1

3Blue1Brown clip shows a false proof on $\pi=4$, as we fold a square onto a circle, with the frame

What exactly is the distinction between $lim(len(c))$ and $len(lim(c))$? I heard vaguely about continuity and Euclidean vs Taxicab distance but these don't explain it to me. Following the integral definition of arc length and it involving the curve's derivative, I do see the argument on the velocity vector, but can't make a connection to this. I'd expect $lim$ and $len$ to commute.

What would be the exact mathematical expression for each? (Or for a close enough alternative example, showing lack of commutativity with full mathematical expressions)

insipidintegrator
  • 6,271
  • 3
  • 11
  • 38
OverLordGoldDragon
  • 208
  • 1
  • 10
  • 1
    https://math.stackexchange.com/questions/2679894/what-is-the-proof-behind-lim-fgx-f-lim-gx – insipidintegrator Jul 14 '22 at 14:21
  • 1
    why do you expect $\lim$ and $\text{len}$ (typically defined using an integral, which itself is a limiting process) to commute? THe fact that two limits do not commute is not a flaw in analysis, in fact it is one of the more difficult and exciting features of analysis, and gives us a whole bunch of stuff to study. There are usually conditions under which such commutations are possible, but it's not always possible, and for the circle and $\pi=4$ troll proof, commuting is not valid. – peek-a-boo Jul 14 '22 at 14:21
  • 1
    See [The staircase paradox, or why $\pi\ne4$](https://math.stackexchange.com/q/12906/13130) *AND* [What is the history of staircase or =4 paradox?](https://hsm.stackexchange.com/q/11240/264) *AND* [Lower semicontinuity of length of graph: $L(g)\le\liminf_{n\to\infty}L(f_n)$](https://math.stackexchange.com/q/1793172/13130) *AND* [these other questions](https://math.stackexchange.com/questions/linked/12906?lq=1). – Dave L. Renfro Jul 14 '22 at 14:25
  • Thanks all for the references, I'll try to look through sometime. It'd help if the list was narrower to pinpoint the commutativity issue. – OverLordGoldDragon Jul 14 '22 at 15:15
  • @peek-a-boo I'm certainly not calling it a flaw, it's indeed intriguing. I just don't see how Euclidean and Taxicab distances come to disagree in this particular limit, and can't think of an example where distance computation commutation fails. – OverLordGoldDragon Jul 14 '22 at 15:16
  • The example $\frac{1}{n}\sin(n^2 x)$ is discussed in: Nancy Edwards, [*An instance of intuition and lengths of limiting curves*](https://www.kappamuepsilon.org/Pentagon/Vol_31_Num_1_Fall_1971.pdf), **The Pentagon** 31 #1 (Fall 1971), 22-25 & 45. The following discusses a similar example, along with when such a paradox doesn't occur (roughly, for some $a \leq b$ we have $a \leq f'_n(x) \leq b$ for all $n$ and $x),$: John Anthony Terilla, [*On the convergence of arclength*](https://www.jstor.org/stable/community.30005363), **Journal of Undergraduate Mathematics** 23 (1991), pp. 25-27. – Dave L. Renfro Jul 15 '22 at 07:08
  • A rather detailed historical discussion of curve length is given in: Gilbert E. Traub, [**The Development of the Mathematical Analysis of Curve Length from Archimedes to Lebesgue**](https://www.google.com/search?q=%22The+development+of+the+mathematical+analysis+of+curve+length+from+Archimedes+to+Lebesgue%22&filter=0), Ph.D. dissertation (under Kenneth Philip Goldberg), New York University, 1983, ix + 455 pages (degree awarded in 1984). – Dave L. Renfro Jul 15 '22 at 07:20
  • Hi @OverLordGoldDragon I found a similar question to yours: https://math.stackexchange.com/questions/12906/the-staircase-paradox-or-why-pi-ne4 – insipidintegrator Jul 21 '22 at 07:52
  • Re: duplicates - I'm sure somewhere there lies the answer to my question, but I'm asking on SE to get an SE answer and save time, not to have to dig through extended text. – OverLordGoldDragon Jul 23 '22 at 00:18

0 Answers0