how to find orthonormal vector with condition?

56 visualizzazioni (ultimi 30 giorni)
Sara
Sara il 23 Set 2025 alle 8:44
Commentato: Sara il 25 Set 2025 alle 10:50
Hello,
This should be simple, but I can't figure it out. I need to create an orthonormal vector whose x component is as close to 0 as possible. So far I know how to get the orthonormal basis of my vector but how can I find a vector within the basis that meets the condition of being as close to 0 as possible?
a = [x y z]; % my vector
n = null(a.'); % orthonormal basis
  2 Commenti
Torsten
Torsten il 23 Set 2025 alle 10:28
Modificato: Torsten il 23 Set 2025 alle 11:20
If y = 0, take [0, 1, 0]. If y ~= 0, take [0,-z,y]/sqrt(y^2+z^2).
Sara
Sara il 25 Set 2025 alle 10:50
Thanks! This was what I needed.

Accedi per commentare.

Risposte (2)

Matt J
Matt J il 23 Set 2025 alle 13:03
n=[0,null([y,z])']

John D'Errico
John D'Errico il 23 Set 2025 alle 13:04
Modificato: John D'Errico il 23 Set 2025 alle 14:03
Confusing question. Which probably means I'm misunderstanding your goal here, or that I am myself confused, which is not uncommon. But I'll take a shot.
Consider a vector v.
v = randn(1,4)
v = 1×4
0.5393 0.4693 0.4517 -1.3974
<mw-icon class=""></mw-icon>
<mw-icon class=""></mw-icon>
I want to find a vector u orthogonal to v, but it must have the property that u(1)==0? That is simple.
U = null(v(2:end))
U = 3×2
-0.2930 0.9064 0.9342 0.2036 0.2036 0.3702
<mw-icon class=""></mw-icon>
<mw-icon class=""></mw-icon>
The columns of U contain 2 vectors, such that in each case if I expand them by padding a zero element as the first element, the new vectors will be orthogonal to v, AND they themselves have unit norm.
u1 = [0;U(:,1)]
u1 = 4×1
0 -0.2930 0.9342 0.2036
<mw-icon class=""></mw-icon>
<mw-icon class=""></mw-icon>
u2 = [0;U(:,2)]
u2 = 4×1
0 0.9064 0.2036 0.3702
<mw-icon class=""></mw-icon>
<mw-icon class=""></mw-icon>
dot(v,u1)
ans = 1.3878e-16
dot(v,u2)
ans = -2.7756e-16
norm(u1)
ans = 1
norm(u2)
ans = 1.0000
And to within floating point trash, they satisfy the requirements. u2 and u2 are not unique of course. But u1 and u2 form an orthonormal set with first element zero, that are each normal to v.
It would have been a somewhat more interesting question if the requirement was to find an orthonormal set with the property that u(1) was some other given number, perhaps 1/2. (Its not too difficult to modify the scheme I used in that case.) But zero is pretty easy.

Categorie

Scopri di più su Dynamic System Models in Help Center e File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by