This is to explain further what I said in class about the laws of vector spaces. Since the kind of reasoning involved is a little advanced, this won't be on the midterms or final, but it's worth looking at to see what the issues are.
The laws used in the definition of a vector space are
some of the laws true in
for every
.
Additional laws not part of the definition include basic ones
such as
0 but also miscellaneous ones that could be
proved, such as
.
Near the beginning of the course I mentioned this fact:
Proposition. The defining laws of vector spaces over R are enough
to imply all the laws that hold in
for all
, in
the sense that any vector space does satisfy all these laws.
(The same is true for any field
in place of
,
but let's stick to
.)
At the time it was not clear how a statement like this could be proved. But now it can be. First, a proof that doesn't quite work:
?? ``Proof.'' If
is a vector space over
, then
, so all the laws of
hold in
.
The trouble with this attempted proof is that a vector space
might not be finite dimensional and so might not be isomorphic
to
, no matter how
is chosen. (Notice that in
the attempted proof,
comes out of nowhere.)
Valid proof. Given any law that holds in
for all
, we want to try to verify it in
. The law has a
certain number of variables, say
. We could write the law
using variable symbols
. To show that
the law holds in
, we need to check that the law is true
for all possible ways of putting specific vectors for
. With specific vectors
,
we can take the subspace
that they span. We know that
, being the span of a finite list of vectors, is
finite-dimensional, say of dimension
(with
).
Then
, so the law in question, being true in
, is true in
for those specific vectors.
That's the proof. Notice that each time a different law and
list of specific vectors is checked, a different
is
used.