Reduced Incidence Matrix

Let G be a connected
digraph with “n” nodes and “b”
branches. Let Aa be the Incidence
Matrix of G. The (n-1) x b matrix
A obtained by deleting any one
row of Aa is called a Reduced-
Incidence Matrix of G.
 1

e1
1
2 KCL Equations:
e2 e3
3
3
1 i1 + i2 − i6 = 0
−i1 − i3 + i4 = 0
2
2
6 4 5 3 −i2 + i3 + i5 = 0
4

node Branch no. i1

no. 1 2 3 4 5 6 i 
 2
Ai = 0 ⇒ 1  1

1 0 0 0 −1 i3 0
2 −1
 0 −1 1 0 0   
i4  = 0
3  0 −1 1 0 1 0 i5  0
 
i6 
A 0
i
A is called the reduced Incidence Matrix
of the diagraph G relative to datum node 4 .
 Reduced Incidence Matrix A
Let G be a connected digraph with “n”
nodes and “b” branches. Pick any node as the
datum node and label the remaining nodes
arbitrarily from 1 to n-1 . Label the branches
arbitrarily from 1 to b.
The reduced incidence matrix A of G
is an (n-1) x b matrix where each row j
corresponds to node j , and each column k,
corresponds to branch k, and where the jkth
element ajk of A is constructed as follow:
1 , if branch k leaves node j

ajk = −1 , if branch k enters node j
0 ,
 if branch k in not connected to node j
Note:
Each branch k connected to the

datum node has only one

non-zero entry in the kth column

of A

The (n-1) KCL equations derived

from Ai = 0
are linearly independent.
Theorem

Ai = 0
gives the maximum possible
number of linearly-independent
KCL equations for a connected
circuit.
Relationship between A and Aa

Let Aa be the n x b Incidence

matrix of a connected digraph G with

“n” nodes and “b” branches.

By deleting any row

corresponding to node m from Aa, we
obtain the reduced incidence matrix
A of G relative to the datum node m .
 e1 1
i1
v1 + i2 +
− D1
v2
D2 −
− e2 − v3 + 3
v6 i3
D6 D3 e3
+ 2 i4
i5
+ D4 D5 +
v4 v5
− −
1
4
2
1
3 3
KVL around closed node sequence:
2

4 1 3 2 1 : v2 + v3 − v1 = 0
6 5
2 3 4 2 : − v3 + v5 − v4 = 0
4 1 3 4 2 1
: v2 + v5 − v4 − v1 = 0
These 3 KVL equations are not linearly-independent because the
3rd equation can be obtained by adding the first 2 equations:
(v2 + v3 − v1 ) + (−v3 + v5 − v4 )
= v2 − v1 + v5 − v4 = 0
 1 Choose 4 as datum node for digraph G
e1
2
1 KCL Equations:
e2 3 e3
2 3 1 i1 + i2 − i6 = 0
4 5
2 −i1 − i3 + i4 = 0
6
3 −i2 + i3 + i5 = 0
4
node Branch no. i1
no. 1 2 3 4 5 6 i 
 2
Independent 1  1 1 0 0 0 −1 i3 0
KCL Equations 
2 −1 0 −1 1 0 0   0
 i
  =
Ai = 0 ⇒ 
4
3  0 −1 1 0 1 0 i5  0
 
Independent i6 
KVL Equations A 0
i
 v1   1 −1 0 v1 = e1 − e2
v   1 0 −1 v2 = e1 − e3
 2   e 
 v3   0 −1 1  1  v3 = −e2 + e3
 =  e2  ⇒
v4   0 1 0  e  v4 = e2
 v5   0 0 1  3  v5 = e3
    e
 v6   −1 0 0 v6 = −e1
v A T
Note:
Since each equation k from the
system of KVL equations generated
from

T
v =A e
contains the variable vk, which
appears in no other equations,
this system of “b” KVL equations
are guaranteed to be
linearly independent.
 KVL
each branch k connected between

node m and the datum node 0

has a KVL equation involving

only one node-to-datum voltage em:

vk = em
 How to write An Independent
System of KCL and KVL
Equations
Let N be any connected circuit and let the digraph
G associated with N contain “n” nodes and “b”
branches. Choose an arbitrary datum node and
define the associated node-to-datum voltage
vectore , the branch voltage vector v , and the
branch current vector i . Then we have the
following system of independent KCL and
KVL equations.
(n-1) Independent KCL Equations :

Ai = 0
b Independent KVL Equations :

v = A e
T
 ik
+ For each 2-terminal resistor Rk,
vk Rk we have 2 unknown
variables {vk, ik}
_

ij ik
+ For each 3-terminal resistor,
+
we have 4 unknown
vj vk
variables {vj, vk, ij, ik}
_ _

ij ik
For each 2-port resistor,
+ +
vj vk we have 4 unknown
_ _
variables {vj, vk, ij, ik}

∴
There are more unknown variables
than equations derived from
constitutive relations of the resistors.