PDA

View Full Version : Projecting a vector onto a plane.


dcl
Aug29-04, 12:10 AM
Heya's
how would one go about spanning a vector say 'u' onto a plane spanned by vectors v1 and v2.

I have a formula for projecting a vector onto say a subspace w:
projw(u) = <u,v1>v1 + <u,v2>v2 + .... <u,vn>vn
But I'm unsure how to use this for when I need to project the vector onto a plane spanned by 2 other vectors.

Thanks.

humanino
Aug29-04, 03:30 PM
dcl : I am confused by the fact that you provide the correct formula !

Say you have a vector \vec{V}=\{V_i\} with components indiced by i in a general n dimensional linear (vector) space : i\in \{0,1,2,\cdots ,n\}. Say in this n dimensional space you have a plane defined by two vectors \vec{u}^{(1)} = \{u^{(1)}_i\} and \vec{u}^{(2)} = \{u^{(2)}_i\}. Then the straightforward application of your formula leads to the projection P(\vec{V}) of the vector \vec{V} onto the plane spanned by \vec{u}^{(1)} and \vec{u}^{(2)} :

P(\vec{V}) = \sum_{i=1}^2 \langle \vec{u}^{(i)},\vec{V} \rangle \vec{u}^{(i)} = c_1 \vec{u}^{(1)} + c_2 \vec{u}^{(2)}

with c_1 = \langle \vec{u}^{(1)},\vec{V} \rangle = \sum_{i=1}^n u^{(1)}_i V_i and c_2 = \langle \vec{u}^{(2)},\vec{V} \rangle = \sum_{i=1}^n u^{(2)}_i V_i

dcl
Aug29-04, 03:57 PM
Thanks for that, guess it was simpler than I thought. :)

robphy
Aug31-04, 10:24 PM
If I'm not mistaken,
in projw(u) = <u,v1>v1 + <u,v2>v2 + .... <u,vn>vn ,
the v1,...,vn should be mutually-orthogonal unit vectors
since a projection must satisfy projw(projw(u))=projw(u).

schemeng
Oct19-10, 11:11 PM
How could you create a matrix that performs this transformation? In other words, what matrix would project any vector V into the plane spanned by v1 and v2?