1
$\begingroup$

I am looking for an kind of polynomial such as Legendre properties that polynomial sequence of orthogonal polynomials such as bellow image. Could you suggest to me one polynomial? Is B-spline correct? Thanks

enter image description here

$\endgroup$
  • $\begingroup$ What does "similar properties" mean? $\endgroup$ – Martin R Jul 2 '15 at 18:37
  • $\begingroup$ It is "polynomial sequence of orthogonal polynomials" $\endgroup$ – Jame Jul 2 '15 at 18:38
  • 1
    $\begingroup$ Does this help? en.wikipedia.org/wiki/Orthogonal_polynomials $\endgroup$ – eigenchris Jul 2 '15 at 18:39
3
$\begingroup$

We can actually invent infinitely many! From what i gather you are interested in orthogonal polynomials, meaning polynomials $P_i$ such that there exists $a,b$ such that

$$ \int_a^bP_i(x)P_j(x) dx = 0 \forall i \ne j$$

To generate such polynomial sequences we can start with some initial Polynomial $P_0$ it can be any polynomial of your choice (An example is $x^2+3x$) and then pick a suitable a,b (lets say 0,1).

Then to get the next polynomial in teh sequence note it will have degree one greater than our current one. So $P_0 = x^2 +3x$ $P_1$ that we are trying to calculate will have the form $a_0 + a_1x + a_2x^2 + a_3x^3 $. Then we can write the integration law:

$$ \int_a^b P_0 P_1 dx = 0 $$

Which for us is

$$ \int_0^1 (x^2 + 2x)(a_0 + a_1x + a_2x^2 + a_3x^3) dx = 0 $$

And this yields a single equation in 4 unknowns after integration

$$ \frac{1}{3}a_0 + \frac{1}{4}a_1 + \frac{1}{5}a_2 + \frac{1}{6}a_3 + \frac{2}{2}a_0 + \frac{2}{3}a_1 + \frac{2}{4}a_2 + \frac{2}{5}a_3 = 0 $$

That can be cleaned up to

$$ \frac{4}{3}a_0 + \frac{11}{12}a_1 + \frac{14}{20}a_2 + \frac{17}{30}a_3 = 0 $$

Any four numbers $a_0 ... a_3$ you pick to satisfy that generates an orthogonal polynomial to the $P_0$ we selected.

Like so $P_2$ can be found by putting conditions on $P_0$ and $P_1$ such that it is orthogonal to both. Like this we can repeatedly generate one orthogonal polynomial after another to make whatever orthogonal sequence we choose.

$\endgroup$
  • $\begingroup$ Thank you. Do you hear about Haar basis function. Could you based on your explanation and find first few Haar basis function as above image in my question?en.wikipedia.org/wiki/Haar_wavelet $\endgroup$ – Jame Jul 2 '15 at 19:14
  • $\begingroup$ Your link gives there general formula right there. Anyone should be able to find them. The proof of orthonormality is also given $\endgroup$ – frogeyedpeas Jul 2 '15 at 19:16
  • $\begingroup$ The tools above are overkill if that's your intention. $\endgroup$ – frogeyedpeas Jul 2 '15 at 19:16
  • $\begingroup$ It is does not look like your tool. What is P(x) in the Haar case. Could you put some example in your answer? $\endgroup$ – Jame Jul 2 '15 at 19:24
  • $\begingroup$ Looking a bit more into it. Those aren't even polynomials. So the technique above doesn't apply. They are much more artificial (perhaps specially invented for some real life application). Do you want a proof of their orthogonality? Etc... They are orthogonal but they aren't polynomials and so asking how to construct them from polynomials won't make too much sense $\endgroup$ – frogeyedpeas Jul 2 '15 at 19:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.