Let $S$ be a subset of a vector space. I conjecture that $S$ is linearly dependent if and only if there exists a proper subset $S' \subset S$ such that $\operatorname{span}(S')=\operatorname{span}(S)$. Is this the case?

If this is the case, I see an analogy between this and the definition of infinite set as a set which is bijective with a proper subset of itself. Could someone elaborate on this if it is true?

  • 1
    $\begingroup$ Yes. Say you have $x \in S - S'.$ Then $x$ can be expressed as a finite linear combination of elements in $S'$ (since $\text{span} (S)=\text{span}(S')$ and $x\in S.$ $\endgroup$ – matt biesecker May 16 '15 at 4:23
  • $\begingroup$ There is an analogy between cardinality and dimension. Both measure size. Any infinite cardinal has a proper subset of the same size, and an infinite-dimensional vector space has a proper subspace with the same size. (Not sure where choice axioms come into play here.) $\endgroup$ – anon May 16 '15 at 4:31
  • $\begingroup$ @anon: There is no mention of cardinality here, just a strict subset. $\endgroup$ – copper.hat May 16 '15 at 4:40
  • $\begingroup$ @copper.hat OP does mention cardinality ("the definition of an infinite set as ..."). That is what I was responding to. $\endgroup$ – anon May 16 '15 at 4:56
  • $\begingroup$ What I was getting at is that with the definitions of both linear dependence and infinite cardinality we say that a set $X$ has property $P$ if there is a proper subset $X' \subset X$ such that $\{X, X'\}$ has property $Q$. In the former case, property $P$ is linear dependence and property $Q$ is both elements have the same span. In the latter case property $P$ is infinite cardinality and property $Q$ is there exists a bijection between both elements. $\endgroup$ – agcha May 16 '15 at 9:55

If $S$ is linearly dependent, then we can write $\sum_k \alpha_k x_k =0$ for some $\alpha_k$ not all zero and $x_k \in S$. Without loss of generality, we can assume that $\alpha_1 \neq 0$, and so we have $x_1 \in \operatorname{sp} \{x_2,...\}$. Then let $S' = S \setminus \{x_1\}$, which is a proper subset, and clearly $\operatorname{sp} S = \operatorname{sp} S'$.

For the other direction, if $\operatorname{sp} S = \operatorname{sp} S'$ with $S'$ being a proper subset of $S$, then pick $x \in S \setminus S'$.

If $x=0$, then $S' \cup \{0\}$ is linearly dependent and hence so is $S$, otherwise we have $x =\sum_k \alpha_k x_k$ for some $x_k \in S'$. Hence $\sum_k \alpha_k x_k + (-1) x = 0$ and so $S$ is linearly dependent.


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.