Dimensionality, Rangespace & Nullspace Problem

In summary, the problem is trying to find a basis for a transformation matrix, and determining the dimensionality of the matrix. Once the dimensionality is known, the problem is to compare it to the dimensionality of the matrix having just one dimension.
  • #1
e(ho0n3
1,357
0
[SOLVED] Dimensionality, Rangespace & Nullspace Problem

Homework Statement
Prove (where A is an n x n matrix and so defines a transformation of any n-dimensional space V with respect to B, B where B is a basis of V) that [itex]\dim(R(A) \cap N(A)) = \dim R(A) - \dim R(A^2)[/itex]

The attempt at a solution
If I determine the basis of [itex]R(A) \cap N(A)[/itex], I can determine its dimensionality and then compare it with [itex]\dim R(A) - \dim R(A^2)[/itex].

I've been unsuccessful at finding a basis. Also, given that [itex]\dim R(A) = m[/itex], is there a way to determine what [itex]\dim R(A^2)[/itex] is?
 
Physics news on Phys.org
  • #2
Let [tex] \{Av_1, \ Av_2, \ ..., \ Av_k\} [/tex] be a basis for [tex]R(A)\cap N(A)[/tex] and [tex] \{A^2u_1, \ A^2u_2, \ ..., \ A^2u_m\} [/tex] be a basis for [tex]R(A^2)[/tex], with [tex] v_1, \ v_2, \ ..., \ v_k, \ u_1, \ u_2, \ ..., \ u_m \in V[/tex].

We claim that X = [tex]\{Av_1, \ Av_2, \ ..., \ Av_k, \ Au_1, \ Au_2, \ ..., \ Au_m\}[/tex] is a basis for [tex]R(A)[/tex].

Proof:
1) X spans [tex]R(A)[/tex].

Take any [tex] w \in R(A)[/tex]. Since [tex]Aw \in R(A^2)[/tex], so [tex]Aw = b_1 A^2u_1 \ + b_2 A^2u_2 \ + ... \ + b_m A^2u_m[/tex], for some scalars [tex]b_1, \ b_2, \ ..., \ b_m[/tex].

Thus, [tex]A(w \ - b_1 Au_1 \ - b_2 Au_2 \ - ... \ - b_m Au_m) = \mathbf{0}[/tex] and so [tex] w \ - \ b_1 Au_1 \ - \ b_2 Au_2 \ - \ ... \ - \ b_m Au_m \in R(A) \cap N(A)[/tex] (Why?)
Complete the proof yourself :)

Hence, every element in [tex]R(A)[/tex] can be expressed as a linear combination of [tex]Av_1, \ Av_2, \ ..., \ Av_k, \ Au_1, \ Au_2, \ ..., \ Au_m[/tex] and so X spans [tex]R(A)[/tex].

2) X is a linearly independent set.
We solve the homogeneous equation [tex]c_1 Av_1 \ + \ c_2 Av_2 \ + \ ... \ + \ c_k Av_k \ + \ c_{k+1} Au_1 \ + \ c_{k+2} Au_2 \ + \ ... \ + \ c_{k+m} Au_m = \mathbf{0}[/tex], where [tex]c_1, \ c_2, \ ..., \ c_{k+m}[/tex] are scalars.

Now, [tex]A(c_1 Av_1 \ + \ c_2 Av_2 \ + \ ... \ + \ c_k Av_k \ + \ c_{k+1} Au_1 \ + \ c_{k+2} Au_2 \ + \ ... \ + \ c_{k+m} Au_m) = \mathbf{0}[/tex], and thus...
Complete the proof yourself: I suggest first showing that [tex]c_{k+1} \ = c_{k+2} \ = ... \ = c_{k+m} \ = 0[/tex]

Since the homogeneous equation has only the trivial solution, X is a linearly independent set.

This proves our claim and the result follows.
 
  • #3
Neat. I wish I had your intuition.

pizzasky said:
Thus, [tex]A(w \ - b_1 Au_1 \ - b_2 Au_2 \ - ... \ - b_m Au_m) = \mathbf{0}[/tex] and so [tex] w \ - \ b_1 Au_1 \ - \ b_2 Au_2 \ - \ ... \ - \ b_m Au_m \in R(A) \cap N(A)[/tex] (Why?)
Complete the proof yourself :)

w is in R(A) and so is [tex]- b_1 Au_1 - \cdots - b_m Au_m[/tex] because it's a linear combination of members of X. Thus [tex]w - b_1 Au_1 - \cdots - b_m Au_m[/tex] is in R(A). Since [tex]A(w - b_1 Au_1 - \cdots - b_m Au_m) = \mathbf{0}[/tex], [tex]w - b_1 Au_1 - \cdots - b_m Au_m[/tex] is also in N(A).

Now, [tex]A(c_1 Av_1 \ + \ c_2 Av_2 \ + \ ... \ + \ c_k Av_k \ + \ c_{k+1} Au_1 \ + \ c_{k+2} Au_2 \ + \ ... \ + \ c_{k+m} Au_m) = \mathbf{0}[/tex], and thus...
Complete the proof yourself: I suggest first showing that [tex]c_{k+1} \ = c_{k+2} \ = ... \ = c_{k+m} \ = 0[/tex]

Let [tex]\mathfrak{v} = c_1 Av_1 + \cdots + c_k Av_k[/tex] and let [tex]\mathfrak{u} = c_{k+1} Au_1 + \cdots + c_{k+m} Au_m[/tex]. [tex]A(\mathfrak{v} + \mathfrak{u}) = A\mathfrak{v} + A\mathfrak{u} = A\mathfrak{u} = \mathbf{0}[/tex].

The last equality above plus the fact that [tex]\{A^2u_1, \ldots, A^2u_m\}[/tex] is linearly independent means that [tex]c_{k+1} = \cdots = c_{k+m} = 0[/tex]. Thus [tex]\mathfrak{v} + \mathfrak{u} = \mathbf{0}[/tex] reduces to just [tex]\mathfrak{v} = \mathbf{0}[/tex] and since [tex]\{Av_1, \ldots, Av_k\}[/tex] is a linearly independent set, [tex]c_1 = \cdots = c_k = 0[/tex] as well. Ergo, X is linearly independent.
 

FAQ: Dimensionality, Rangespace & Nullspace Problem

What is dimensionality?

Dimensionality refers to the number of independent variables or features in a dataset. It can also refer to the number of dimensions in a vector space.

What is rangespace?

Rangespace, also known as column space, is the set of all possible linear combinations of the columns of a matrix. In other words, it is the span of the columns of a matrix.

What is nullspace?

Nullspace, also known as kernel, is the set of all vectors that when multiplied by a given matrix result in the zero vector. In other words, it is the set of all solutions to the homogeneous equation Ax=0.

What is the dimensionality of rangespace?

The dimensionality of rangespace is equal to the rank of the matrix, which is the maximum number of linearly independent columns in the matrix.

What is the relationship between rangespace and nullspace?

The rangespace and nullspace are complementary subspaces, meaning that they have no vectors in common and together span the entire vector space. This relationship is known as the fundamental theorem of linear algebra.

Back
Top