Upload
lamtram
View
220
Download
0
Embed Size (px)
Citation preview
MATA33 Determinant notes, mostly stolen from Prof Eric Moore
Definition: determinant (i) The determinant of a 1 by 1 matrix [a] is a. (ii) Suppose a definition is provided for a n−1 by n−1 determinant. Define the determinant of the n*n matrix by:
=∑ 1 det
where is the matrix obtained from A by deleting the ith
row and jth
column. The matrix is sometimes called the
ijth
minor matrix of A. Think of this approach as the definition by expansion along the first row or expansion by minors along the first row.
nnnn
n
n
aaa
aaa
aaa
...
......
......
......
...
...
det
21
21221
11211
Properties of Determinants Let A be a n x n matrix. If A has a row or column of zeros, det A = 0. If A is a triangular matrix, det A is the product of the elements on the main diagonal. det I = 1.
det A = det AT
.
Interchanging any two rows (or columns) multiplies the
determinant by −1.
Properties of Determinants
Since det A = det AT
let’s state the remaining properties only
in terms of rows.
If a row of A is multiplied by a constant k, the
determinant of the resulting matrix is k det A.
o Note that det(kA)= kn det A.
Let A, B, C be n × n matrices that are identical except
that the ith
row of A is the sum of the ith
rows of B and C
then det A = det B + det C.
o but det (E+F) ≠ det E + det F in general.
o This property + previous property =⇒
determinant is linear in each row.
If two distinct rows are identical then det A = 0.
If two distinct rows are proportional then det A = 0. .
Adding a multiple of one row to another row does not
change the value of the determinant.
det(AB) = (det A) (det B).
The determinant of a matrix consists of sums and
products of its entries. If the entries are polynomials in
some variable, say x, then the determinant is a polynomial
in x. Often it is of interest to know the values of x that
make the determinant zero.
A square matrix is invertible if and only if det A ≠ 0.
The homogeneous linear system AX = 0 has the unique
solution X = 0 if and only if det A ≠ 0.
Let be an n × n matrix. We define the cofactor of aij denoted cij by
1
The n × n matrix C = [cij] is called the cofactor matrix of A.
(Recall that is the matrix obtained from A by deleting
the ith
row and jth
column and is sometimes called the ijth
minor
matrix of A. det can be called the ijth
minor of A or the
minor of the element aij of A)
In this context, a cofactor is sometimes called a signed minor.
Note that cij is a scalar (real number) but ij is an
(n − 1) × (n − 1) matrix.
We can now restate the definition of determinant in terms of cofactors. (i) The determinant of a 1 by 1 matrix [a] is a.
(ii) Suppose a definition is provided for a n−1 by n−1 determinant. Define = ∑ = a11c11 + a12c12 + …….. + a1nc1n
where cij is the cofactor of aij.
nnnn
n
n
aaa
aaa
aaa
...
......
......
......
...
...
det
21
21221
11211
Cofactor Method of Inverting a Matrix The transpose of the cofactor matrix C of A is called the
classical adjoint of A and denoted by adj A; i.e., adj A = CT
. Theorem: If A is any square matrix, then A ( adj A ) = (det A ) I = ( adj A ) A. In particular, if det A ≠ 0, the inverse of A is given by
,
where adj A = CT
, C being the cofactor matrix of A. (Comment: this is the classical mathematician’s way of talking about matrix inverses, and is good for discussing properties of inverses. Faster methods including reduction are used more in practice.)
Example on Matrix Inversion by Cofactors
3 52 7
1
: 7 25 3
21 10
7 5
2 321 10
7/11 5/112/11 3/11
Does this work as an inverse of A? Yes: look below
711
511
211
311
3 52 7
1 00 1
Her
Fo
(A
=(
res a big
or exam
-1)1,2
(-1)2+1 d
Fr
g picture
mple, on
, /
det
rom M
e of the c
ne elem
Mathwo
cofactor
ment of
1Typee
/detA
orld.W
method:
f the inv
1
equationher
A= det
Wolfram
:
verse
re.
t
m.com:
/
/d
:
detA
Up
Ad
M
Ad
p-dow
dd cop
ultiply
dd the
wn tric
ied col
y as sho
produ
ck for
lumns
own be
ucts, sig
2*2 or
to the
elow (p
gns po
r 3*3 m
right:
picture
sitive (
matric
e from
(down)
ces (no
m Cliff’
) or ne
ot hig
s notes
egative
her n)
s)
e (up):
)
M
MATA3
Solu Not i = (1 =(1- = (1 =(1- if x=
33 Win
tion
inverti
-x)(-1-
-x)(1+x
-x)(1+
-x)(1+x
=3, -1,
nter 2
ible if
-x)[(2-
x) [(2-
+x) [21
x)(7+x
1 or -
2008 F
0=det
-x)(-6-
-x)(6+
1-4x -
x)(3-x)
7
Final Q
t A
-x) – 9]
x)+9]
-x2 ]
)
Q6
]
]
MATA33 Final Summer 2009 MCQ8
Assume A is an n x n matrix where n ≥ 2: Exactly how many of the following mathematical statements are always true ? (i) det(2A) = 2n det(A) (ii) det(In + A) = 1 + det(A) (iii) If AX = 0 has infinitely many solutions, then A is not invertible. (iv) If AB = AC, then B = C (v) If R is the reduced form of A and R has exactly n non-zero rows, then A is invertible. (a) 1 (b) 2 (c) 3 (d) 4 (e) 5 (B) Solution (i) No: every element is *2 so det(2A)=2n det(A) (ii) No – lots of counterexamples (iii) Yes – if A-1 exists then A-1 multiplies zero vector to get the unique trivial solution, all zeroes (iv) No – eg A is the zero matrix (v) Yes – there’s n unknowns, n lin indept eqns
MATA33 W09 T2 LAQ1 (Cont)
Alternative derivation of M-1, using for x=6 that detM=2(6-7)2=2.
2 6 62 7 62 7 7
1
: 7 2 00 2 26 0 2
2
7 0 62 2 00 2 2
2
3.5 0 31 1 00 1 1
Which checks with Prof Grinnell’s answer
So(i) (ii)unBu(iiichan(iv)(v)
olution False
) n×n snique sout for Bi) mayboosing
nd the rv) it’s in) true fr
since wset of nolution
B=0, X=be was
g n differest zern the syfrom th
MAT
we are linear iff =0 is c more derent mros, canyllabuse reduc
TA33
learninr equatif P islearly odirectly
matricesn build s that Pction m
S09 T2
ng ‘P iions PXs inverone soly discus, each
d an invPT invermethod
2 MCQ
invertibX=B irtible lution.
ussed inh n*1 averse frrtible ifof inve
Q5
ble iff din n un(e.g. te
n 2009,and withrom theff P inversion
detP≠0nknownext pag
, but byh a singe solutivertible
0’ ns has age 282).
y gle 1 ions Y.e
a .
MATA33 S09 T2 LAQ3b (Alternative solution)
Alternative derivation of F-1,
71 2 05 8 00 0 7
1
: 56 35 014 7 00 0 18
8 ∗ 7 2 ∗ 5 ∗ 7
56 14 035 7 00 0 18
126
4/9 1/9 05/18 1/18 00 0 1/7
Which checks with Prof Grinnell’s answer by reduction
Cramer’s Rule
If A is an invertible n × n matrix, the solution to the system
AX = B
of n equations in the variables x1,x2, ...., xn is given by:
,
,
,
where, for each k, Ak is the matrix obtained from A by
replacing the kth.
column of A by B.
Gi
Cr
Π1
Sta
iven n l
ramer's
1=
atemen
linear e
rule gi
nt of C
equatio
ives, x1
Cramer
ons with
1 = Π1,
r’s rule
h n var
define
e (from
riables
ed belo
m Wiki
ow:
ipedia)
.
)
Thx1
(1)thedet(2)det
Wsys
Π1
he abovcorrect
) Addine determtA = de) Multitermin
e start bstem, h
1=
V
ve Cramtly usin
ng a muminantetAT) iplying ant by
by subhence: t
Verific
mer’s rung two
ultiple t. And
all elemk. An
stitutinthis ch
cation
ule expproper
of a rowd same
ments ond sam
ng for anges Π
of Cra
pressiorties of
w to anfor co
of one me for a
Π1 to
amer’s
on Π1 caf determ
nother rlumns
columna row.
from
rule
an be sminants
row do(remem
n by k m
m the eq
shown ts
oes not mber th
multip
quations
to give
changehat
lies the
s of the
e
e
e
e
Nox2,
Π1
ow sub, the thi
1=
Ver
tract frird colu
rificatio
rom theumn m
on of C
e first cmultiplie
Crame
columned by x
er’s rul
n the sex3, and
le (con
econd mso on
nt)
multipl
lied by
No
. Π
ow we
Π1=
Ver
factor
rificatio
out the
on of C
e x1 in t
Crame
the first
er’s rul
t colum
le (con
mn of th
nt)
he nummerator:
In k-twi
So
the samth coluill be eq
o Cram
Ver
me waumn of tqual to
er’s rul
rificatio
y, if ththe ma xk. As
le is ve
on of C
he colutrix of a resu
erified.
Crame
mns ofthe sys
ult we g
er’s rul
f b's isstem ofget that
le (con
s used tf equatit
nt)
to replaions the
ace thee result
e t