1

Let $a$, $b$, $c$,$d$, $e$, $f$ distinct complex number on the same circle. Then the chords $ab$, $cd$, $ef$ are concurrent ( possibly at infinity) if and only if $$ \det\left( \begin{matrix} 1&a+b&ab\\ 1&c+d&cd\\ 1&e+f&ef \end{matrix}\right)=0 $$

The condition is invariant under complex affine transformations, so one may assume $a$, $\ldots$, $f$ on the unit circle.

Note that if moreover $b=\bar a = \frac{1}{a}$ and so on, then the chords $ab$, $cd$, $ef$ are perpendicular to the $x$ axis, so parallel. Clearly the determinant is $0$ also.

Clarifies a bit the calculations showing that some chords of a regular $n$-gon ar concurrent.

Provided as reference. I know the answer to my question and will add it soon.

$\bf{Added:}$ Consider a regular $30$-gon like in the picture.  regular 30-gon, 5 concurrent chords

We have $5$ concurrent chords. We can check this by showing that the matrix

$$\left( \begin{matrix} 1&0&-1\\ 1&z+z^{18}&z^{19}\\ 1&z^4+z^{24}&z^{28}\\ 1&z^6+z^{26}&z^{32}\\ 1&z^{12}+z^{29}&z^{41} \end{matrix}\right)$$

has rank less then $3$, where $z$ is a primitive root of $1$ of order $30$. One checks that the vector $(1+z^{13}, -z-z^{12}, 1+z^{13})$ is in its kernel. Note the cyclotomic polynomial $\phi_{30}(z)= 1+z-z^3-z^4-z^5+z^7+z^8$.

$\bf{Added:}$ There is a similar formula in this post. This one seems easier to fathom, and also remember.

$\bf{Added:}$

  1. A great for intersecting chords in regular $n$-gons is the paper by Bjorn Poonen and Michael Rubenstein ( kindly provided by @Gerry Myerson:)

  2. The formula of @Jean Marie:, and why it is really equivalent: the rows of the determinant are $(\sqrt{a b}, \frac{1}{\sqrt{a b}}, \sqrt{\frac{a}{b}} + \sqrt{\frac{b}{a}})$, and one needs a factor of $\sqrt{a b}$ to get the one above.

  3. Let's sketch the solution: the lines $ab$, $cd$, $ef$ are concurrent if and only the mixed product

$$\langle (1,a, \bar a)\times (1,b,\bar b), (1 c, \bar c)\times (1, d, \bar d), (1, e, \bar e)\times ( 1, f, \bar f) \rangle = 0$$

(basic linear algebra). Now if $|a|=\ldots = |f|=1$, then $$(1,a, \bar a)\times (1,b,\bar b)= (1, a, \frac{1}{a}) \times ( 1, b, \frac{1}{b})$$ and we get the result.

orangeskid
  • 56,630
  • 2
    This website is not for challenges. – Gerry Myerson Feb 01 '24 at 23:51
  • 3
  • You haven't actually asked a question. 2. Answer-your-own-question is mainly intended for situations where you post a question, not knowing the answer, and then you figure out the answer, and post it. It's not intended for challenges.
  • – Gerry Myerson Feb 02 '24 at 00:13
  • Try https://math.meta.stackexchange.com/questions/29043/area-51-proposal-mathematical-challenges and the items linked there. – Gerry Myerson Feb 02 '24 at 00:16
  • Does this answer your question? Concurrent triples of chords in a regular n-gon $\quad$ Not quite a duplicate because they focus on regular n-gon, but your determinant appears there too and the solution can be easily modified to the general case. – Calvin Lin Feb 02 '24 at 01:29
  • @Calvin Lin: . Probably equivalent, modulo some calculations. I find this formula easier to remember, hence posted as a reference. Will provide a solution later on. – orangeskid Feb 02 '24 at 01:34
  • 1
    Bjorn Poonen and Michael Rubenstein, The number of intersection points made by the diagonals of a regular polygon, is a thorough investigation of the problem suggested by its title. It is available at https://math.mit.edu/~poonen/papers/ngon.pdf – Gerry Myerson Feb 02 '24 at 02:11