2

I would like to see a construction of the value of $\pi$ from first principles using e.g. a circle, without $\sin$ or $\cos$ and the usual relations for circumference or area.

I would consider e.g. a quarter of a circle, where you move in the $x$ direction from $0$ to $R$ (the radius of a circle), defining an infinitesimal section of the circumference, in order to get an integral expression for the circumference of $\frac14$ of a circle, corresponding to $\frac{\pi R}2$.

I tried to do so, but could not find a ‘simple’ solution for that. Any ideas?

Addendum:

  • I do not want any ‘series’ or something to calculate $\pi$;
  • I want somewhat the ‘proof’ that the circumference of a circle is indeed $2\pi r$ using geometric principles only (like Pythagoras and integration…)
Alex
  • 132
  • 3
    How do you define the length of a curved arc without a "series or something"? How do you define integration without a "series or something"? These are all, intrinsically, limit related concepts. – lulu Feb 22 '18 at 15:52
  • You do not want 'series' but 'integration'. – S. D. Z Feb 22 '18 at 15:52
  • 1
    There can be no proof of that, because that is the definition of $\pi$. – Arnaud Mortier Feb 22 '18 at 15:53
  • What is a "construction" of a value of a number? Do you want to compute $\pi$ within some predefined error $\epsilon$, that is, roughly, to some number of decimal places? – Travis Willse Feb 22 '18 at 15:54
  • 1
    You need a definition of $\pi$ in order to prove the the circumference is $2 \pi r$. Are you instead looking for a proof that the ratio of the circumference of a circle to the diameter is a constant? – Jonathan Feb 22 '18 at 15:54
  • What you can prove is that the ratio circumference/radius does not depend on the radius. – Arnaud Mortier Feb 22 '18 at 15:54
  • @Jonathan I think we are making the same point :) – Arnaud Mortier Feb 22 '18 at 15:55
  • I'm taking this to mean that the OP doesn't want some magical series, such as the power series for $\arctan 1,$ but instead wants a geometrically constructed sequence. – Thomas Andrews Feb 22 '18 at 15:55
  • Seems like a reasonable question. However, it is hard to avoid series for this as you need to measure the length of a unit circle. – copper.hat Feb 22 '18 at 16:02
  • It's a good idea to study some of the (very convincing) pseudo-paradoxes that depend on casual definitions of the lengths of curves. Here for example, or the well known false proof that $\sqrt 2 = 2$ (use the staircase function to approximate the diagonal of the unit square). – lulu Feb 22 '18 at 16:11
  • No no, I want to find an integral function, in which I express a part of the circumference $ds$ as part of $dx$, so that the integral of that from x=0 to 1 gives $pi/2$.. – Alex Feb 22 '18 at 17:04
  • Do you mean something like this (or this)? – Hans Lundmark Feb 22 '18 at 18:06
  • Yes I think so. I have to check it tomorrow, but this looks like the expression I am after... – Alex Feb 22 '18 at 18:08
  • No, not what I want. It uses sin/cos and the knowledge that the arc length of a section is $\pi$. And the other is 'I know the length is'... But that is what I want to calculate... – Alex Feb 23 '18 at 06:12
  • I believe my answer proves from scratch what $\pi$ is. All the missing steps are super easy to fill in. I believe some people have a tendency to read an answer and be like "I can't understand it" but don't realize that if they studied it long enough, they would be able to see what they're saying and that it wasn't written in a too complicated way to understand. I believe my answer is not that hard to interpret. It seems pretty clear to me. If you can't understand it, there's nothing more I can do for you unless you tell me which part you're trying to see why is true through a comment. – Timothy Aug 23 '20 at 01:35
  • I think I figured out what your question was. It was to use the fact that $\pi$ is the area of a circle with radius 1 rather than the fact that $\pi$ was the ratio of circumference to diameter. However, I think it would be easier for people reading the question if you made it clear that that's what your question is. After reading Gerry Myerson's answer, I then saw that you didn't want to use trigonometry. I don't think you even said that in your question. I understand people don't always figure out in advance what type of answer they're going to end up liking. – Timothy Aug 23 '20 at 04:29
  • I kind of sort of don't like the feeling of knowing that neither of the other answers solved your problem and then after I write my answer, I still hear nothing about what you think of my answer. If you don't think it's good, I understand why you might want to say nothing. It would be nice if I had the type of brain that would take you saying so as valuable feedback to be like "Now I learned that some people tend to keep missing thinking of it the right way. There's nothing I can do to solve your problem but I learned that that's the case which is useful." and knew how to be like "I know this – Timothy Aug 27 '20 at 02:43
  • problem can be solved. I'm not giving up on it." However, the truth is I have aspergers and I'm good at interpreting short simple things on the spot like a young and inexperienced child where that's one of the first things I learn from. Almost all my past is buried in the sand. I keep doing my own thinking independently of earlier and then letting it quickly get buried in the sand and doing it again and again independently of earlier. I would probably in actuality get a bad feeling from hearing it and be unable to help it and maybe am better off not knowing. However, I would probably also – Timothy Aug 27 '20 at 02:49
  • quickly get over it until I saw the comment again. – Timothy Aug 27 '20 at 02:50

3 Answers3

3

I think that what you're after is what Archimedes did in his text Measurement of a Circle. There, he proved, using geometric principles and no series, that the area of any circle is equal to a right-angled triangle in which one of the sides about the right angle is equal to the radius, and the other to the circumference, of the circle. If we define $\pi$ as the number such that the perimeter of a circle is $\pi$ times its diameter, then what this means is that the area of a circle with radius $r$ is $\pi r^2$.

  • He used the so-called “method of exhaustion”, which is just another infinite process. I fear the OP would not be satisfied with that either. (I'm not sure there is an answer the OP would find acceptable). – Matthew Leingang Feb 22 '18 at 16:25
  • 2
    @MatthewLeingang I'm not sure either. – José Carlos Santos Feb 22 '18 at 16:26
  • upvoting anyway; seems to be the best possible answer – Matthew Leingang Feb 22 '18 at 16:27
  • Not what I actually wanted. In the end, using a circle of radius 1, I want a mathematical expression (an integral probably), that gives me $\pi/2$ or something. At (0,1) I move some $dx$ towards positive x. But at the point (dx, 1) the distance to the center os larger than 1, so I have to go to the point (dx, 1-h). And so on. Using rectangular triangles and pythagoras, I want to find an expression for $ds$ (an circumference element), so that ds = f(dx). I integrate this from 0 to 1 to get $\pi/2$... – Alex Feb 22 '18 at 17:00
  • 1
    @Alex $\int_{-1}^1 \sqrt{1-x^2} dx$? – Jonathan Feb 22 '18 at 23:57
  • How do you get to that equation? How to derive it? – Alex Feb 23 '18 at 06:13
  • You're supposed to answer the question yourself, not have reading a link required to get the answer. The actual answer itself doesn't answer the question. It doesn't calculate $\pi$. It just proves that the area of a circle with radius $r$ is $\pi r^2$. – Timothy Aug 23 '20 at 01:41
0

The usual formula for arc length is $$\int_a^b\sqrt{1+\left({dy\over dx}\right)^2}\,dx$$

From $x^2+y^2=1$ we get $y=\sqrt{1-x^2}$, so $${dy\over dx}={-x\over\sqrt{1-x^2}}$$ so we are looking at $$\int_0^1\sqrt{1\over1-x^2}\,dx$$ Now the problem is, the usual way to evaluate this integral is to make the substitution $x=\sin u$, $dx=\cos u\,du$, but OP doesn't want to use trig. Well, you have to use something to get $\pi$ to fall out of an integral, you can't get it just using polynomials and such, so I'm afraid you're sod out of luck.

Gerry Myerson
  • 179,216
0

If you take on blind faith that the area of a circle with radius 1 is $\pi$, then you can calculate $\pi$ without trigonometry by taking the power series of $2\sqrt{1 - x^2}$ and integrating it from -1 to 1. The problem is any proof from first principles that the area of a circle with radius 1 is $\pi$ uses trigonometry. First I will calculate $\pi$ directly from its technical definition without using the assumption that the area of a circle with radius 1 is $\pi$. Next, I will prove from first principles that the area of a circle with radius 1 is $\pi$. I don't consider this image a proof from first principles that the area of a circle with radius 1 is $\pi$.

enter image description here

Here's a calculation of $\pi$ directly from the definition.

It's obvious that $\pi = 6 \times \sin^{-1}(\frac{1}{2})$. It turns out that the derivative of $\sin^{-1}$ is an elementary function. All we have to do is figure out the power series for the derivative of $\sin^{-1}$ centered at 0 and then integrate each term to get the power series for $\sin^{-1}$. In general, $\frac{d}{dx}f^{-1}(x) = \frac{1}{f'(f^{-1}(x))}$. So $\frac{d}{dx}\sin^{-1}(x) = \frac{1}{\cos(\sin^{-1}(x))} = \frac{1}{\sqrt{1 - x^2}} = (1 - x^2)^{-\frac{1}{2}}$. To take the power series of this centered at 0, you first take the power series of $(1 + x)^{-\frac{1}{2}}$ centered at 0 and then substitute $-x^2$ for $x$. Now the first derivative of this is $-\frac{1}{2}(1 + x)^{-1\frac{1}{2}}$. Then the second derivative is $(-\frac{1}{2})(-1\frac{1}{2})(1 + x)^{-2\frac{1}{2}}$. Now to get its power series, you divide the derivatives by the factorials to get $(1 + x)^{-\frac{1}{2}} = 1 - \frac{1}{2}x + \frac{1}{2}(\frac{3}{4})x^2 - \frac{1}{2}(\frac{3}{4})(\frac{5}{6})x^3 ...$ Now substituting $-x^2$ for $x$, we get $(1 - x^2)^{-\frac{1}{2}} = 1 + \frac{1}{2}x^2 + \frac{1}{2}(\frac{3}{4})x^4 + \frac{1}{2}(\frac{3}{4})(\frac{5}{6})x^6 ...$ Now finally, $\sin^{-1}(x) = \int_0^x(1 - t^2)^{-\frac{1}{2}}dt = x + \frac{1}{2}(\frac{1}{3})x^3 + \frac{1}{2}(\frac{3}{4})(\frac{1}{5})x^5 + \frac{1}{2}(\frac{3}{4})(\frac{5}{6})(\frac{1}{7})x^7$. So $\pi = 6 \times \sin^{-1}(\frac{1}{2}) = 6(2^{-1} + \frac{1}{2}(\frac{1}{3})(2^{-3}) + \frac{1}{2}(\frac{3}{4})(\frac{1}{5})(2^{-5}) + \frac{1}{2}(\frac{3}{4})(\frac{5}{6})(\frac{1}{7})(2^{-7}) ...)$.

Here's a proof that the area of a circle with radius 1 is $\pi$.

Using the substitution rule in reverse, we get $\int_{-1}^12\sqrt{1 - x^2} = 2\int_{-1}^1\sqrt{1 - x^2} = 2\int_{-1}^1\cos(\sin^{-1}(x)) = 2\int_{\sin(-\frac{\pi}{2})}^{\sin(\frac{\pi}{2})}\cos(\sin^{-1}(x)) = 2\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\cos(\sin^{-1}(\sin(\sin'(x)))) = 2\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\cos^2(x) = 2\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\frac{\cos(2x) + 1}{2} = \pi$

Image source: Area of a circle $\pi r^2$

Timothy
  • 793