# Why does this “incorrect” Chebyshev function approximation work better than the correct one?

I recently had the need to approximate this function

$$f\left(x\right)=\begin{cases} \log\left(\frac{\pi+2\arcsin\left(x\right)}{\pi}\right), & x<0\\ -\log\left(2-\frac{\pi+2\arcsin\left(x\right)}{\pi}\right), & x\ge0 \end{cases}$$

which looks like this

by a Chebyshev series and found something very strange in a popular numerical way of computing the Chebyshev coefficients. There is nothing special about this particular function except that the series converges very slowly due to the function going to ±∞ at ±1. Indeed it is not hard to find functions which have slow Chebyshev convergence—this one also converges slowly due to the discontinuous first derivative at zero:

$$g\left(x\right)=\begin{cases} \frac{1}{(10\left|x\right|+1/2)}-\frac{2}{21}, & \left|x\right|\le1\\ 0, & \text{otherwise} \end{cases}$$

which looks like this

Looking for a way to evaluate series like this that do not involve quadrature lead me to an algorithm in Numerical Recipes, chebft, (scroll to the third page) http://www.aip.de/groups/soe/local/numres/bookcpdf/c5-8.pdf or http://rosettacode.org/wiki/Chebyshev_coefficients. This appears to be essentially identical to the GNU Scientific Library function gsl_cheb_init http://git.savannah.gnu.org/cgit/gsl.git/tree/cheb/init.c.

The Numerical Recipes book—but not the GSL docs—are very clear that this routine works only if the series converges quickly; if that is the case, one computes the series to a "high" order $N$ which is essentially exact in its approximation and then truncates that series to the first $M$ coefficients. The assumption is that the truncated terms will be neglible.

This is great except for these slowly-converging series which can require hundreds or thousands of terms to converge to say five significant digits of the true coefficients as computed by e.g. an exact closed-form expression for $f(x)$ which happens to involve the sine integral.

FOLLOWING IS THE CRUX OF THIS NOTE.

(1) Run one of these routines, chebft or presumably gsl_cheb_init, using a "high" order $N_1$, then examine the first $M < N_1$ coefficients. Then run the routine again with a significantly larger $N_2$, and again look at the first $M$ coefficients. They are different, in some instances by quite a bit, especially in those of order approaching $M$.

(2) The function approximation using the $M$ coefficients from the $N_1$ case is better than the function approximation using $M$ coefficients from the N2 case, even though the $M$ coefficients from the $N_1$ case are rather different than the true Chebyshev coefficients, and more different from the true coefficients than the first M coefficients from the $N_2$ case, which $M$ coefficients approach the correct coefficients as $N_2$ gets large.

(3) Implied in (2) is this: Using $M = N_1$ coefficients gives a much better approximation than using the first $M$ coefficients of the actual Chebyshev series if $M$ is "small!"

Here is an example for $f(x)$ above. The even-order coefficients are zero. First, the coefficients are listed then plotted as 20*log10(abs(coefficient)), that is, decibels or dB, then the difference between $f(x)$ and its approximation using the listed coefficients is plotted.

$N = 19, M = 19$

Order Coefficient
1     1.67593635160246E+00
3     6.12793162859104E-01
5     3.25435579886791E-01
7     2.14992666832514E-01
9     1.47380048482564E-01
11    1.05479939128339E-01
13    7.35400746724925E-02
15    4.92768119919009E-02
17    2.80197009109371E-02
19    9.31058725663306E-03 FIG 3 FIG 4

$N = 199, M = 19$

Order Coefficient
1     1.73837707065426E+00
3     6.75682619059153E-01
5     3.89256226291883E-01
7     2.80219340502524E-01
9     2.14597792673852E-01
11    1.75260255015454E-01
13    1.46673501808019E-01
15    1.26540465300740E-01
17    1.10595097004330E-01
19    9.83860349363951E-02 FIG 5 FIG 6

The error plots are clipped at $±0.5$ so there is the possibility that I have chopped off something important, especially since I stopped the plot at $1.0e-12$ inside of $±1$ to prevent blowing things up. The non-clipped plot for $N_1=19$ shows the error there to be about $±24$ with the $N_2 = 199$ version being slightly smaller, about $±23$; so there is the slight possibility that the vast majority of the error has been hidden in this extremely small region near $±1$. However, $g(x)$ generates very similar (in nature) coefficient and error plots in showing this phenomenom but with slow convergence near zero and it has no infinities that I am "hiding".

Here are the first 19 coefficients computed from the sine integral that Mathematica discovered for me—they should be the exact Chebyshev coefficients for $f(x)$:

Exact Results

Order Coefficient
1     1.745308598921210
3     0.682614597669615
5     0.396189107568285
7     0.287153573392024
9     0.221533832064617
11    0.182198549038367
13    0.153614508509876
15    0.133484632575528
17    0.117542886658270
19    0.105337895207024


Please study these three coefficient lists carefully.

The rapid roll-off of the coefficients in the right-hand 10 percent or so of the coefficient plots is present no matter how many coefficients are computed, e.g., $20,000$ and more. It's a little hard to see in the 19-coefficient plot but it is there, in the last non-zero data point.

QUESTION

Why does this happen? Why do these incorrect series coefficients for rather severely wrong Chebyshev approximations do such a better job at approximating the underlying functions, at least in these case of slow Chebyshev convergence? Does the taper in the $10$ percent of the RHS of the plots amount to a windowing effect as is well known in time series and signal processing? I have not tested this on any "fast-converging" cases.

Does anyone have an explanation for the unexpected good performance of this "new" series inadvertently provided by these canned routines? Is the $M=N$ usage optimal in any sense?

I apologize for the length of this post especially if I have treaded upon some protocol or courtesy—I'm new to the world of Stackexchange.

• I don't find any of this unexpected -- perhaps you should explain a bit more why you find it unexpected. – joriki Aug 18 '15 at 6:37
• I believe that I have been clear in my explanation of the problem. If there is any point of confusion, please let me know how I can help. Otherwise, please enlighten me with your insight. – Jerry Aug 20 '15 at 10:43
• The situation is asymmetric. If someone finds something unexpected, she or he is usually in a position to point out what they would have expected, and why. If someone doesn't find something unexpected, there's often not much to say about why -- there just isn't anything unexpected to point to. If you believe you've explained why you find the results unexpected, perhaps you could point to the specific point in the text where you did? – joriki Aug 20 '15 at 10:52
• When one misuses an algorithm by applying it to a situation for which it is not designed, a poor outcome usually results. In the present case, not only do poor results not accrue but results that are far better are obtained as compared to even the prescribed usage. – Jerry Aug 25 '15 at 4:16
• The salient point of my post is to invite an explanation for this behavior. I have offered a tentative line of explanation by suggesting that there might be a windowing effect analogous to that which is commonly used in truncating series of trigonometric polynomials to mitigate against the Gibbs phenomenon, perhaps spreading the error out as is typical of Chebyshev-type errors rather than localizing it around a discontinuity or other non-smooth area as is typical of truncated Fourier series. – Jerry Aug 25 '15 at 4:16