11.08.2013 Views

Untitled - Cdm.unimo.it

Untitled - Cdm.unimo.it

Untitled - Cdm.unimo.it

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

70 Polynomial Approximation of Differential Equations<br />

Thus, <strong>it</strong> is natural to define, for n ≥ 1, the operator Aw,n = Iw,n − Πw,n−1. The<br />

polynomial in Pn−1<br />

(4.2.1) Aw,nf = Iw,nf − Πw,n−1f, n ≥ 1, f ∈ C 0 ([−1,1]),<br />

takes the name of aliasing error.<br />

Depending on the smoothness of f, this error tends to zero when n → +∞. The proof<br />

of this fact and the characterization of the aliasing error is given in section 6.6.<br />

Another type of aliasing error is obtained by setting<br />

(4.2.2) Ãw,nf = Ĩw,nf − Πw,nf, f ∈ C 0 ([−1,1]),<br />

where n ≥ 1. This is also studied later on.<br />

4.3 Fast Fourier transform<br />

Let us examine what happens by applying the discrete Fourier transform in the special<br />

case of the Chebyshev basis. By virtue of (1.5.6), (3.1.4), (3.4.6) and (2.2.12), the entries<br />

of Kn in (4.1.5) take the form<br />

(4.3.1) kij =<br />

(−1) i<br />

n<br />

<br />

1 if i = 0, 0 ≤ j ≤ n − 1<br />

i(2j + 1)π<br />

cos ×<br />

2n<br />

2 if 1 ≤ i ≤ n − 1, 0 ≤ j ≤ n − 1<br />

For the same reason, due to (3.1.11), (3.5.7) and (3.8.5), the entries of ˜ Kn are<br />

(4.3.2) ˜ kij =<br />

(−1) i<br />

n<br />

<br />

ijπ<br />

cos ×<br />

n<br />

⎧<br />

1<br />

2 if i = 0, j = 0 or j = n<br />

⎪⎨<br />

1<br />

1<br />

if i = 0, 1 ≤ j ≤ n − 1<br />

if 1 ≤ i ≤ n − 1, j = 0 or j = n<br />

2<br />

1 ⎪⎩ 2<br />

1<br />

if 1 ≤ i ≤ n − 1, 1 ≤ j ≤ n − 1<br />

if i = n, j = 0 or j = n<br />

if i = n, 1 ≤ j ≤ n − 1<br />

.<br />

.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!