Coordinates and Basis
MA
2071
Purpose of these notes and assignment:
illustrate the concepts and skills
involved with coordinates and coordinate systems (also know as basis)
Definition:
given a vector space V, a basis (or coordinate system) is a set of vectors {v1, v2, v3,
. . . vk}
with two properties:
1) any vector
x can be written as a unique combination of the
vectors in V, or
x = c1v1 + c2v2+ . . . + ckvk equation (1)
2)
the basis vectors are independent
of one another. None can be written as a
combination of the others.
Remarks:
be careful to distinguish between vectors
and scalars. The vi s are vectors while the cis are scalars
(numbers). The latter are called coordinates.
Example 1: { (1,1) , (1,-1)
} is a basis for the vector
space R2. In the notation
above, v1 = (1 ,1) and v2 = (1,
-1). We know that it is a basis because
we can satisfy the two properties of the definition as follows:
a) any vector in R2 can be written as a combination of
them. Let x = (x , y)
be an arbitrary vector in R2. The equation
X = (x,y)
= c1v1 + c2v2 = c1(1
,1) +
c2 (1 , -1)
is really a system of
two equations and two unknowns. This is important to recognize. This can be better seen by looking at it
component by component:
x
= c1 +
c2
y
= c1 - c2
Does this system have
a solution?? It is a nonhomogeneous system so the possibility
exists that it doesn't. Since x and y
can literally be any numbers, we have
to be ready for all possibilities!!!
We solve it the same
way we have been solving systems throughout the course. Set up the augmented
matrix and reduce it to Final Form.
Here, the system in matrix form looks like:
which
means that the augmented matrix is
(
you have to get used to the
fact that x and y are treated as constants while the c's
are the variables.) Now you should be
able to look at this system and see that it will always have a solution, no
matter what x and y are. Why? How do you
end up in a "no solution" situation?
You would have to have, on the bottom row a situation like
where whatever is in
the place of & is not
0. Can this happen here? With one row
operation (adding -1 times Row1 to Row2 ) we
arrive at
which
should tell you it will have a unique solution. If it doesn't, we'll go another couple of steps and get Final Form:
This Final Form shows
that no matter what x and y are, we will have a unique solution. This means that any vector (x ,
y) can be written as a combination
of (1 , 1) and (1, -1). The first part of the definition has
been satisfied.
All this can be used
to summarize an Algorithm
for Finding Coordinates of a vector relative to a basis:
Problem:
find the coordinates of a vector x
relative to a basis {v1,
v2, v3, . . . vk}
step 1: set up the augmented matrix (v1 v2 v3 .
. . vk | x) This matrix has the basis vectors as
columns
on the left
and the given vector x on
the right
step 2: reduce it to Final Form (either with Gauss
step 3: Interpret the Final Form
step 4: Read them off the
matrix.
Note: if you had a row of 0's appear indicating
either no solution or infinitely many, then either you made an arithmetic
mistake or you did not really have a basis to begin with
Notation for future reference: the
coefficient portion with the basis vectors as columns will always be denoted by
P in this course.
That is,
P = (v1 v2 . . . vk)
In matrix terms, we are solving the problem Pc = x
(key equation!)
Example 1: find the coordinates of ( 6,3,6) relative
to { (1,0,-1) ,
(0,1,1), (4,0,5) }
step 1: set up the augmented matrix (P | x)
step 2: reduce it to Final
Form
step 3 & 4 :
interpret the Final Form
the coordinates are c1 = 2,
c2=3, c3
= 1
this means that (6,3,6) = 2(1,0,-1) + 3 (0,1,1) + 1( 4,0,5) which can be easily checked
Example 2: find the coordinates of ( 6,3,6) relative
to { (1,0,-1) ,
(0,1,1), (2,3,1) }
step 1: set up the augmented matrix (P | x)
step 2: reduce it to Final
Form
step 3 & 4 :
the bottom row tells us we have no solution. The vector (6,3,6) cannot be built out of the three given
vectors.
There is actually
more information here…. the third column in the Final Form is telling us that
the vector (2,3,1) is really 2 times the
first vector, (1,0,-1) plus 3 times the vector (0,1,1). ( the
reader is urged to check this detail)
The three vectors given were not independent; one
was a combination of the others.
For those who like a
geometric view of things, the three
vectors , {
(1,0,-1) , (0,1,1), (2,3,1)
} , all lie in the same plane.
The vector (6,3,6) does not lie in that plane. There is no way to have vectors in the plane add up to a vector which is
not in the plane!
Example 3: find the coordinates of ( 9,6,1) relative
to { (2,1,-1) ,
(3,1,2), (2,3,1), (9,6,1) }
step 1: set up the augmented matrix (P | x)
step 2: reduce it to Final Form
step 3: interpret the Final
Form
It
tells us the problem has a solution. It also tells us
there is an
arbitrary variable (c4). This is then telling us that
we don't
need the fourth vector
(9, 6 , 1). In fact the 4th
column of the Final Form
is telling
us that the vector (9, 6 , 1) is a combination of the first three,
specifically it
is 2 * (2
, 1, -1) + 1*(3, 1, 2) + 1*(2, 3, 1).
The
upshot of all this is we don't have a basis;
the vectors are not independent of one
another. (the fix to
this would be to throw out the 4th vector – then we would have a
vector).
Discussion: How do we know if a set of vectors forms a
basis for Rn??
none can be a
combination of the others. This means we
want the rank of the P matrix to be
the same as the number of vectors (k above)
we need to be able to
write any vector in Rn as a combination of
them which amounts to saying we must
always be able to solve the problem Pc = x. So we don't want to have a row of
zeroes appear in the Final Form of P because we can't assume there will be a 0
in the reduced form of x, the right hand side.
This means the Final Form of P must indicate a rank of n for it.
In order that we have no
arbitrary variables in the solution for the c's
because if we do then the vectors in the claimed basis won't be independent –
there will be repetition within them.
Thus there must be n columns in P – any
more than that will cause arbitrary variables.
Putting all this
together means that P must be nxn and it must have
rank of n. So its Final Form has only one possibility, the Identity matrix, In
! Anything else signifies that either
there is no solution to some problems or the vectors are not independent.
Summary: for a set {v1,
v2, v3, . . . vk} to be a basis for Rn we must have
any other result indicates we don't have a basis. There are either too
many vectors, too few vectors or some vector is a combination of the others.
If you do have a basis then to find the coordinates of x relative to
the basis, {v1, v2, v3,
. . . vk} ,
solve the problem
Pc = x (which symbolically is c = P-1x )
Its Final Form is guaranteed to
be (In | c ) The
coordinates will be on the right hand side as in Example 1
Homework for Tuesday
page 272: problems 1-3 (decide if a set is a basis or
not), problems 7 and 8 (finding
coordinates)
also: a plane through the origin has a basis
consisting of { (1, 0 , 4) , (0,1,9)
} On the plane, I am at
the location (3,5). Where am I in three dimensional space (R3)
??