Академический Документы
Профессиональный Документы
Культура Документы
1
For that reason, the image of A, im(A), is often elementary matrices are invertible, so the row op-
called the column space of A; it’s the subspace of erations preserve rank.
F m spanned by the columns of A. Elementary column operations will multiply a
What about the row space of A, the subspace matrix by an elementary matrix on the right, so
of F n spanned by the rows of A? We’ll look at they’ll preserve rank, too. q.e.d.
that next, and we’ll find that the row space and
the column space have the same dimension. (In By means of elementary row and column oper-
some textbooks, the dimensions of these two vector ations, you can always transform a matrix to one
spaces are called the column rank and the row rank, in which there are only 0’s and 1’s, and the 1’s, if
then they show the two ranks are equal.) there are any, start at the upper left corner and are
placed down and to the right, for example
Composition with isomorphisms doesn’t
1 0 0 0
change the rank. Let’s first look at this in terms 0 1 0 0
of linear transformations, then translate our results 0 0 0 0
into statements about matrices.
The transformed matrix has the same rank as the
Theorem 5. Composing a linear transformation original one, and its rank is equal to the number of
with an isomorphism doesn’t change rank. 1’s in it. That can be read either as the number of
T nonzero rows, or as the number of nonzero columns.
Proof. Let V → W be a linear transformation, and From this observation, we can derive the following
R S
let V 0 → V and W → W 0 be isomorphisms. Then theorem.
R
we’ll show that the two compositions T ◦ R : V 0 →
T T S Theorem 7. The rank of a matrix is equal to the
V → W and S ◦ T : V → W → W 0 have the same
rank of its transpose. In other words, the dimension
rank that T has.
of the column space equals the dimension of the row
For the first one, every vector in V is in the image
space, and both equal the rank of the matrix.
of R, so the image of T ◦ R is identical to the image
of T , so they have the same rank.
S
For the second one, the isomorphism W → W 0 Math 130 Home Page at
restricts to an isomorphism im(T ) → im(S ◦ T ) http://math.clarku.edu/~djoyce/ma130/
since it’s a one-to-one correspondence that pre-
serves the vector space operations. Therefore im(T )
and im(S ◦ T ) have the same dimension. Thus,
rank(T ) = rank(S ◦ T ). q.e.d.