is equal to (−1)
n
Q. Part (b) of the lemma follows.
Similar arguments can be applied to more general determinants. LetXpq,
Ypq,Zpq, andOpqdenote matrices withprows andqcolumns, whereOpq
is null and let
An=
∣
∣
∣
∣
Xpq Yps
Orq Zrs
∣
∣
∣
∣
n
, (3.3.14)
wherep+r=q+s=n. The restrictionp≥q, which impliesr≤s, can
be imposed without loss of generality. IfAnis expanded by the Laplace
method taking minors from the firstqcolumns or the lastrrows, some
of the minors are zero. LetUmandVmdenote determinants of orderm.
Then,Anhas the following properties:
a.Ifr+q>n, thenAn=0.
b.Ifr+q=n, thenp+s=n,q=p,s=r, andAn=XppZrr.
c. Ifr+q<n, then, in general,
An= sum of
(
p
q
)
nonzero products each of the formUqVs
= sum of
(
s
r
)
nonzero products each of the formUrVr.
Property (a) is applied in the following examples.
Example 3.2. Ifr+s=n, then
U 2 n=
∣
∣
∣
∣
En, 2 r Fns Ons
En, 2 r Ons Fns
∣
∣
∣
∣
2 n
=0.
Proof. It is clearly possible to performnrow operations in a single step
andscolumn operations in a single step. RegardU 2 nas having two “rows”
and three “columns” and perform the operations
R
′
1
=R 1 −R 2 ,
C
′
2 =C 2 +C 3.
The result is
U 2 n=
∣
∣
∣
∣
On, 2 r Fns −Fns
En, 2 r Ons Fns
∣
∣
∣
∣
2 n
=
∣
∣
∣
∣
On, 2 r Ons −Fns
En, 2 r Fns Fns
∣
∣
∣
∣
2 n
=0
since the last determinant contains ann×(2r+s) block of zero elements
andn+2r+s> 2 n.