Take the 2-minute tour ×
Mathematica Stack Exchange is a question and answer site for users of Mathematica. It's 100% free, no registration required.

I have two 5th-order (6th-order?) Chebyshev approximations:

f1[x_] = Sum[a[n]*ChebyshevT[n,x],{n,0,5}]
f2[x_] = Sum[b[n]*ChebyshevT[n,x],{n,0,5}]

and f1[1] == f2[-1]. I now "combine" these functions so f1's domain is mapped to [-1,0] and f2's domain is mapped to [0,1]. The Chebyshev approximation of this combination is:

coeff[n_] := coeff[n] = 2/Pi*(
 Integrate[f1[x*2+1]/Sqrt[1-x^2]*ChebyshevT[n,x],{x,-1,0}] +
 Integrate[f2[x*2-1]/Sqrt[1-x^2]*ChebyshevT[n,x],{x,0,1}]
)

(coeff[0] actually should be divided by 2, but that doesn't affect my question)

Mathematica won't give me a general formula here (getting rid of "coeff[n_] :=" fails), but I believe there is one. What is it?

I tried using "f1[1] == f2[-1]" to simplify, but it doesn't help.

I calculated coeff[0] through coeff[12]. The multiplier for each of the a[i] is below (the b[i] are easily calculated from the a[i]).

I noticed the multiplier is of the form (a+b*Pi)/c (for a,b,c all integers) for the first few entries in each sequence (ie, i < j in some sense), and then become a/Pi for integer a.

The first sequence (for a[0]) has alternating 0's and may be a special case (and is also easy to solve).

I tried using oeis.org on the numerators and denominators to no avail.

Expand[Table[D[coeff[i],a[j]],{j,0,5},{i,0,12}]] // TeXForm

$ \left( \begin{array}{ccccccccccccc} 1 & -\frac{2}{\pi } & 0 & \frac{2}{3 \pi } & 0 & -\frac{2}{5 \pi } & 0 & \frac{2}{7 \pi } & 0 & -\frac{2}{9 \pi } & 0 & \frac{2}{11 \pi } & 0 \\ 1-\frac{4}{\pi } & 1-\frac{2}{\pi } & -\frac{4}{3 \pi } & \frac{2}{3 \pi } \ & \frac{4}{15 \pi } & -\frac{2}{5 \pi } & -\frac{4}{35 \pi } & \frac{2}{7 \pi } & \frac{4}{63 \pi } & -\frac{2}{9 \pi } & -\frac{4}{99 \pi } & \frac{2}{11 \pi } & \frac{4}{143 \pi } \\ 5-\frac{16}{\pi } & 4-\frac{38}{3 \pi } & 2-\frac{16}{3 \pi } & -\frac{22}{15 \pi } & \frac{16}{15 \pi } & -\frac{2}{21 \pi } & -\frac{16}{35 \pi } & \frac{58}{315 \pi } & \frac{16}{63 \pi } & -\frac{122}{693 \pi } & -\frac{16}{99 \pi } & \frac{202}{1287 \pi } & \frac{16}{143 \pi } \\ 25-\frac{236}{3 \pi } & 21-\frac{66}{\pi } & 12-\frac{188}{5 \pi } & 4-\frac{182}{15 \pi } & -\frac{44}{35 \pi } & \frac{10}{7 \pi } & -\frac{28}{45 \pi } & -\frac{34}{105 \pi } & \frac{76}{165 \pi } & \frac{38}{693 \pi } & -\frac{964}{3003 \pi } & \frac{14}{429 \pi } & \frac{1492}{6435 \pi } \\ 129-\frac{1216}{3 \pi } & 112-\frac{5278}{15 \pi } & 72-\frac{3392}{15 \pi \ } & 32-\frac{3518}{35 \pi } & 8-\frac{2624}{105 \pi } & -\frac{254}{315 \pi } & \frac{64}{45 \pi } & -\frac{1334}{1155 \pi } & \frac{64}{495 \pi } & \frac{8482}{15015 \pi } & -\frac{2752}{9009 \pi } & -\frac{4054}{15015 \p\ i } & \frac{1856}{6435 \pi } \\ 681-\frac{32092}{15 \pi } & 605-\frac{5702}{3 \pi } & 420-\frac{9236}{7 \pi } & 220-\frac{4838}{7 \pi } & 80-\frac{15836}{63 \pi } & 16-\frac{2258}{4\ 5 \pi } & -\frac{316}{693 \pi } & \frac{86}{77 \pi } & -\frac{1132}{819 \pi } & \frac{310}{429 \pi } & \frac{1228}{4095 \pi } & -\frac{1822}{3003 \pi } & \frac{788}{13923 \pi } \\ \end{array} \right) $

share|improve this question
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.