1

Consider $n$ equally sized circles $c_i$ in the plane with their centers $x_i$ in the vertices $v_i$ of a regular $n$-gon and all meeting in the center of that $n$-gon.

Examples:

  • Fig. 1: circle pattern for $n=5$:

enter image description here

  • Fig 2: circle pattern for $n=15$:

enter image description here

For each pattern, the circles partition the plane into many cells with cell corners defined by circle intersection points (marked blue) and circle arcs as cell border lines. Most cells are bounded by 4 intersection points with exceptions for the innermost and outermost cells.

Observation: all those arcs have the same length, except for the innermost (having half of that length) and the outermost triangular cells (having one double-length arc).

It clearly has to do with all those symmetries, but I failed to find a proper argument for a proof as the cells themselves are not all congruent.

  • Fig 3: shows a detail for $n=9$ with 4 consecutive circles all having point $O$ in common and rotated by the same angle. Due to symmetry $|AB|=|AB'|$ and $|BC|=|B'C|$. But, why is $|AB|=|BC|$?

enter image description here

Any clue is highly appreciated.

KnarfB
  • 21
  • Similar question about "zonotopes" here involving polygons and lozenges instead of circles and "circular cells" – Jean Marie Mar 25 '24 at 20:36
  • Any reaction to my comment ? – Jean Marie Mar 27 '24 at 22:00
  • 1
    Yeah, many thanks. Most references don't show proof though. Found an interesting proof of colinearity for intersection points B, B' and the red/blue yellow/gree in Fig. 3 (in German): https://walser-h-m.ch/hans/Miniaturen/K/Kreisfaecher/Kreisfaecher.htm – KnarfB Mar 28 '24 at 11:05

1 Answers1

1

We fix a circle (here, the one with center $B$), and we rotate the circle by an angle (to get the circle centered in $B'$. By construction, we easily see that $ABCD$ is a diamond and that the intersection $C$ is placed at the same angle that the circle was rotated.

enter image description here

Because rotations of the other circles are regular, the intersections are regularly spaced on the circle.

The center $A$ is never counted as such an intersection, so the two "one half" arcs are just one full arc. We can see the point $D$ as the intersection of the circle with itself, so where we have a double-length arc, we can see it as just two arcs.

caduk
  • 5,621
  • 1
  • 5
  • 23
  • Thanks, I see your point. But in general, point A is not identical to the center O of the pattern. I have added Fig. 3 above to clarify that situation. – KnarfB Mar 26 '24 at 22:17
  • 1
    @KnarfB maybe it is not clear, but the bottom left circle I have represented is not on your figure, I just added it for the construction (but should have probably hid it), so A is really the centre of the figure. – caduk Mar 26 '24 at 22:34