8
$\begingroup$

Examining the indefinite integral formulations of inverse trig functions I notice some things

$$\arcsin(x)=\int_0^x \frac{1}{\sqrt{1-z^2}}dz$$

$$\arccos(x)=\int_x^1 \frac{1}{\sqrt{1-z^2}}dz$$

We can say that these functions "split" the range of integration $[0..1]$ at $x$.

Is there a visualization which expresses this relationship graphically?

I mean, other than just drawing the graph, is there a visualization which meaningfully shows the relationship, in an insightful and intuitive way?

Similarly,

$$\arctan(x)=\int_0^x\frac{1}{z^2+1}dz$$

$$\mathrm{arccot}(x)=\int_x^\infty\frac{1}{z^2+1}dz$$

Again, these "split" the range $[0..\infty]$ at $x$. Is there a digram for this?

Similarly,

$$\mathrm{arcsec}(x)=\int_1^x\frac{1}{z\sqrt{z^2-1}}dz$$

$$\mathrm{arccsc}(x)=\int_x^\infty\frac{1}{z\sqrt{z^2-1}}dz$$

These split the range $[1..\infty]$ at $x$. Is there a diagram for this?

$\endgroup$
1
$\begingroup$

Suppose the length of the hypotenuse of a right triangle is $1$ and the length of one leg is $x.$ Then the angle opposite the side of length $x$ is $\arcsin x,$ and the angle between that side and the hypotenuse is $\arccos x.$ Since the sum of the two small angles of a right triangle is $\pi/2,$ we have the identity $\arcsin x + \arccos x = \pi/2.$ The same kind of argument shows that $\arctan x + \operatorname{arccot} x = \pi/2$ and $\operatorname{arcsec} x + \operatorname{arccsc} x = \pi/2.$

$\endgroup$
0
$\begingroup$

Here is a "visual" explanation for $ \arccos $ and $ \arcsin $.

Use $ \gamma(t) = (1-t,\sqrt{1-(1-t)^2}) $ where $ 0 \le t \le 1 $ to parameterize the upper right part of the unit circle. (You can find this parameterization by a simple application of Pythagoras Theorem.) Then fix $ x = \cos(\alpha) $ where $ 0 \le \alpha \le \frac{\pi}{2} $. Now calculate $ \alpha $ in Radian by

$$ \alpha(x) = L(\gamma | _ {[0,1-x]}) = \int_{0}^{1-x} {|\gamma'(s)|ds} = \int_{0}^{1-x} {\sqrt{1 + \left(\frac{-(1-s)}{\sqrt{1-(1-s)^2}}\right)^2}} ds = \int_{0}^{1-x} {\sqrt{\frac{1-(1-s)^2+(1-s)²}{1-(1-s)^2}}} ds = \int_{0}^{1-x} {\sqrt{\frac{1}{1-(1-s)^2}} ds} = \int_{1}^{x} {\frac{-1}{\sqrt{1-s^2}} ds} = \int_{x}^{1} {\frac{1}{\sqrt{1-s^2}} ds} $$

But the function which assign's to $ x $ the value $ \alpha $ such that $ x = \cos(\alpha) $ is by definition $ \arccos $. Hence we have derived

$$ \arccos(x) = \int_{x}^{1} {\frac{1}{\sqrt{1-s^2}} ds} $$

So the integral is interpreted as measuring the length of the unitcircle from $ (1,0) $ to $ (\cos(\alpha),\sin(\alpha)) $. This is exactly the angle $ \alpha $ in radian.

Now we use $ \sin(\alpha) = \cos(\beta) $ where $ \beta = \frac{\pi}{2} - \alpha $. The angle $ \beta $ can be found in the triangle spanned by $ (0,0) $, $ (\cos(\alpha),\sin(\alpha)) $ and $ (0,1) $. As above we calculate the angle in radian by

$$ \beta(x) = L(\gamma|_{[1-x,1]}) = \cdots = \int_{1-x}^{1} { \sqrt{\frac{1}{1-(1-s)^2}} ds } = \int_{x}^{0} { \frac{-1}{\sqrt{1-s^2}} ds } = \int_{0}^{x} { \frac{1}{\sqrt{1-s^2}} ds } $$

You can interpret this integral as measuring the length of the unitcircle from $ (\cos(\alpha),\sin(\alpha)) $ to $ (0,1) $. This is exactly the angle needed to complete $ \alpha $ to $ \frac{\pi}{2} $.

So the reason of $ [0,1] $ beeing split at x is the relation $ \sin(\alpha) = \cos(\beta) $ where $ \beta = \frac{\pi}{2} - \alpha $ in combination with the analytic definition of an angle.

Note that we have also "proofed" the following result:

$$ \frac{\pi}{2} = \alpha(x) + \beta(x) = \int_0^1{\frac{1}{\sqrt{1-t^2}}dt} $$

$\endgroup$
0
$\begingroup$

For me, the visual has always been the derivation process. $$I=\int_0^x\frac{dt}{(1-t^2)^{1/2}}$$ Suppose $t=\sin u$, then $dt=\cos u\,du$ $$I=\int_{t=0}^{t=x}\frac{\cos u}{(1-\sin^2u)^{1/2}}du$$ $$I=\int_{t=0}^{t=x}\frac{\cos u}{(\cos^2u)^{1/2}}du$$ $$I=\int_{t=0}^{t=x}\frac{\cos u}{\cos u}du$$ $$I=\int_{t=0}^{t=x} du$$ $$I=u|_{t=0}^{t=x}$$ $$I=(\arcsin t)|_{0}^{x}$$ $$I=\arcsin x$$ I don't know man. Something about watching those integrals go from nasty to beautiful just really did the trick for me.

$\endgroup$
  • $\begingroup$ The derivation is nice but your integrals have no ranges. Where in your derivation is the $\int_0^x$ part? $\endgroup$ – spraff Oct 16 '18 at 21:26
  • $\begingroup$ @Spraff How about now? $\endgroup$ – clathratus Oct 16 '18 at 22:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.