Definition. If $W_1 \subseteq W_2$ are subspaces of vector space $V$, then we say that $v_1, \ldots, v_m \in W_2$ are linearly indepdent mod $W_1$ if $$ a_1 v_1 + \cdots a_m v_m \in W_1 \text{ implies } a_i=0, \forall i$$

I'm having some trouble getting the intuition for this definition. I know that one way to think of it is "linear independence, but with $W_1$ as the zero vector," but this still isn't very clear for me.

I tried doing a toy example with $W_2 =\mathbb{R}^3$ and the $x$-, $y$-, and $z$-axes, where $W_1$ is the $x$-axis. I think that any two vectors will be linearly independent mod $W_1$ iff their projections onto the $yz$-plane are linearly independent in that subspace. (Can someone confirm this?)

In fact, it seems we can replace the $yz$-plane with any subspace $W'_1$ such that $W_2 = W_1 \oplus W'_1$ (which can be found by "extending" a basis for $W_1$ to one for $W_2$).

So in general, my "intuition" for linear independence mod $W_1$ is that the projections of $v_1, \ldots, v_m$ onto $W'_1$ are linearly independent. Does this hold?

Thanks in advance!

  • $\begingroup$ Yes, you are right. $\endgroup$ – Boris Novikov Aug 28 '13 at 6:48

You are right. But a better statement would be that linear independence mod $W_1$ means that the (canonical) images of the vectors in the quotient space $W_2/W_1$ are linearly independent.

Any complementary subspace $W_1'$ projects isomorphically ot the quotient $W_2/W_1$, which explains why it makes no difference whether projecting to $W_1'$ (parallel to$~W_1$) or to $W_2/W_1$. But the quotient space is defined without needing the choice of a complementary subspace.


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.