ECE 802-601 Homework 1: Orthonormal Bases and Unitary Transforms, Assignments of Electrical and Electronics Engineering

The homework assignment for a university course on electrical and computer engineering (ece), specifically ece 802-601, from spring 2008. The assignment covers topics such as orthonormal bases, linear independence, gram-schmidt orthogonalization, inner products, unitary transforms, and parseval's theorem. Students are required to complete problems involving square integrable functions, symmetric and antisymmetric functions, and fourier transforms.

Typology: Assignments

Pre 2010

Uploaded on 07/23/2009

koofers-user-l7g
koofers-user-l7g 🇺🇸

9 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
ECE 802-601 Homework 1
Spring 2008
Due January 29, 2008
1. Consider the space of square integrable functions on the interval ],[
π
π
, ],[
2
π
π
L,
and the associated othonormal basis given by ,...2,1,
2
)sin(
,
2
)cos(
,
2
1=
n
nxnx
πππ
Consider the following two subspaces, S: space of symmetric functions on ],[
π
π
and
A: space of antisymmetric functions on ],[
π
π
.
a) Show how any function ],[)( 2
π
π
Lxf can be written as )()()( xfxfxf as +
=
,
where AxfSxf as )(,)(.
b) Give orthonormal bases for S and A.
2. Let for and
n
ntty =)( K,2,1,0=n11
t.
a) Show that
{
is a linearly independent set in
}
K,,, 210 yyy ]1,1[
2
L.
b) Write a MATLAB function that implements the Gram-Schmidt orthogonalization
process. Apply it to
{
and plot the first four orthonormal vectors. Given a
finite or countably infinite set of linearly independent vectors , we can construct an
orthonormal set with the same span as follows:
}
K,,, 210 yyy
i
y
i
x
Start with
0
0
0y
y
x= .
Then, recursively set
0011
0011
1,
,
xxyy
xxyy
x><
>
<
=
Therefore, i
k
iikk
kk
kk
kxxyv
vy
vy
x><=
=
=
1
0
,,
3. Inner products are preserved by unitary transforms.
a) Prove the general Parseval’s theorem, for any
in a vector space and an orthonormal basis,
*
2121 ,,, >><<>=< ii
i
xyxyyy
21 ,yy
{
}
K,,, 321 xxx .
b) Using the result in part (a), show that
c) For )/(
)/sin(
)( Tt
Tt
thT
π
=, prove that
{
}
Zn
TnTth
)( is an orthogonal basis of the space
U of functions whose Fourier transforms are bandlimited in ]/,/[ TT
π
π
.
pf2

Partial preview of the text

Download ECE 802-601 Homework 1: Orthonormal Bases and Unitary Transforms and more Assignments Electrical and Electronics Engineering in PDF only on Docsity!

ECE 802-601 Homework 1 Spring 2008 Due January 29, 2008

1. Consider the space of square integrable functions on the interval [− π , π], L 2 [− π, π],

and the associated othonormal basis given by , 1 , 2 ,... 2

sin( ) , 2

cos( ) , 2

n nx nx

Consider the following two subspaces, S: space of symmetric functions on [− π , π]and

A: space of antisymmetric functions on [− π , π].

a) Show how any function f ( x )∈ L 2 [− π, π]can be written as f ( x )= fs ( x )+ fa ( x ),

where f (^) s ( x )∈ S , fa ( x )∈ A.

b) Give orthonormal bases for S and A.

  1. Let y (^) n ( t )= tn for n = 0 , 1 , 2 ,Kand− 1 ≤ t ≤ 1.

a) Show that { y 0 , y 1 , y 2 ,K}is a linearly independent set in L 2 [− 1 , 1 ].

b) Write a MATLAB function that implements the Gram-Schmidt orthogonalization

process. Apply it to {^ and plot the first four orthonormal vectors. Given a

finite or countably infinite set of linearly independent vectors , we can construct an

orthonormal set with the same span as follows:

y 0 , y 1 , y 2 ,K^ }

y i x i

  • Start with 0

0 (^0) y

y x =.

  • Then, recursively set 1 1 0 0

1 1 0 0 1 ,

y y x x

y y x x x −< >

  • Therefore, (^) i

k

i

k k i k k

k k k v y x x y v

y v x = < > −

=

1

0

  1. Inner products are preserved by unitary transforms.

a) Prove the general Parseval’s theorem, for any

in a vector space and an orthonormal basis,

< 1 , 2 >=∑ < 1 , i >< 2 , i >

i

y y y x y x

y 1 , y 2 { x^1 , x 2 , x 3 ,K }.

b) Using the result in part (a), show that

c) For ( / )

sin( / ) ( ) t T

t T hT t

= , prove that { h^ T ( t − nT )} n^ ∈ Z is an orthogonal basis of the space

U of functions whose Fourier transforms are bandlimited in [ − π / T , π/ T ].

  1. In this problem, you will explore the effect of the length of the window and the overlap between consecutive windows on the resulting spectrogram. Let R be the length of the window and L be equal to the lapse between consecutive windows, i.e. the amount of overlap equals to R-L. Load the speech file mtlb.mat in MATLAB (this is part of MATLAB). Fix the sampling frequency as 7418 as provided by the file and the number of FFT points to 512. a) In this part of the problem, you will explore the effect of R for a fixed L. Fix L to 10 and explore the effect of R for R=32,64,128,256. Determine the value of R that is the most appropriate for this signal. b) Using the R value you determined in part (a), determine the effect of L for L=1,10,100.
  2. One approach to avoid the time-frequency resolution limitations of a short-time Fourier transform is to adapt the window size to the signal content. In this problem, you

will explore this option using a simplified algorithm. Let be the spectrogram

computed using the window function. For the following quadratic

chirp test signal,

S j ( t , ω )

g (^) j ( t )= 2 − j /^2 g ( 2 − jt )

t=-2:0.001:2; y=chirp(t,100,1,200,'q'); compute the STFT at four different scales and then choose the best scale for each time-

frequency point ( t , ω)such that the amplitude at that point is a maximum. Combine the

‘best’ points from each spectrogram to form your final synthesized time-frequency representation. Summarize your observations on how the choice of the ‘best’ scale changes across time and frequency.