Linear algebra - dimension and intersection

Click For Summary

Homework Help Overview

The discussion revolves around a linear algebra problem concerning the codimension of subspaces within a finite-dimensional vector space. The original poster seeks to prove a relationship involving the codimension of the intersection of multiple subspaces.

Discussion Character

  • Exploratory, Assumption checking, Mathematical reasoning

Approaches and Questions Raised

  • The original poster expresses confusion and seeks guidance on proving the statement by induction, questioning the use of certain properties of codimension and dimension.
  • Some participants question the validity of the original statement by providing counterexamples and exploring specific cases, such as when the intersection is trivial.
  • Others suggest reconsidering examples and provide additional scenarios to clarify the relationship between dimensions and codimensions.

Discussion Status

The discussion is ongoing, with participants exploring various interpretations and examples. Some have offered insights into the structure of the proof, while others are still grappling with the implications of the codimension relationship.

Contextual Notes

Participants are navigating through specific cases and examples, including dimensions of subspaces and their intersections, while also addressing potential misconceptions about the properties of codimension.

zwingtip
Messages
20
Reaction score
0

Homework Statement


let V be a finite dimensional vector space of dimension n. For W \leq V define the codimension of W in V to be codim(W) = dim(V) - dim(W). Let W_i, 1 \leq i \leq r be subspaces of V and S = \cap_{i=1}^{r}W_i. Prove:

codim(S) \leq \sum_{i=1}^{r} codim(W_i)

Homework Equations


dim(U + V) = dim(U) + dim(V) - dim(U \cap V)<br /> \sum_{i=1}^{r} codim(W_i) = \sum_{i=1}^{r} (n - dim(W_i)

The Attempt at a Solution



I'm completely lost here. I know I need to prove this by induction. Any tips to point me in the right direction? Can I use the fact that codim(S) = n - dim(\cap_{i=1}^{r} W_i and dim(U \cap V) &lt; dim(U) + dim(V) assuming dim(U + V) \neq 0?
 
Physics news on Phys.org
only had a quick look, but not convinced this is true..

take the case when say dim(W1) = dim(W2) = 1, in Rn, then codim(W1) = codim(W2) = n-1

now say the intersection S is only the zero vector then dim(S) = 0 and codim(S) = n >n-1> Can up with oher higherdimensional arguments similarly
 
Still doesn't make sense to me. What if the intersection contains more than just the zero vector? Sorry for being stupid. Help?
 
actually i think my exampel was wrong as
codim(S) = n < codim(W1) + codim(W1) = 2n-2, which works for for n>=3

now take R3, say you have 2 planes P1 & P2 that intersect in a line L, then
dim(R3) = 3
dim(P1) = dim(P2) = 3, then codim(P1) = codim(P2) = 2
dim(L) = 1, codim(L) = 2

so
codim(L) = 2 <= codim(P1) + codim(P2) = 1 + 1 = 2
 
I think I've got it with an induction proof from the base cases r=1, r=2 and the inductive step being that it holds for r>2 then manipulating it algebraically for an entire page of 5mm graph paper. THanks.
 

Similar threads

Replies
1
Views
1K
Replies
15
Views
2K
Replies
34
Views
3K
  • · Replies 10 ·
Replies
10
Views
3K
Replies
8
Views
2K
  • · Replies 2 ·
Replies
2
Views
4K
Replies
1
Views
3K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K