Proving symmetry of metric (single linkage between clusters using arbitrary dissimilarity measure)

201 Views Asked by At

I am told to assume that our dissimilarity measure $d$ satisfies the properties required of it, what seems to be the definition of a metric:

$d(x,y) \geq0 $ and $d(x,y)=0 \Longleftrightarrow x=y$

$d(x,y)=d(y,x)$

$d(x,z)\leq d(x,y) + d(y,z)$

Given that, we define single linkage to be $$d(A,B)=\min_{x\in A, y\in B}\{d(x,y)\}$$ We are told to assume that our dissimilarity measure $d$ satisfies the properties above.

I want to prove (or contradict...) $$d(A,B)=d(B,A)$$

Since I know $d(x,y)=d(y,x)$ it seems like it ought to be straightforward.

Suppose we have $x\in A , y\in B$, then $d(x,y)=d(y,x)$ still holds, as far as I can see.

$\Rightarrow \min\{d(x,y)\}=\min\{d(y,x)\}$

$\Rightarrow \min_{x\in A, y\in B}\{d(x,y)\} = \min_{x\in A, y\in B}\{d(y,x)\} $

$\Rightarrow d(A,B)=d(B,A)$ as required.

Does this constitute an adequate proof? Can someone help me make it a little more rigorous? Is there something obvious I'm missing?