English:
Let be an matrix and be an matrix.
By assumption, any vector satisfying also satisfies .
That is, if we denote the null space (kernel) of matrix as and that of matrix as , then:
This implies the following inequality for the dimensions of the kernels (nullities):
From the Rank-Nullity Theorem, for any matrix :
Since matrices and both have columns:
Substituting these into the inequality:
This completes the proof.
English:
By assumption, .
For any subspace , let denote its orthogonal complement. Taking the orthogonal complement of both sides reverses the inclusion relation:
For any matrix, the orthogonal complement of its null space is its row space. That is, if is the row space of matrix , then .
Thus, the inclusion can be rewritten as:
This means every row vector of matrix can be expressed as a linear combination of the row vectors of matrix .
Let be the -th row vector of , and be the -th row vector of .
For each , there exist scalars such that:
Define an matrix where the -th entry is .
Then, the -th row of the product is , which equals .
Therefore, holds.
This proves that there exists an real matrix satisfying .