I have been working on the following problem for several days:
Given L(x):=(x^2-1)* diff( T(n,x) ) / ( x- c) where \(T(n,x)\) is the nth order Chebyshev
polynomial and c is constant.
I try to evaluate diff( L(x),x) and diff(L(x),x,x,x,x) at certain x,including
x=c.
How can I get simplifed results? Has anyone written packages that rewrite expressions
such as diff(T(n,x),x),etc in terms of { T(i,x),i=0...n-1 )?
Wanjun Mi asks how to differentiate Chebyshev polynomials. The code to do this is on p. 68 of my book "Essential Maple". I’ve just re-typed this here (it’s short enough that that is easier than finding my package, so watch out for errors):
`diff/T` := proc(k, expr, x) local j, ans; if not type(k, 'integer') then 'diff(T(k,expr), x)' elif k=0 then 0 elif k<0 then diff(T(-k,expr),x) elif k=1 then T(0,expr)*diff(expr,x) else ans := -k*((-1)^(k-1)+1)/2*T(0,expr); for j from 0 to trunc((k-1)/2) do ans := ans + 2*k*T(k-1-2*j, expr); od; ans*diff(expr, x) fi end:
Now diff(T(5,x), x) returns the derivative in terms of T(0,x), T(2,x), and
T(4,x).
More on Chebyshev polynomials coming up in Symbolic Recipes Vol II (as soon as Vol I is finished).
I don’t have packages for your problem, but may be some hints are useful for a novice:
1) Start with command: with(orthopoly); to have a bulk or orthonormal polynomials available. 2) Then avoid the name L in your session (it's reserved for another polynomial) 3) Never define a function like this: L(x):=(x^2-1)*... Use instead: f:=x->(x^2-1)* ... ; Then use f(x), f(2.5), ... 4) To simplify relations which contain diff( L(x), x, x, ... ) you have either to make n in T(n,x) explicit e.g. 5 or 10 or so or (you don't have to write a package for that) define an equation like (I hope I did not overlook that this relation is already present somewhere inside Maple Rel. 3): eq:=diff(T(n,x),x)= (... * T(n,x) + ... * T(n-1,x))/(1-x^2); (see the handbook of Abramowics/Stegun for the details) and then use '' substitute '', e.g. subs(subs(n=m,eq),diff(T(m,x),x)); (do you understand the meaning of the "double" substitution?).