Previous | Next --- Slide 9 of 52
Back to Lecture Thumbnails
rbunny

It should be symmetric otherwise <u, v> won't be the same as <v, u>

0x484884

If we take the transpose of each side, we get v transpose times A transpose times u on the left and the same thing on the right (since its just a 1x1 matrix) and since <u,v> = <v,u> we get the same inner product when we use A transpose.

Osoii

So if we transpose u and v, u1v1 changes to v1u1, its value is not changed, same for u2v2. But u1v2 changes to v1u2, and u2v1 changes to v2u1. To make this inner product symmetric, the coefficients of u1v2 and u2v1 must be the same (in this case, they are both 1)

ecdeo

In 2D, with a matrix [[ A B ][ C D ]] (row major)

u·v = Au1v1 + Bu1v2 + Cu2v1 + Du2v2
v·u = Au1v1 + Bu2v1 + Cu1v2 + Du2v2.

Since u·v = v·u and A, B, C, D are constants in terms of u and v, we have

Bu1v2 = Cu1v2
Cu2v1 = Bu2v1

so B = C

atarng

The inner product matrix is symmetric to maintain the requirement that <u, v> = <v, u>

jlessioh

Wow I never would have thought to represent dot and cross products with matrices instead of just vectors.