Let $W_1,W_2,W_3$ be linear subspaces of finitely generated vector space V. Assume that $W_1\cap W_2=W_1\cap W_3=W_2 \cap W_3=\{\theta \}$. Is $W_1+W_2+W_3$ is always direct sum?

I think that the answer is NO and looking for a contradiction.

Please help, thank you!

  • $\begingroup$ Is $\theta$ meant to be zero? $\endgroup$ – Sammy Black Feb 18 '14 at 20:01

I'm assuming that by $\{ \theta \}$ , you meant $\{ \mathbf 0 \}$, the subspace consisting of just the zero vector.

In that case, the answer is yes, the sum is a direct sum (i.e., each thing in $W_1 + W_2 + W_3$ can be written in a unique way as $a + b + c$, where $a \in W_1$, $b \in W_2$, and $c \in W_3$).

Quick proof:

Lemma: If $U$ and $S$ are subspaces of $V$ with $U \cap S = \{ \mathbf 0 \}$, then $U + S$ is a direct sum

Proof: Suppose $x \in U + S$, and $x = u + s= u' + s'$, where $u, u' \in U$ and $s, s' \in S$. The $u + s = u' + s'$, so $u-u' = s - s'$. Since $u - u' \in U$, and $s - s' \in S$, we have that $u-u' = s-s' \in U \cap S = \{ \mathbf 0 \}$. Hence $u - u' = s - s' = 0$. Thus the decomposition of $x$ is unique.

Now: apply this lemma to $U = W_2$ and $S = W_3$ to conclude that every element of $Z = W_2 + W_3$ can be expressed in only one way as a sum of elements of $W_2$ and $W_3$.

Then observe that $W_1 \cap Z = W_1 \cap (W_2 + W_3) = (W_1 \cap W_2) + (W_1 \cap W_3) = 0$, and apply the lemma to $U = W_1$ and $S = Z$. Then any element in $U + Z = W_1 + W_2 + W_3$ can be expressed uniquely as $a + d$, where $a \in W_1$ and $d \in Z$. And $d$ can be expressed uniquely as $d = b+c$, where $b \in W_2$ and $c \in W_3$, by the prior application of the lemma.

We're done!


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.