Distance Between a Subspace and a Vector

In summary, the conversation discusses finding the distance between a vector and a subspace in a Euclidean space. The answer involves finding the projection of the vector onto the orthogonal complement of the subspace and using the Gram-Schmidt orthogonalization procedure to find the distance. An example is provided and the answer is confirmed to be correct.
  • #1
Sudharaka
Gold Member
MHB
1,568
1
Hi everyone, :)

I just want to confirm my answer to this question.

Question:

Find the distance between a vector \(v\) and a subspace \(U\) in a Euclidean space \(V\).

Answer:

Here what we have to find essentially, is the length of the projection of \(v\) to the orthogonal compliment, \(U^\perp\) of \(U\). Hence if we can find a orthogonal basis of \(U^\perp\); say \(\{e_1,\,e_2,\,\cdots,\,e_n\}\) then the distance is given by,

\[d=\left|\frac{v.e_1}{e_1. e_1}e_1+\cdots+\frac{v.e_n}{e_n. e_n}e_n\right|\]

To find the orthogonal basis we might want to use the Gram-Schimdt orthogonalization procedure.

Let us take an example. Let \(v=(1,\,2,\,3,\,4,\,5)\) and the subspace \(U\subset \mathbb{R}^5\) given by,

\[x_1+2x_2+3x_3+4x_4=0\]

\[5x_1+6x_2+7x_3+8x_4=0\]

Therefore we have that \(v_1=(1,\,2,\,3,\,4,\,0)\mbox{ and }v_2=(5,\,6,\,7,\,8,\,0)\) as linearly independent vectors of \(U^\perp\). Now using the Gram-Schimdt orthogonalization procedure we get,

\[e_1=v_1=(1,\,2,\,3,\,4,\,0)\]

\[e_2=v_2-\frac{v_2 . e_1}{e_1. e_1}e_1=\left(\frac{8}{3},\,\frac{4}{3},\,0,\,-\frac{4}{3},\,0\right)\]

Now that we have found a orthogonal basis for \(U^\perp\) we can find the distance as,

\[d=\left|\frac{v.e_1}{e_1. e_1}e_1+\frac{v.e_2}{e_2. e_2}e_2\right|=|e_1|=\sqrt{30}\]

Am I correct? :)
 
Physics news on Phys.org
  • #2
Yup. All correct.
 
  • #3
I like Serena said:
Yup. All correct.

Wow, that's great. I have finally understood something. Thanks very much for the confirmation. :)
 

FAQ: Distance Between a Subspace and a Vector

What is the definition of distance between a subspace and a vector?

The distance between a subspace and a vector is the shortest distance between the vector and any point on the subspace. It is also known as the perpendicular distance or the minimum distance.

How is the distance between a subspace and a vector calculated?

The distance between a subspace and a vector can be calculated using the formula d = ||v - projSv||, where v is the vector and projSv is the projection of the vector onto the subspace.

Is the distance always positive between a subspace and a vector?

Yes, the distance between a subspace and a vector is always positive. This is because the distance is calculated as the length of the shortest line segment between the vector and the subspace, which cannot be negative.

Can the distance between a subspace and a vector be zero?

Yes, the distance between a subspace and a vector can be zero. This occurs when the vector lies on the subspace, meaning that the vector is already a part of the subspace.

How can the distance between a subspace and a vector be used in real-life applications?

The concept of distance between a subspace and a vector is commonly used in machine learning and data analysis. It can help in determining the similarity between data points and identifying outliers. It can also be used in optimization problems, such as finding the shortest distance between a point and a line or a plane.

Back
Top