Using the metric Euclidean tensor for polar coordinates, calculate the arc length for the curve:
$x^1=2a\cos t$
$x^2=\sin t$ where $0\leq t\leq2\pi$
If I use the formula:
$L=\int_{a} ^{b} \sqrt{\left(\cfrac{dx^1} {dt} \right)^2+\left(\cfrac{dx^2} {dt}\right) ^2}=\sqrt{\left| \delta_{ij} \cfrac{dx^1} {dt}\cfrac{dx^2} {dt}\right|} $
I get an elliptic integral, can anyone help me to find the arc lenght? and How could you interpret the result geometrically?
Thanks in advanced.
As you probably noticed, the result is $$L=2 \left(E\left(-4 a^2\right)+\sqrt{4 a^2+1} E\left(\frac{4 a^2}{4 a^2+1}\right)\right)$$ What you can do is to approximate $L$ using $[2n,2n]$ Padé approximants built around $a=0$.
For example $$P_2=2\pi\, \frac {1+\frac{159 }{44}a^2+\frac{453 }{176}a^4 } { 1+\frac{115 }{44}a^2+\frac{125 }{176}a^4}$$ $$P_3=2\pi\, \frac {1+\frac{39575 }{7116}a^2+\frac{20621}{2372}a^4+\frac{129235 }{37952}a^6} {1+\frac{32459 }{7116}a^2+\frac{34741 }{7116}a^4+\frac{79037 }{113856}a^6 } $$ $$P_4=2\pi\, \frac {1+\frac{133542997 }{17725732}a^2+\frac{1325913585 }{70902928}a^4+\frac{1210596065 }{70902928}a^6+\frac{4808786003 }{1134446848}a^8 } {1+\frac{115817265 }{17725732}a^2+\frac{915821721 }{70902928}a^4+\frac{553597479 }{70902928}a^6+\frac{777708891 }{1134446848}a^8 }$$
The table shows how good (or bad) are these approximations $$\left( \begin{array}{ccccc} a & P_2 & P_3 & P_4 & \text{exact} \\ 0.00 & 6.283185307 & 6.283185307 & 6.283185307 & 6.283185307 \\ 0.25 & 6.659164725 & 6.659167214 & 6.659167222 & 6.659167222 \\ 0.50 & 7.639438651 & 7.640368153 & 7.640394783 & 7.640395578 \\ 0.75 & 8.969481052 & 8.985328348 & 8.986601081 & 8.986713343 \\ 1.00 & 10.44445389 & 10.52687833 & 10.53873159 & 10.54073433 \\ 1.25 & 11.91631116 & 12.15523595 & 12.20520464 & 12.21838971 \\ 1.50 & 13.29114874 & 13.78900721 & 13.92449946 & 13.97441783 \\ 1.75 & 14.52204955 & 15.36792068 & 15.64828398 & 15.78301361 \\ 2.00 & 15.59531652 & 16.85165250 & 17.33739745 & 17.62855113 \end{array} \right)$$
Edit
To better see the gain, consider the norm $$\Phi_n=\int^2_0 \big[L-P_n\big]^2\,da$$ which leads to $$\Phi_2=0.983 \qquad \Phi_3=0.114 \qquad \Phi_4=0.013$$ that is to say almost a factor of $10$ when $n \to n+1$.