0
$\begingroup$

I was working on a problem about geometric medians and I had an idea for a divide and conquer solution, but it would only work if a set of points, when split into two disjoint sets, and those sets respective geometric medians calculated and a line passed between, it holds that the geometric median of the overall set of points lies on that line. When graphing it and visualizing it, seems possible, and when I used existing algorithms and experimented, it seems to be very close but not quite on the same line. I am not sure if this is due to error in calculation of the geometric median or not.

I have some example Python code demonstrating how close the numbers are here: Geometric Median comparison

Any expertise in this matter would be appreciated.

$\endgroup$
1
$\begingroup$

If $A$ and $B$ are two set of points, the geometric median of $A \cup B$ does not necessarily lie on the line passing between the geometric mean of A and the geometric mean of $B$.

In the following example the points of $A$ are red and the one in $B$ is blue. The geometric median of $A$ is the red square. The green line connects the geometric means of $A$ and $B$. The black square is the geometric mean of $A \cup B$, i.e., the Fermat point of the triangle whose vertices are in $A \cup B$.

Geometric mean

$\endgroup$
  • $\begingroup$ Oh wow...how did I miss such an obvious case. Thank you! $\endgroup$ – hLk Jun 7 at 1:41
  • $\begingroup$ Doing some more testing and it looks like the more points you have the closer it falls to the line. Never quite on it though $\endgroup$ – hLk Jun 7 at 3:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.