You are on page 1of 1

A homogeneous system of linear equations can be written as Your statement of the problem specifies that

be but the problem is true in greater generality than that.

How do we make a new system of equations whose equations are linear combinations of the original system? We
act linearly on the rows of a matrix by multiplying it on the left. To get a matrix whose rows are linear
combinations of the rows of we must have a matrix such that The statement "if each equation in
each system is a linear combination of the equations in the other system" is symmetric, so we must also have
invertible so we can write

So the theorem is this: If there is an invertible such that then and have the same
solution.

The proof is very simple. If we can left multiply by to get And if we can left
multiply by to get So iff meaning the systems are equivalent.

When I say that the problem is true in greater generality than for matrices, I include that nothing about this
proof requires that and even be square.
A few comments:

When I referred to a 'system of linear equations', I referred to the m linear equations in n variables represented by
a general m x n matrix.

Secondly I am familiar with the theorem that if C is an invertible square matrix and the matrix product A = CB
exists then Ax = 0 and Bx = 0 have exactly the same solutions. This is what you have proved Kent.

I suppose I should clarify my question to fit it with your language. If we have two m x n matrices A and B and Ax =
0 and Bx = 0 have exactly the same solutions, how do we show that there exists an invertible m x m matrix C for
which A = CB?

This problem seems more difficult than the one you've solved, if we are to find an elementary solution (see below).

The reason I framed the question as before is it's in the first chapter of my linear algebra book (Hoffman and
Kunze). Though I am familiar with invertible matrices and so on, the question was posed in the exercises before
matrices were even introduced. Hence I suppose there must be a more elementary solution which does not utilize
the language of matrices. But that is obviously no problem.

You might also like