see also rank-nullity theorem
linear algebra, higher linear algebra
(…)
For a linear operator between vector spaces, where is finite-dimensional of dimension , then the dimensions of the kernel and the image of are related by:
If is a finite rectangular matrix then the row rank (the dimension of the span of its rows (“row space”)) plus the dimension of the solution space of equation (“nullspace”) yields the number of columns. From the other side, column rank (the dimension of the space of columns) plus the dimension of the solution space of equation (or, equivalently, ) yields the number of rows.
Notice that now we have two statements as a rectangular matrix can be thought as a linear operator in two ways: acting by left multiplication on column vectors or acting by right multiplication on row vectors.
Sometimes additional statements are included by some authors.
Most often the additional part quoted is that the row rank is the same as the column rank. This is a somewhat nontrivial statement in our formulation (dimension of the space of rows/columns); in some expositions of the theory the definitions of ranks are such that this statement is automatic. Namely, the rank is the minimal dimension such that the -matrix can be written in the form where is , that is is , and is an matrix. Then the transposed matrix can be written as . This is nothing than the image factorization of a linear map and its transpose which both go through the same intermediate space.
The following 2 refereces have somewhat nonstandard exposition of linear algebra to be amenable to constructive/algorithmic treatments useful in numerical linear algebra:
Last revised on May 21, 2024 at 13:17:32. See the history of this page for a list of all contributions to it.