# Homework Help: Proving a Linear System

1. Apr 14, 2010

### annoymage

1. The problem statement, all variables and given/known data

let u1 , u2 , u3 $$\in$$ R3 be such that

Au1T = u1T , Au2T = $$\frac{1}{2}$$u2T , Au3T = $$\frac{1}{3}$$u3T

Suppose u1 , u2 , u3 $$\neq$$ 0

Show that

u1 , u2 , u3 are linearly independent

2. Relevant equations

a1u1 + a2u2 + a3u3 = 0

a1 = a2 = a3 = 0 for it to become linearly independent

3. The attempt at a solution

AuiT = uiT iff ui = ui AT for all i=1,2,3

implies

AT( u1 + $$\frac{1}{2}$$ u2 + $$\frac{1}{3}$$ u3) = 0

how to proof this is linear independent?
gimme clue please T_T

Last edited: Apr 14, 2010
2. Apr 14, 2010

### Staff: Mentor

There's more to the definition of linear independence. For example, there is always a solution for a1, a2, a3 in the equation above no matter what vectors you have. It doesn't matter whether they're independent or dependent.
Here's a clue - what is the definition of linear independence?

3. Apr 14, 2010

### annoymage

hmm, i still in "i dont know" owho..

im not good in english, so what you mean is that

a1 a2 a3 always have a solution because Av=0 , the v is always have solution, for all A and 0 inside matrix mxn. right?

and what i know the definition of linear independece is what i state

a1u1 + a2u2 + a3u3 = 0

a1 = a2 = a3 = 0 for it to become linearly independent.

hehe, can give me other clue about other definition.. clueless right now, sorry T_T

4. Apr 14, 2010

### Staff: Mentor

There is more to the definition of linear independence, and this is a subtlety that escapes very many students.

Here are two examples that illustrate what I'm talking about.

1. u = <1, 0, 0>, v = <0, 1, 0>, w = <1, 1, 0>
Consider the equation au + bv + cw = 0. Clearly a = b = c = 0 is a solution. Does that make these vectors linearly independent?

2. u = <1, 0, 0>, v = <0, 1, 0>, w = <1, 0, 1>
Consider the equation au + bv + cw = 0. Clearly a = b = c = 0 is a solution. Does that make these vectors linearly independent?

5. Apr 14, 2010

### Staff: Mentor

Your English is OK, and is probably better than my attempts to speak your language. We're not talking about matrices here; we're just talking about what it means for a set of vectors to be linearly dependent or linearly independent.

For vectors u, v, and w, and constants c1, c2, and c3, the equation c1u + c2v + c3w = 0 always has a solution for the constants c1, c2, and c3. No matter what the vectors u, v, and w are. It doesn't matter whether they're linearly dependent or linearly independent. That equation always has a solution. So there must be something else that allows us to make a distinction between linearly independent and linearly dependent vectors. That's what I'm trying to get you to realize.

6. Apr 14, 2010

### irycio

Oh c'mon, the definition is very simple. Since i'm not familiar with typing the mathematic formulas yet, let me write it for 3 vectors, but it's easy to generalise:
x,y,z are linearly independent iff:
(ax+by+cz=0) =>( a=b=c=0)
In Mark44's example 1 there is another solution: a=b=1, c=(-1). However, there is no other solution for vectors shown in (2), hence they are independent.

Nothing subtle :P

E: Well, apparently Mark44 was faster in answering his own post :P.

7. Apr 14, 2010

### annoymage

no, mark44's is correct. i realised something wait let me refrase my definition. its hard to use this symbols

8. Apr 14, 2010

### irycio

I didtn't write he was incorrect. I just didn't agree with his calling the deffinition of linear independence "subtle".

9. Apr 14, 2010

### annoymage

ooooooooooo, if that, hmm like at question 1st

1. (a,b,c) = (-t,-t,t) for all t (real number)

proving that this is linear dependent
but (0,0,0) is also a solution

hmm, i get it what you are trying to say..

let me refrase my definition

a1u1 + a2u2 + a3u3 = 0

for at least ai $$\neq$$ 0 , i=1,2,3

then it is linearly dependent

otherwise, it is linear independent

am i right?

10. Apr 14, 2010

### annoymage

owho, sorry irycio ;P

11. Apr 14, 2010

### annoymage

but, how do i solve the question then? should i assume that the statement is linear independent and make contradiction?

12. Apr 14, 2010

### Staff: Mentor

Nope, that's not it. Let's look at my first example.
u = <1, 0, 0>, v = <0, 1, 0>, w = <1, 1, 0>
I can see by inspection that au + bv + cz = 0 has a solution a = b = c = 0. Does that make vectors u, v, and w linearly independent? The answer is no.
I disagree. It's very subtle. Students always focus on the equation (e.g. for 3 vectors) au + bv + cz = 0 and completely miss the subtlety of the distinction between linear independence and linear dependence.

As I said before the equation always has a solution regardless of whether the vectors are linearly dependent or linearly independent. The subtlety comes with whether the solution for the constants is unique (a = b = c = 0) or not unique; i.e., that there are nonzero solutions for the constants. If the solution is unique, the vectors are linearly independent. If the solution is not unique, the vectors are linearly dependent.

13. Apr 14, 2010

### Staff: Mentor

This is a different question from the first one you posted.

(a, b, c) is a vector, and a single vector is linearly independent as long as it's not identically the zero vector. If u = <a, b, c> = <-t, -t, t> then the equation c1 u = 0 has only a single solution for arbitrary values of t, c1 = 0.

I'm not sure about what you're asking, though.

14. Apr 14, 2010

### irycio

Oh of course it always does have the zero-solution, but in my opinion it just is not definition.

What I wrote in thefragment which You first quoted means literally, that the vectors are independent ONLY IF zero is the ONLY solution. Your first example has another one (1,1,-1) and hence those vectors are NOT independent.

I believe we totally agree on the linear dependencies between vectors. It's just the definition I was told concentrated more on the uniqueness of the solution than on the equation SUM=0 itself (as important as it is, though .

15. Apr 14, 2010

### Staff: Mentor

I don't understand what you're saying here.
That's not what you wrote, which was
This is the part after iff. In my counterexample I supplied an equation ax + by + cz = 0 that is obviously true, and came up with a solution a=b=c=0 (which are true statements), so the implication is true. The problem is that according to your definition the vectors x, y, and z are linearly independent when in fact they are not.
Right. So your definition is defective. I recognize that you understand the difference between linear dependence/independence. All I'm saying is that your definition doesn't work. It should say that the equation ax + by + cz = 0 has exactly one solution for the constants a, b, and c. That is exactly the subtlety I've been talking about, and the one that has escaped many generations of linear algebra students.
That is the correct definition - the uniqueness of the solution. The fact that a linear combination of vectors adds up to the zero vector is completely immaterial.

16. Apr 14, 2010

### irycio

You're right, my definition "(ax+by+cz=0)..." didn't mean what I wanted it to mean.

17. Apr 14, 2010

### annoymage

im still twisting my brain, how to do this,

hmm,

u1(AT - I) = 0

u1$$\neq$$ 0 => (AT - I) = 0

am I in the right path? ;P

18. Apr 14, 2010

### annoymage

irycio, can you help me ? T_T

mark44, are you off9 ?

someone please , ngahaha im begging you

19. Apr 14, 2010

### Staff: Mentor

Do you know about eigenvalues, eigenvectors, and similar matrices? For this problem you have Au1 = u1, Au2 = (1/2)u2, and Au3 = (1/3)u3, where u1, u2, and u3 are not zero vectors.

You need to show that u1, u2, and u3 are linearly independent. I.e., show that the only solution of the equation c1u1 + c2u2 + c3u3 = 0 is c1 = c2 = c3 = 0.

To do this, you need to be able to say something about the matrix A.

20. Apr 14, 2010

### annoymage

i see i see, let me do some reading, n then try solve it, after i go bath. owhoho..
my body all sticky, later if i get the proof, i'll post to let you check it.