1

I was inspired to return to an old problem I came up with after seeing

this question.

This was the problem of finding the analogs of other trigonometric functions which would parametrize a certain graph $f(x,y)=1$ or $y=f(x)$ taking the polar curve angle $\theta=\theta_0$ and finding the length of the curve for one intersection of $f(x,y)+0$. These are also called the abcissa and ordinate of the intersection.

enter image description here

This just boils down to finding trigonometric functions which parametrize the curve f(x,y). This is easy, but the goal is have the distance from the intersection of the angle with the curve to the x,y-axis be the these functions $(x(\theta), y(\theta))$. Here is how to find such functions. For convention, let $x(\theta)=\cos_{f(x,y)=0}(\theta)=\cos_f(x),y(\theta)=\sin_{f(x,y)=0}(\theta)=\sin_f(x)$. Sorry for the notation, but it is intuitive. Here is how to find such functions easily:

$$f(x,y)=0,y=x\tan(\theta)\implies f(x,x\,\tan(\theta))=0\implies x=\cos_{f(x,y)=0}(\theta),y=\sin_{f(x,y)=0}(\theta)$$

There is not an easy inverse function for both arguments. However, when y is given explicitly you can solve for the alternate trigonometric functions which sometimes works through recursion.

$$\frac{f(x)}{x}=\tan(\theta)\implies x=f^{-1}(x\,tan(\theta))= f^{-1}(f^{-1}(f^{-1}(…x…\,\tan(\theta))\,\tan(\theta))\,\tan(\theta))=\cos_f(\theta)\implies y=\sin_f(\theta)=\tan(\theta) f^{-1}(f^{-1}(f^{-1}(…x…\tan(\theta))\tan(\theta))\tan(\theta))$$

Unless you use an inversion theorem, then you cannot always solve for the inverse of y=f(x), unless if there is a closed form inverse. For example if you modeled the trigonometric equations with y=x tan(θ) and f(x)=\cos(x), you would find that $x(\theta)=\cos_{\cos(x)(θ)}=\cos_f(x), y(\theta)=x(\theta)\tan(\theta)$ would be the parametrization for $\cos(x)$ for $(x(\theta),y(\theta))$. This uses the cardinal sine/ sinc function. Note $x_0$ here is just a parameter as a consequence of the nesting:

$$\frac{\sin(x)}{x}=\text{sinc}(x)=\tan(\theta)\implies \cos_f(x)=\sin^{-1}(\sin^{-1}(…x_0 \tan(\theta))\tan(\theta)…),\frac{\cos(x)}x=\tan(\theta)\implies \cos_f(x)=\cos^{-1}(\cos^{-1}(…x_0\tan(\theta))\tan(\theta)…)$$

Here are the results in this graph. Notice that lengths of the legs of the right triangle, like in the diagram, will give the derived modified cosine and sine functions. These new trigonometric functions are now based on sine and cosine graph instead of the circle.

Of course we can just use an inversion theorem, in the link above, to find the alternate trigonometric functions. Here is a formula that will not need to be used very much as we can usually find these alternate trigonometric functions. This is the general result for a function with explicit $y=f(x)$ with the Lagrange Inversion Theorem. This uses the Bell polynomials and the rising factorial

$$\frac {f(x)}x=\tan(\theta)\implies x(\theta)=\cos_f(x)\left(\text{Inverse of }\frac yx\right)[\tan(\theta)]=a+\sum_{n=1}^\infty\lim_{x\to a}\frac{d^{n-1}}{dx^{n-1}}\left[\left(\frac{x-a}{\frac{f(x)}{x}-\frac{f(a)}{a}}\right)^n\right]\frac{\left(\tan(\theta)-\frac{f(a)}{a}\right)^n}{n!}=\sum_{k=0}^\infty c^{-n}_1\sum_{K=1}^{k-1}(-1)^K k^{(K)}B_{k-1,K}\left(\frac{c_2}{2c_1}, \frac{c_3}{3c_1},…, \frac{c_{n-k+1}}{(n-k+1)c_1}\right),\frac{f(x)}x=\sum_{n=0}^\infty c_n\frac{x^n}{n!},n\ge 2$$

All I did was use linked Lagrange Inversion theorem formulas.

As a side note, I will leave you with the alternate trigonometric functions for $a|x|^n+b|x|^n=1$. Here is an interactive graph The analogue for these trigonometric functions is just $$\mathrm{x(\theta)=\pm\frac{1}{\sqrt[n]{a+b|tan(\theta)|^n}},y(\theta)= \pm\frac{tan(\theta)}{\sqrt[n]{a+b|tan(\theta)|^n}}}$$

Here is another special case if one takes $f(x)=ax^2+bxy+cy^2+dx+ey+f=0$ in this fixed graph. Notice how the formula for a=c=1,b=d=e=0,f=-1 is a special case with the unit circle with the usual x=cos(θ),y=sin(θ). These types of analogues for the trigonometric functions are:

$$x(\theta)=-\frac{\sqrt{{(e\tan(\theta)+d))^2-4f・(c\tan^2(\theta)+b\tan(\theta)+a)}}+e\tan(\theta)+d}{2(c\tan^2(\theta)+b\tan(\theta)+a)},y(\theta)=x(\theta)tan(\theta)$$

The secant and cosecant analogues are just $\frac{1}{x(\theta)}$ and $\frac{1}{y(\theta)}$. The tangent and cotangent analogues are almost always $\tan(θ)$ and $\cot(θ) $because $\mathrm{y(\theta)=x(\theta)tan(\theta)}$.

These graphs now have the analogue (co)secant graphs. Please see the graphs for examples of trigonometric function analogues. The question now is how to solve $$y=x\tan(\theta),f(x,y)=0\implies f(x,x\tan(\theta))=0$$ for $x$ in exact form, not necessarily closed form, using any method except approximation. Please correct me and give me feedback!

Arctic Char
  • 16,007
Тyma Gaidash
  • 12,081
  • First of all, do you assume $f(0,0)=0$ ? – Jean Marie Jul 31 '21 at 21:21
  • @JeanMarie No, please see this graph where $f(0,0)=a|0|^m+b|0|^n-1=-1\ne0$. f(x,y)=0 just means a function which is implicitly written for y. It is just a more general case of $y=f(x), y=f(x)+c \implies f(x,y)=y-f(x)=0,f(x,y)=y-f(x)-c=0 $ Tell me if this helps. – Тyma Gaidash Jul 31 '21 at 21:33
  • 1
    I don't understand why you use implicit equations $f(x,y)=0$ of your curves instead of considering their polar representation $r=r(\theta)$ which is much more adapted to your issue. – Jean Marie Aug 01 '21 at 07:55
  • @JeanMarie I have actually never thought of using polar coordinates. I could probably find a way to solve a few special cases and even help evaluate those hard to find in f(x,y)=0 form. A few functions are better in f(x,y)=0 form. You can go ahead and solve in polar form as an answer if you like for. Still,$f(r,\theta)=0,\theta=\theta_0$ may still be hard to solve, in general on cannot solve for x(θ)=sin(θ) for all f unless they have some inversion theorem, for example. Thanks – Тyma Gaidash Aug 01 '21 at 11:43
  • I mean something else than using polar coordinates. I mean that you can turn the things in the other way by focusing your study on functions having an explicit function $r=\varphi(\theta)$ expression ; for cardioids, it is $r=a (1+\cos( \theta))$ ; for lemniscates, it is $r=\pm a \sqrt{\cos(2 \theta)}$. Do you know that Gauss has developed a specific trigonometry for lemniscates ? – Jean Marie Aug 01 '21 at 12:17
  • @JeanMarie No, I know very basic spherical and hyperbolic geometry, but not leminscate geometry. Actually, of you use this method for y=ln(x), you would get this with x=$\mathrm{\sqrt{\cot(\theta)},y=\sqrt{cot(\theta)}tan(\theta)}$ and not the usual hyperbolic functions. This method is slightly different, like in the question graphic. – Тyma Gaidash Aug 01 '21 at 12:19
  • http://recreatiimatematice.ro/arhiva/corespondente/RM22019REISNER.pdf?i=1 (it is in french ; but you should understand a good part of it) – Jean Marie Aug 01 '21 at 12:24
  • @JeanMarie Thanks for the help. Gauss developed a whole new mini trigonometric field of study whereas, this method takes the function literally instead of a circle, a square pattern, for instance. It just modifies the functions, but the leminscate functions do not seem easy to write in terms of cos and sine. Does this make sense? – Тyma Gaidash Aug 01 '21 at 12:29
  • I have been working on these issues from the point of view of conformal mapping (indeed mapping a disk onto a square) but I am not a specialist of these questions. Besides, I will be frank: it is difficult for me to follow the line of thought of your question because I am lost in the numerous details. Could you capture in a few (final) sentences which kind of general result you are looking for ? – Jean Marie Aug 01 '21 at 13:07
  • .@JeanMarie Basically, I just need a way to solve f(x,x tan(θ))=0 for x. You can also find a way to solve {r=r(θ), y=x tan(φ)}, φ is the angle of rotation. All I did was give examples of specific cases of $\frac{f(x)}{x}=\tan(\theta)$ and f(x,y)=0, y=x tan(θ) solved in the question. This is the setup to solve for x=sin(φ) which will give the alternate trigonometric functions. – Тyma Gaidash Aug 01 '21 at 14:57
  • By the way, I got these graphical results by using the cartesian equation for the leminscate, even though polar would have been better, and solving it with a system of y=x tan(θ). Maybe these can also be considered leminscate trigonometry? – Тyma Gaidash Aug 01 '21 at 15:50
  • I have had a look at your nice graphical results. Please note that you can simplify the "green" expressions by using sec=1/cos etc. No idea if it can be considered on the same foot as for example functions based on Gauss "lemn" function. – Jean Marie Aug 01 '21 at 17:20
  • 1
  • @Arctic Did you find this old post from the editing sandbox? – Тyma Gaidash Oct 21 '21 at 15:19
  • No, I came by this when I review the edit by islamm. – Arctic Char Oct 21 '21 at 15:24

0 Answers0