Barycentric coordinates on an affine space

84 Views Asked by At

In the book 'Geometry of Quantum States' the author states that in an $n$-dimensional affine space, select $n+1$ points $x_i$ so that an arbitrary point $x$ can be written as:

$x=\mu_0x_0+\mu_1x_1+.........\mu_nx_n$ , where $\mu_0+\mu_1+....+\mu_n=1$

My doubt is that in any $n$-dimensional space, we can define any point as a sum of $n$ independent coordinates as the dimension of basis is $n$. So why choose $n+1$ points here as one of these points has to be linearly dependent on the others?

1

There are 1 best solutions below

0
On BEST ANSWER

I've been working on the same problem. My thoughts goes as follows.

Define line as affine space $ (\mathbb A, V) $ with $ \dim V = 1 $. It can be described (why?) as $ \{P + \alpha v \mid \alpha \in \mathbb K\} $, where $ v $ is a basis of $ V $ and $ P $ is any point. Now we can prove following proposition: if $ P_1 \neq P_2 $, then $ \{P + \alpha v \mid \alpha \in \mathbb K\} = \{\mu_1 P_1 + \mu_2 P_2 \mid \mu_1 + \mu_2 = 1\} $.

For $ \dim V = 2 $ remember that there exist unique plane passing through 3 points, if they are not on the same line (it is crucial; why?). Prove similar proposition for plane — affine space of dimension $ 2 $.

Now $ \dim V = n $. What conditions we have to impose on $ (O, v_1, \dots, v_n) $ and $ (P_1, \ldots, P_{n + 1}) $ to get the equality as earlier? From proof it should be clear why we take exactly $ n + 1 $ points and what conditions should be.