Let $S$ be a vector space. Suppose that for any positive integer $n$, there exists a linearly independent subset $S_n ⊆ V$ of size $n$.

How do you show that $S$ is not finite dimensional but infinite-dimensional.

I understand that to be finite dimensional then it must have a finite then there exists a finite set that such at $\operatorname{Span}(T)=S$. So would it be right to prove by contradiction that there is not a finite spanning set that satisfies $\operatorname{Span}(T)=S$, and hence I could argue that the vector space $S$ is infinite dimensional.

How would I show that $S$ does not have a spanning set $T$, such that $\operatorname{Span}(T)=S$?

  • 6
    $\begingroup$ If it was finite dimensional it would have a finite basis. $\endgroup$
    – copper.hat
    Mar 5 '18 at 22:42

Try to prove the following statement:

If $V$ is a vectorspace and $A = \{x_1, \ldots, x_n\}$ is a subset of $V$ containing $n$ (different) vectors. Suppose $A$ is a generating set for $V$, then every subset of $V$ with more than $n$ elements is a linearly dependent subset.

Given: a vector space $V$ such that for every $n \in \{1, 2, 3, \ldots\}$ there is a subset $S_n$ of $n$ linearly independent vectors.

To prove: $V$ is infinite dimensional.

Proof: Let us prove this statement by contradiction: suppose that $V$ has finite dimension $k$. Fix a basis $\{v_1, \ldots, v_k\}$, then it is given that we can find a set $S_{k+1}$ of $k +1 < \infty$ vectors which are linearly independent. However, the above statement shows that the elements have to be linearly dependent, a contradiction. Hence it must be that $V$ is infinite dimensional.

  • $\begingroup$ How can one assume that by assumption there is a set k+1<infinity elements such that the elements are linearly independent? Thanks $\endgroup$ Mar 6 '18 at 14:02
  • $\begingroup$ @Mathamainia see edit. If this is not clear, I will try to explain :). Also, note that I denoted the vector space by $V$, since I think using $S$ would be confusing together with the $S_n$. $\endgroup$
    – Student
    Mar 6 '18 at 14:18
  • $\begingroup$ Sorry would you mind please explaining $\endgroup$ Mar 6 '18 at 14:41
  • $\begingroup$ @Mathamainia have you read what I have edited? The statement you try to prove says you have a vector space $V$ such that for each $n$ there is a subset $S_n$ of $n$ linearly independent vectors. You try to prove that $V$ is infinite dimensional. I prove this by contradiction, by assuming that $V$ has finite dimension and then use that it is given that I can find a subset with more vectors which are linearly independent. $\endgroup$
    – Student
    Mar 6 '18 at 14:43
  • $\begingroup$ Surely any set Sk+1 of k+1<∞ vectors would be are linearly dependent by the theorem you stated above, and as we are given S is a linearly independent set of vectors then this causes contradiction and hence V is infinite dimensional. (rather than the other way) $\endgroup$ Mar 18 '18 at 15:42

I will use the following definition of "finite dimensional":

Definition A vector space $V$ is finite dimensional if $V$ has a finite spanning set. Otherwise, $V$ is infinite dimensional.

Note: the definition does not presume that we know what "dimension" means.

One has the following elementary theorem about matrices:

Theorem 1 If a matrix $A$ has more columns than rows, then there exists an non-zero solution to $A x = 0$.

Proof: Row reduction.

The theorem has the following consequence:

Theorem 2 Suppose $V$ is finite dimensional with a spanning set of cardinality $n$. Then any linearly independent set in $V$ has cardinality $\le n$.

Proof. Suppose $\{v_1, \dots, v_n\}$ is a spanning set and $\{w_1, \dots, w_m\}$ is linearly independent. Since $\{v_1, \dots, v_n\}$ is spanning, there exist $a_{i, j}$ in the underlying field $K$ such that $w_j = \sum a_{i, j} v_i$. Let $A$ denote the matrix $(a_{i, j})$. If $m >n$, there exists $$x = \begin{bmatrix}x_1 \\ \vdots \\ x_m \end{bmatrix} \ne 0$$ such that $A x = 0$. Check that this implies $\sum_j x_j w_j = 0$, contradicting the linear independence of $\{w_1, \dots, w_m\}$.

With a tiny bit more effort, one can prove the existence of bases and that any two bases have the same cardinality:

Definition A basis of $V$ is a set that is both spanning and linearly independent.

Theorem 3 A finite dimensional vector space has a basis.

Sketch: Choose a spanning set of minimal cardinality and show that it is also linearly independent.

Theorem 4 Any two bases of a finite dimensional vector space have the same cardinality.

Proof. Follows from Theorem 2.

Definition The cardinality of a basis is called the dimension of $V$.

Theorem Let $V$ be a finite dimensional vector space with dimension $n$. Then any spanning set of $V$ has cardinality $\ge n$ and any linearly independent subset of $V$ has cardinality $\le n$.

Proof. Follows from Theorem 2.


So: assume that for every $n$ there exists a linearly independent subset $S_n$ of $V$ of cardinality $n$. Show that there exists a countable subset of $V$ that is linearly independent.

Consider $S'_n = S_1 \cup \ldots \cup S_n$. Note that $$S'_1\subset S'_2 \subset \ldots$$ and inside every $S'_n$ there exists a linearly independent subset of cardinality $n$.

Now start an inductive construction. Inside $S'_1$ there exists a linearly independent subset $S''_1$ of cardinality $1$. Suppose that we have constructed $S''_n$ inside $S'_n$ of cardinality $n$. Then we can enlarge it to $S''_{n+1}$ inside $S'_{n+1}$ of cardinality $n+1$. We obtain $$S''_1 \subset S''_2 \subset S''_3 \subset \ldots $$ every $S''_n$ is linearly independent of cardinality $n$. We get $S''=\cup_n S''_n$ linearly independent and countable.


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.