I am reading some lectures on linear algebra which are on Russian language and few moments have confused me and I was trying to understand them correctly but I failed to do it.

Firstly, let me give you some preliminary definitions and theorems from my lectures:

Definition: A vector space $V$ is called finite-dimensional, if it has a basis consisting of finitely many vectors. Otherwise, we call the space to be infinite-dimensional.

Theorem: In finite-dimensional vector space each basis has the same number of vectors.

The proof of this theorem is based on the following lemma:

Lemma: Let $\{e_1,\dots,e_m\}$ and $\{f_1,\dots,f_n\}$ be two linearly independent system of vectors such that the second system is contained in the linear span of the first. Then $n\leq m$.

Definition: The dimension of finite-dimensional vector space $V$ is the number of elements in each basis of $V$. If $V$ is infinite-dimensional, then we write $\dim V =\infty$.

Then he is proving the following statement which confuses me.

Statement: Subspace $W$ of finite-dimensional space $V$ is finite-dimensional and $\dim W\leq \dim V$.

Proof: Since $V$ is finite-dimensional then $\dim V=m$ and let $\{e_1,\dots,e_m\}$ basis of $V$. Suppose that $\dim W>m$ then $W$ contains linearly independent vectors $f_1,\dots,f_n$ with $n>m$. Then $\{f_1,\dots,f_n\}\subset \langle e_1,\dots,e_m\rangle=V.$ But this contradicts to the above lemma. Hence $\dim W\leq \dim V$.

I understood the idea of the proof but cannot understand some technical moments.

Question 1: If $\dim W>m$ then why $W$ contains linearly independent vectors $f_1,\dots,f_n$ with $n>m$. Intuitively I know this but can anyone show it rigorously?

Question 2 (sorry for stupid question): Suppose we have shown that $\dim W\leq \dim V$ then how it follows that $W$ is also finite-dimensional?

  • $\begingroup$ If there were not $n$ linearly independent vectors in $W$, then you would take as many as you could and they would have to span $W$ (i.e. would be a basis with less than or equal to $m$ elements). $\endgroup$ – Morgan Rodgers Nov 12 '19 at 19:39
  • $\begingroup$ For question 2, if $\dim{W} = \infty$ that would be bigger than $\dim{V}$. $\endgroup$ – Morgan Rodgers Nov 12 '19 at 19:41
  • $\begingroup$ Do you have the definition "A vector space is $n$-dimensional if it has a basis with $n$ elements"? $\endgroup$ – 79037662 Nov 12 '19 at 19:42
  • $\begingroup$ @MorganRodgers, to be honest I didn't understand your first comment. Could you explain it one more time? $\endgroup$ – ZFR Nov 12 '19 at 19:59
  • $\begingroup$ Sure, if you take a collection of linearly independent vectors in $W$, starting with a single vector, then at each step either 1: the collection you have spans all of $W$, and so is a basis for $W$ or 2: they don't span all of $W$ so you can add a vector to get a larger collection of linearly independent vectors. You just do this until you have more than $n$. $\endgroup$ – Morgan Rodgers Nov 12 '19 at 20:02

Question 1: Suppose $\dim W > m$. By your second definition, the dimension of a finite-dimensional vector space is the number of elements of each basis of this vector space. This surelly works for vector subspaces too. So, if $W$ is finite dimensional, each basis of $W$ has more than $m$ elements. But we know that the elements of a basis must be linearly independent, by the definition of a basis. So, we conclude that $W$ contains linearly independent vectors $f_{1},...,f_{n}$, for some $n>m$. If $W$ is infinite-dimensional, the statement still holds. To see this, suppose that there is no set of linearly independent vectors $f_{1},...,f_{n}$ in $W$. Then, every set of $n$ vectors is linearly dependent, and thus $\dim W < n$, which contradicts the fact that $W$ is infinite-dimensional.

Question 2: If $\dim W \le \dim V$ then $\dim W \le m$, since $m = \dim V$. Thus, the dimension of $W$ must be, at most, equal to $m$, which is a finite number. It follows that $W$ is finite-dimensional, once its dimension if finite.

  • $\begingroup$ Regarding question 1: You said "by your second definition.." but what if $W$ is infinite-dimensional? $\endgroup$ – ZFR Nov 12 '19 at 20:43
  • $\begingroup$ Yes, I was editing it! haha thanks! $\endgroup$ – IamWill Nov 12 '19 at 20:44
  • $\begingroup$ Regarding question 2: You have shown that $\dim W\leq m$ but how it follows that $W$ is finite-dimensional by these definitions? $\endgroup$ – ZFR Nov 12 '19 at 20:48
  • $\begingroup$ Because every basis of $W$ must have at most $m$ vectors, and then you apply your second definition again. $\endgroup$ – IamWill Nov 12 '19 at 20:51
  • $\begingroup$ It is still not so clear to me. You said that each basis of $W$ must have at most $m$ vectors. Why? What if some basis has $>m$ vectors. $\endgroup$ – ZFR Nov 12 '19 at 20:57

By definition, if $n=\dim W$ then $n$ is the number of elements in (any) a basis of $W$.

Suppose $W \subset V$ and $p=\dim V $ is finite. Let $n=\dim W$, then there is a basis $b_1,...,b_n$ for $W$.

If $V \setminus W$ is non empty we can find some $b_{n+1} \in V \setminus W$ and the resulting collection $b_1,...,b_{n+1}$ is linearly independent.

We can continue this process (with $\operatorname{sp} \{b_1,...,b_{n+1} \}$ in place of $W$) a finite number of times. When it terminates, the resulting collection will be a basis of $V$ from which it follows that $n \le p$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.