Let be invertible.
We will do both parts at the same time by giving a left and right inverse for . Recall that
Note that the order of multiplication is reversed when we take transposes. Since is invertible, exists and
Since , transposing both sides of the equations above yields
In summary,
has a left and right inverse, so it's invertible, and we have shown that the left and right inverse are both , i.e., .
Let and be matrices such that .
Note that is surjective: given , we have
since matrix multiplication is associative. If we set , then , so because was arbitrary, this shows that is surjective.
Because is a square matrix, surjective is equivalent to invertible, so is invertible.
Now that we know that is invertible, we know that exists. Thus, multiplying both sides of by ,
Hence, is equal to the invertible matrix , so itself is invertible with .
Let , be linear maps between vector spaces, where . If , then and are both invertible with and .
To prove this, let's show that is surjective. The proof is the same as before: let , and notice that
As before, setting gives , so is surjective, hence invertible. I'll reprove this here in case you haven't seen this fact:
Because is surjective, it follows that . Since is finite-dimensional, we may apply rank-nullity to get
so is injective. Thus, is a bijection, and the inverse of a linear map is linear, so is an isomorphism. (Recall that isomorphism here means that is a linear bijection and that is a linear map.)
The same calculations as above still work:
Alternatively, you could have shown that is injective, hence invertible.
Let be an matrix and be an matrix such that . Show that is injective.