Chebyshev polynomial approximation

Click For Summary
The discussion focuses on finding the quadratic least squares Chebyshev polynomial approximation for the function g(z) = (15π/8)(3-z^2)√(4-z^2) over the interval z ∈ [-2,2]. The transformation to the variable t is necessary, with the mapping t = z - 1 leading to t ∈ [-1, 1]. Participants clarify that the method of least squares involves minimizing the norm squared of the difference between the function and a linear combination of Chebyshev polynomials, rather than simply projecting the function onto these polynomials. There is confusion regarding the appropriate mapping for z to ensure that t remains within the correct bounds. Ultimately, the correct approach requires reevaluating the transformation to maintain the integrity of the Chebyshev polynomial approximation.
says
Messages
585
Reaction score
12

Homework Statement


Find the quadratic least squares Chebyshev polynomial approximation of:
g(z) = 15π/8 (3-z^2)√(4-z^2) on z ∈ [-2,2]

Homework Equations


ϕ2(t) = c0/2 T0(t) +c1T1(t)+c2T2(t)

T0(t)=1
T1(t)=t
T2(t)=2t2-1

Cj = 2/π ∫ f(t) Tj(t) / (√(1-t2) dt

where the bounds for the integration are -1 to 1.

The Attempt at a Solution


The first thing I know I need to do is transform into t [-1,1], so:

t = z-1 ⇒ z = t+1

at z = 0, t = -1
at z = 2, t = +1
t ∈ [-1,1]

g(z) = (15π/8) (3-z2)√(4-z2)

g(t) = (15π/8) (3-(t+1)2)√(4-(t+1)2)

I'm a bit confused at this point. Hoping someone can be of help. How I've proceeded is below

2/π ∫ g(t) Tj(t) / (√(1-t2)) dt

bounds of integration are again from -1 to 1

c0 = 2/π ∫ (15π/8) (3-(t+1)2)√(4-(t+1)2) (1) / (√(1-t2)) dt

c1 = 2/π ∫ (15π/8) (3-(t+1)2)√(4-(t+1)2) (t) / (√(1-t2)) dt

c2 = 2/π ∫ (15π/8) (3-(t+1)2)√(4-(t+1)2) (2t2-1) / (√(1-t2)) dt

Once i evaluate c0, c1, c2 I can then put them in the equation ϕ2(t) and then sub the z value back in and get the Chebyshev polynomial approximation. At least I think that's how it's done.
 
Physics news on Phys.org
says said:
t = z-1 ⇒ z = t+1

at z = 0, t = -1
at z = 2, t = +1
t ∈ [-1,1]
Your problem statement states that z ∈ [-2,2], not [0,2].

Also, what you have done is to project the function on the Chebyshev polynomials, which is not really the method of least squares (although, of course, the result is the same). The method of least squares would be to consider the norm squared of ##f-T##, where ##T## is some linear combination of the Chebyshev polynomials and then minimise this with respect to the coefficients of the linear combination.
 
so at z=-2, t=-3
z=2, t=1

that's if:
t=z-1, z=t+1
t ∈ [-3,1]

I don't see how this part is relevant to the problem as i never use it. That is unless t is supposed to be ∈ [-1,1] and I'm supposed to find a different euation for z that satisfies t ∈ [-1,1]?
 
The Chebyshev polynomials are relevant for the interval [-1,1] so you must find a different mapping from z to t.
 

Similar threads

  • · Replies 11 ·
Replies
11
Views
2K
Replies
39
Views
6K
  • · Replies 7 ·
Replies
7
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 2 ·
Replies
2
Views
1K
Replies
40
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K