91Ó°ÊÓ

A subspace is a special subset of a vector space that has some important properties. Imagine a quiet corner in a library where you can think of vectors being arranged in an orderly fashion.
Unlike a random collection of vectors, a subspace:

• Contains the zero vector, meaning it's the starting point or origin of the subspace.
• Is closed under addition, so you can add any two vectors from the subspace and the result will still be in the subspace.
• Is closed under scalar multiplication, meaning if you multiply a vector from the subspace by any scalar (like 2, -3, or even 0.5), the result will also remain in the subspace.

When we talk about a subspace of \( \mathbb{R}^{n} \), we mean a subset of this n-dimensional space where these three properties hold. These conditions help maintain a certain structure that makes calculations like projections manageable. Subspaces are like smaller worlds within the larger world of vectors, governed by simple yet powerful rules.

Vector projection is a way to "shadow" a vector onto a subspace, which helps to find the closest vector in that subspace to the original vector.
Imagine shining a light on a stick standing upright on the ground, creating a shadow. The shadow represents the projection. Mathematically, if you have a vector \( \mathbf{x} \) and you want to "project" it onto a subspace \( W \), you find the vector in \( W \) that is closest to \( \mathbf{x} \).
The projection onto \( W \) is denoted as \( \operatorname{proj}_W(\mathbf{x}) \). This is calculated using the formula:

• Find a vector in \( W \) that has the smallest distance to \( \mathbf{x} \).
• Ensure this vector is closest in terms of direction and magnitude, residing completely within \( W \).

This process simplifies our calculations by ensuring that everything remains within the tidy and structured space of \( W \). Projections are especially useful in minimizing errors and optimizing space usage in applications like computer graphics and statistics.

A fundamental property of vector projections is captivating because it simplifies complex problems. When you project a vector onto a subspace twice, the second projection won't change anything. Here's why:
Consider you have already projected \( \mathbf{x} \) onto \( W \) to become \( \operatorname{proj}_W(\mathbf{x}) \). This new vector is firmly within the subspace \( W \).
Projecting \( \operatorname{proj}_W(\mathbf{x}) \) onto \( W \) again results in no change. This happens because the target is already exactly where it should be. It's like trying to improve perfection: once a vector is perfectly aligned within the subspace, re-projecting it confirms its position rather than altering it.
In mathematical terms, this is expressed as \( \operatorname{proj}_W(\operatorname{proj}_W(\mathbf{x})) = \operatorname{proj}_W(\mathbf{x}) \). Through this property, we understand stability within mathematical operations, ensuring that calculations involving projections are both efficient and predictable.