Instead of saying can be written as combination, you can say ; therefore, is lin. dep and can be written as a linear combination.
if I have a matrix A like this:
A =
when I do a reduced row echelon form rref in matlab it comes out like this:
Well I thought that when there is a pivot in a column, then that column is linearly independent, but I can see simply that column one in matrix A, is a combination of and and .
So why is linearly independent?
Here's another way to look at it:
The mistake is identifying column one in A with column one in the result handed to you by matlab.
A reordering has taken place.
Correction. The above is not stated well at all.
Reordering the columns amounts to a relabeling of the unknowns.
It is performed (conceptually at least) by applying a succession of permutations from the right to A.
The intention here (at least one) is to put the resulting matrix in a "nice" form, i.e., push the non-zero pivots
as far to the left as possible. Certainly, you can see this is going to happen to A simply by eye.
After the first several reduction steps, the second column will be all zeros. So your matlab is going to "fix it" by
relabeling the unknowns. Note also, that there has been a reordering of the rows (which is in effect simply an interchanging
of the equations). This process has no effect on the labeling of the unknowns, and could be carried out (again conceptually) by
a preliminary multiply of an appropriate permutation matrix to A, this time from the left.
This is why you cannot (in general) identify the columns of A with the corresponding columns of
the resulting matrix.
Also, this experience should suggest that it's a good idea on your part to know exactly what any software package
is going to do (or might do) to what you hand it.
There is a correction in my first post. You might take a look at it.
If you carry out the reduction process yourself you'll see that the linearly independent columns of A are 1,3 and 4.
This is not at all difficult; it can be done by eye.
The matrix returned from matlab tells you immediately that there are only three linearly independent columns in A.
But due to other matlab activities, the three nonzero pivots are sitting in columns 1,2 and 3.
The breakdown in the correspondence between the columns of A and the columns of the matlab result is clearly evident.
But how can column 1 be linearly independent if it can be built from columne 2, 3 and 4?
would c2 + c2 + c3 = c4
It seems I'm back to where I started?
If c1 can be built by a linear combination of the othter columns why is it linearly independent?
I'm missing something here.
You need to understand that it doesn't make sense to say that a specific vector is "independent" or "dependent". A set of vectors may be independent or dependent. What your row reduction shows is that (the set of) the first three vectors are independent but the first four are not.
First, you continue to refer to column vector 1, and ask how it can be linearly independent.
When you say that, what do you have in mind? It doesn't make sense to me.
Certainly, if you throw any nonzero vector into a set by itself, you can say it is a linearly independent set.
(After all, for nonzero vector v, what is the only scalar, c, that will make cv = 0?)
But I don't see what it has to do with this problem.
Second, I claimed that the reduction could be carried out by eye.
Well, I'm sorry to say that I didn't do it myself correctly, so where do I get off suggesting that anyone else should be able to do it correctly?
Certainly, column 2 does not become a column of all zeros, as I claimed.
Trying again (with more care), I see that columns 1, 2, 3, 5 and 6 form a linearly independent set. This is in accord with the matlab result. You can see the unit pivots clearly in those five columns.
So the rank of A is 5. Dimension of column space and row space is each 5; dimension of null space is 2; dimension of left-null space is 4.
Now I expect you will object again, since column 1 is still in this set.
And, you will ask again, how can it be linearly independent.
I think you can answer this question yourself by noting that the column 4 vector is not in the above set.
P.S. I don't think I've made any claim about knowing something (or anything) that you don't already know concerning the notion of linear independence.
Have a good day.