[Math] Calculating the length of a curve for $ y=\ln (1-x^2) $

calculusintegration

In calculating the length of $ y=\ln (1-x^2) $ on the interval $ [0, {3\over4}]$, I found that for whatever reason, I end up with invalid domains in the indefinite integral using the given limits of integration.

Using the formula $ \int^b_a \sqrt{1+f'(x)^2} $, I was able to derive

$$ {dy\over dx}=f'(x)=-{2x\over1-x^2} $$
$$ \int^{3\over4}_0 \sqrt{1+(-{2x\over1-x^2})^2}\ dx = \int^{3\over4}_0 \sqrt{1+{4x^2\over1-2x^2+x^4}}\ dx = \int^{3\over4}_0 \sqrt{{x^4+2x^2+1}\over{x^4-2x^2+1}}\ dx $$
$$ = \int^{3\over4}_0 {{x^2+1}\over{x^2-1}}\ dx $$

as the integrand in question. Then by the rule of the sums of integrals:

$$ \int^{3\over4}_0 {{x^2}\ dx\over{x^2-1}} + \int^{3\over4}_0 {{dx}\over{x^2-1}}$$

The clearest line of attack was through the following trigonometric substitution:

$$ x=\sec \theta $$
$$ dx=\sec \theta \tan \theta\ d\theta$$

$$ \int {{\sec^3 \theta \tan \theta\ d\theta}\over{\tan^2 \theta}} + \int {{\sec \theta \tan \theta}\over{\tan^2 \theta}} = \int \sec^2 \theta \csc \theta\ d\theta + \int \csc \theta\ d\theta$$

The former I integrated by parts:

**I.**$$ u=\csc \theta\ \quad dv=\sec^2 \theta\ d\theta$$
$$ du=-\csc \theta \cot \theta\ d\theta\ \quad v=\tan \theta$$

$$ \int \sec^2 \theta \csc \theta\ d\theta = \boxed{\csc \theta \tan \theta + \int \csc \theta\ d\theta}$$

And the latter:

**II.**$${\int \csc \theta\ d\theta= \boxed{-\ln \lvert{\csc \theta + \cot \theta}\rvert}}$$

Hence the complete indefinite integral:

**I.+II.**$$ \int \sec^2 \theta \csc \theta\ d\theta + \int \csc \theta\ d\theta = \boxed{\csc \theta \tan \theta -2 \ln \lvert{\csc \theta + \cot \theta}\rvert}$$

Transforming back to the original variable:

$$ x=\sec \theta \quad\quad \sqrt{x^2-1} = \tan \theta$$
$$ {1\over\sqrt{x^2-1}} = \cot \theta \quad\quad {x\over\sqrt{x^2-1}} = \csc \theta$$

$$ {\csc \theta \tan \theta -2 \ln \lvert{\csc \theta + \cot \theta}\rvert} = \boxed{x – 2\ln \lvert{{x+1}\over{\sqrt{x^2-1}}}\rvert} $$

As one can easily see, for whatever reason the limits produce a negative term under the radical in the natural log. Any insight as to why this is the case would be much appreciated.

Best Answer

..........................................................................

The hitch is here :

enter image description here

Related Question