MFSAS Exam 2014 Solutions

Maths for Signals and Systems Exam 2014-Solutions
1.
a)
(i)
I is the identity matrix, O are zero matrices and F is a matrix that is related to the
special solutions of the system.
The dimensions of the individual matrices are given in the subscripts
Fr( n r )
I r r
R
The subscripts in the individual matrices reveal their
O ( mr )r O ( mr )( n r )
corresponding sizes.
(ii) Due to the special column rearrangement of R the special solution vectors contain the
pivot variables in their first r elements and the free variables in their the last n r
elements. As already mentioned above, each special solution has one free variable
equal to 1 and the other free variables are all zero. Therefore, the null space matrix N
X r( n r )
is given by N
where X r( n r ) is an unknown matrix of size r (n r ) .
I ( n r )( n r )
Knowing that RN O we get:
Fr( n r ) X r( n r )
I r r
RN
O ( m r )r O ( m r )( n r ) I r( n r )
I rr X r( n r ) Fr( n r ) I r( n r ) X r( n r ) Fr( n r )
O ( m r )( n r )
( m r )( n r )
X r( nr ) Fr( nr ) Or( nr ) X r( nr ) Fr( nr )

Fr( n r )
Therefore, N
.
I ( n r )( n r )
(iii) We assume that the echelon form is obtained without any permutations. In case of a
3 4 matrix, the maximum rank is 3. In that case we are given that the dimension of
the null space is 1. Since the rows of the matrix are 4-dimensional, we know
immediately that the dimension of the row space is 3. Therefore, the rank of the matrix
1 0 0 a
is 3. In that case the echelon matrix must be of the form R 0 1 0 b . The null
0 0 1 c
space is obtaine by looking for random vectors x , for which Ax 0 ERx 0 . This
implies Ax 0 ERx 0 the Rx 0 , since the matrix E is a square, full rank
matrix.
4
1 0 0 a 0 4 2a 0 a 2
1 0 0 2
0 1 0 b 2 0 2 2b 0 b 1 R 0 1 0 1
0 0 1 c 0
0 0 1 0
2c 0 c 0
2
(iv) We know that EA R where E Eij is the product of all elimination matrices used
ij
in the procedure. If the rank of matrix A is r then the last m r rows of R are zero
rows. Therefore, from the equation EA R we see that each of the last m r rows of
E multiplied with A from the left gives a zero row vector. This verifies the fact that
the last m r rows of E belong to the left null space, since they satisfy the relationship
xT A 0T . Due to the method that we use to construct E , it can be shown easily that
E is a full rank matrix (rank is m ) and therefore its last m r rows are independent.
Since these rows belong to the left null space and knowing that the left null space has
dimension m r , we can say that the last m r rows of E form a basis of the left null
space.
1
b)
(i) The echelon form has two pivots. Therefore, the rank of the matrix is 2. The rows are 4dimensional and therefore, we have 2 free variables. We solve the system
x
1 0 3 4
0 1 1 0 y 0 for z 1, w 0 and z 0, w 1 .
z
0 0 0 0
w
For z 1, w 0 we have x 3 0 x 3 and y z 0 y 1 .
For z 0, w 1 we have x 4 0 x 4 and y 0 .
3
4
1
0
Therefore, the special solutions are

and and they form a basis for the
1
0

0
1
nullspace.
The row space does not change with elimination and therefore, any two independent
rows of the echelon matrix, for example rows 1 and 2, form a basis for the row space.
(ii) 5(row1)+4(row2)
(iii) A has rank 2 and AT is 4 by 3 so its null space has dimension 3-2=1.
c)
(i) The pivots of A1 are equal to 1/(pivots of A ) because det A1 1/(det A) .
A I
(ii) Multiply row 1 by A1 and add to row 2 to obtain
1
O A
(iii) The determinant is +1. Exchange the first n columns with the last n . This produces a
I A
n
factor (1) n and leaves
which is triangular with determinant (1) . Then
O
(1) n (1) n 1 .
2.
a)
(i)
P A( AT A) 1 AT
(ii) AT A is symmetric and therefore ( AT A) 1 is symmetric. (To prove this we use the
property ( A1 )T ( AT ) 1 .)
PT [ A( AT A) 1 AT ]T ( AT )T [( AT A) 1 ]T AT A( AT A) 1 AT P
P 2 [ A( AT A) 1 AT ][ A( AT A) 1 AT ] [ A( AT A) 1 ( AT A)( AT A) 1 AT ] P
If A is square and invertible its column space is the entire n-dimensional space and
therefore the projection of b onto A should be b . In that case P AA1 ( AT ) 1 AT I .
(iii) If b is perpendicular to the column space of A then Pb AA1 ( AT ) 1 AT b 0 .
(iv) e b Pb , AT e AT b AT p AT b AT Pb 0
b)
(i) The projection matrix P is of the form P A( AT A) 1 AT with A being the column
vector 1 2 4 . Therefore, it projects onto the column space of A which is the line
T
c1 2 4 .
T
22
1
1302
(ii) The error is e b Pb 23 and the distance is e
.
21
21
17
(iii) Since P projects onto a line, its three eigenvalues are 0,0,1. Since P is symmetric, it
has a full set of (orthogonal) eigenvectors, and is then diagonalizable.
c)
(i) We have a set of equations
C 2D 0
CD0
C 1
C D 1
C 2D 1
and therefore the system is
1 2
0
1 1

C 0
1 0 1
D
1 1
1
1 2
1
The system doesnt have a solution since the solutions that is obtained from 2 of the
equations doesnt satisfy the rest.
(ii) The projection matrix is
3/ 5 2 / 5
2 / 5 3 / 10
1/ 5 1/ 5
1 / 10
0
1 / 5
0
and the projection vector is
1/ 5
1/ 5
1/ 5
1/ 5
1/ 5
1 / 5
1 / 10
0
1/ 5 1/ 5
3 / 10 2 / 5
2 / 5 3 / 5
0
0
3 / 10
3/ 5
9 / 10
6 / 5
Approximate solution is C 6 / 10 and D 3 / 10 . Straight line is 6 / 10 3t / 10 .
(iii) error vector is
0 0 0
0 3 / 10 3 / 10

e b p 1 3 / 5 2 / 5

1 9 / 10 1 / 10
1 6 / 5 1 / 5
3.
a)
(i) By solving the system Ax 0 , it is straightforward to see that the null space has
1
1
dimension 1 and its basis is the vector .
1
1
(ii) Matrix B is singular. All rows are identical and therefore the row space is of dimension
1. Therefore, 3 out of 4 eigenvalues of B must be 0. The remaining non-zero
eigenvalue can be found from the trace of B and it is equal to 4. Therefore, the
eigenvalues of B are 4,0,0,0.
3 1 1 1
1 3 1 1
4I B
(iii) AT A
1 1 3 1
1 1 1 3
If x is an eigenvector of B with eigenvalue , then
Bx x 4 x Bx 4 x x (4I B) x (4 ) x AT Ax (4 ) x
Therefore, the eigenvalues of AT A are obtained from the eigenvalues of B , by
reversing the sign and adding 4. Thus, the eigenvalues of AT A are 0,4,4,4.
(iv) The non-zero singular values of matrix A , are the square roots of the eigenvalues of
AT A . Therefore, these are 2,2,2. The matrix AT A is diagonalized through the formula
AT A VV T where A UV T . The matrix V has the eigenvectors of AT A in its
columns.
The eigenvector of AT A that corresponds to 0 is of the form x x x xT with
1
magnitude 4x 2 . If we look for an orthonormal eigenvector then 4 x 2 1 x .
2
T
1 1 1 1
Therefore, a column of V is
.
2 2 2 2
b)
1
We select the first orthogonal direction to be A a 1 and A 2 . Therefore,
0
1 / 2
A
q1
1 / 2 . The second direction is:
A
0
1
1
1 1
1 1 0 1
AAT
1
1
B b T b 0 11 1 00 0 1 1 0 0
2
2
A A
1
0
1 1
0 0 0 1
1
1 1 / 2
0 1 1 1 / 2
2
1
0 1
B
1
1
6
1
4
4
4
3
2
1/ 2
2
1
q2
1 / 2
3
6
1
1/ 2
BBT 2
1 / 21 / 2
T
B B 3
1
1
1

2
1/ 4 1/ 4 1/ 2
2
1 / 2 1 1 / 4 1 / 4 1 / 2
3
1 / 2 1 / 2
1
1 / 4 1 / 4 1 / 2 0
1/ 4 1/ 6
BBT
2
2
c 1 / 4 1 / 4 1 / 2 1 1 / 4 1 / 6
3
3
BT B
1 / 2 1 / 2
1 / 2 1 / 3
1 1
1 1 0 0 1 / 2
AAT
1
c 1 1 0 1 1 / 2
2
AT A
0 0 0 1 0
2 / 3
AAT
BBT
c T c 1 / 3
BT B
A A
1 / 3
0 2 / 3 2 / 3
AAT
BBT
c T c T c 1 1 / 3 2 / 3
B B
A A
1 1 / 3 2 / 3
2
c
3
1
c
1
q3
1
c
3
1
1
2
1
Therefore, Q
2
q1T a 2 , q1T b
R 0
1
2
3
6
0
1
6
1
6
2
6
1
2
3
1
3
1
, q1T c
1
2
, q2T b
3
6
, q2T c
1
6
, q3T c
2
3
2
1
6
2

MFSAS Exam 2014 Solutions

Uploaded by

Copyright:

Available Formats

MFSAS Exam 2014 Solutions

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MFSAS Exam 2014 Solutions

Uploaded by

Copyright:

Available Formats

Maths for Signals and Systems Exam 2014-Solutions

X r( nr ) Fr( nr ) Or( nr ) X r( nr ) Fr( nr )

Therefore, the special solutions are

(i) The pivots of A1 are equal to 1/(pivots of A ) because det A1 1/(det A) .

(i) We have a set of equations

and therefore the system is

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.