Clearing the cruff
At this point, we have constructed a number of objects that feel similar:
- The space $\mathbb{R}^n$.
- The null space $\mathcal Z(A)$ of a matrix.
- The range $\mathcal R(A)$ of a matrix.
Geometrically, they all correspond to hyperplanes in $\mathbb{R}^n$ and they all share similar properties.
As an example, we have already seen any range can be expressed as a null space and that any null space can be expressed a range.
Similarly, the set of $t$ free variables found in solving $A \mathbf{x} = \mathbf{0}$ feel very much like vectors in $\mathbb{R}^t$. We can plug in any values for those free variables and get a solution. There is no constraint on them. Much like $\mathbb{R}^t$ is the set of $t$ values in $\mathbb{R}$ with no constraints on them.
In mathematics, when confronted with the many similar-seeming objects that are not all exactly the same in our current framework, it makes sense to search for an abstraction that can unify all the objects as instances of a single idea.
That is what vector spaces are. Almost everything we have seen is an element of a vector space.
Table of contents
- Fields
- Definitions and examples
- Subspaces
- Spans
- Linear independence
- Bases and dimension
- Rank and nullity
- Inner products
- Gram Schmidt
- QR factorizations