1
$\begingroup$

I need to find an alternative form of $c\cdot\cos(a+b)$. The goal is to get $c$ into the $\cos$ (e.g. $\cos(a\cdot c+b\cdot c)$). Can anyone help me with this problem or knows where i can look further?

Thanks in advance thorb3n

Edit1: Sorry, i thínk i wrote it the wrong way. i mean $c\cdot \cos(a+b)$

$\endgroup$
  • 2
    $\begingroup$ $\cos(ab)$ is not related to $\cos a$, $\sin a$, $\sin b$ and $\cos b$ in a sensible way. $\endgroup$ – user228113 Oct 13 '17 at 8:19
  • 1
    $\begingroup$ Welcome to MSE. Please use MathJax. $\endgroup$ – José Carlos Santos Oct 13 '17 at 8:26
  • $\begingroup$ Why exactly do you want such a form? (I just want more background), and what have you tried? $\endgroup$ – mdave16 Oct 13 '17 at 10:48
1
$\begingroup$

If I understand your question correctly, you want an equation of the form $c \cdot \cos(a+b) = \cos(f(a,b,c))$.

However, no such real-valued function $f$ exists. If such an equation above does exist, then it must also work for my favourite numbers $a=0$, $b=0$ and $c=10$. However, the left side of the equation is $10$, which is outside of the range of $\cos$, so no such function could exist.

$\endgroup$
  • 1
    $\begingroup$ Good answer. Although, no such real valued function could exist, since $\cos(\cdot)$ can be unbounded when it's allowed to take complex values :) $\endgroup$ – Jam Oct 13 '17 at 10:58
  • $\begingroup$ @Jam True, I overlooked it thinking that OP might not be thinking about them. Great answer btw +1 $\endgroup$ – mdave16 Oct 13 '17 at 11:14
1
$\begingroup$

Starting with $c\cdot\cos(a+b)$, for most purposes I can think of, we wouldn't need to change how it's expressed. It's quite simple the way its currently expressed, and neatly tells you the range of output values ($-c$ to $+c$) and the roots of the expression ($a+b=n\pi$).

But if you do want to express it with everything inside the $\cos(\cdot)$, you could try setting the expression as $c\cdot\cos(a+b)=\cos(x)$, and finding $x=\cos^{-1}\left[c\cdot\cos(a+b)\right]$.

We can find approximations for $\cos^{-1}(z)$, such as $\cos^{-1}(z)\approx \frac\pi2-z-\frac{z^3}{6}$, for $-1<z<1$. We can then set $z=\cos{-1}\left[c\cdot\cos(a+b)\right]$.

This implies that: $$c\cdot\cos(a+b)\approx \cos\bigg(\frac{\pi}{2}-c\cdot\cos(a+b)-\frac{c^3}{6}\cos^3(a+b)\bigg)$$

However, this approximation only works for $-1<c<1$. But, as mdave16 points out, $\cos(x)$ is bounded as $-1<\cos(x)<1$, so not all values of $c$ are possible anyway (unless $x$ can be a complex number).

enter image description here

The figure above shows what the approximation looks like in the worst case, for $a=-1.8,c=1$, as we vary $b$. It turns out that $a$ and $b$ don't affect the maximum error but even when $c$ is close to $1$, the absolute error is at most $0.08$.

$\endgroup$
  • $\begingroup$ We can use the Taylor series for $\cos(x)$, and some tricks with a sawtooth wave to add periodicity, and a square wave to get the sign right, to show that $\cos(x)\approx \frac12(-1)^{\lfloor \frac{x}{\pi}+\frac12\rfloor}\left(2-(x-\pi(1+\lfloor \frac{x}\pi-\frac12\rfloor))^2\right)$, with maximum error of $0.23$. Then $c\cdot \cos(a+b)\approx \cos\left(\frac\pi2-\frac{c}2(-1)^{\lfloor\frac{a+b}{\pi}+\frac12\rfloor}\left(2-(a+b-\pi(1+\lfloor \frac{a+b}{\pi}-\frac12\rfloor))^2\right)\right)$, with maximum error of $0.23$, when $c=1$. The function $\lfloor\cdot\rfloor$ is the floor function. $\endgroup$ – Jam Oct 13 '17 at 13:41

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.