198 5. Further Determinant Theory
=
∑
i,j,r
E
ijr,ijr
,
which is equivalent to (5.4.23). The application of a suitably modified the
fourth line of (5.4.4) to (5.4.23) yields (5.4.24). Identities (5.4.27)–(5.4.29)
follow from (5.4.8), (5.4.22), (5.4.24), and the identities
3 crcs=(c
2
r
+crcs+c
2
s
)−(cr−cs)
2
,
6 c
2
r
=2(c
2
r
+crcs+c
2
s
)+(cr−cs)
2
+3(c
2
r
−c
2
s
),
6 c
2
s
=2(c
2
r
+crcs+c
2
s
)+(cr−cs)
2
−3(c
2
r
−c
2
s
).
5.4.5 Three Further Identities
The identities
∑
r,s
(c
2
r+c
2
s)(cr−cs)E
rs
=2
∑
r,s
c
2
rE
rs,rs
+
1
3
∑
r,s,u,v
E
rsuv,rsuv
, (5.4.36)
∑
r,s
(c
2
r
−c
2
s
)(cr+cs)E
rs
=2
∑
r,s
cr(cr+cs)E
rs,rs
−
1
6
∑
r,s,u,v
E
rsuv,rsuv
, (5.4.37)
∑
r,s
crcs(cr−cs)E
rs
=
∑
r,s
crcsE
rs,rs
−
1
4
∑
r,s,u,v
E
rsuv,rsuv
(5.4.38)
are more difficult to prove than those in earlier sections. The last one has
an application in Section 6.8 on the KP equation, but its proof is linked to
those of the other two.
Denote the left sides of the three identities byP,Q, andR, respectively.
To prove (5.4.36), multiply the second equation in (5.4.10) by (c
2
i+c
2
j),
sum overiandjand refer to (5.4.4), (5.4.6), and (5.4.27):
P=
∑
i,j,r,s
(c
2
i
+c
2
j
)E
is
E
rj
+
∑
i,j,r
(c
2
i
+c
2
j
)E
ir
E
rj
=
∑
j,r
E
rj
∑
i,s
(s→j)
c
2
iE
is
+
∑
i,s
E
is
∑
j,r
(r→i)
c
2
jE
rj
+
∑
i,j,r
(c
2
i+c
2
j)
∂E
ij
∂xr