06.08.2013 Views

Linear Algebra (Math 311) – Final

Linear Algebra (Math 311) – Final

Linear Algebra (Math 311) – Final

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Linear</strong> <strong>Algebra</strong> (<strong>Math</strong> <strong>311</strong>) <strong>–</strong> <strong>Final</strong><br />

Problem 1. [30 points] Below you see a matrix A and its reduced rowechelon<br />

form B.<br />

⎛<br />

1<br />

⎜ 0<br />

A = ⎜ 2<br />

⎝ 2<br />

2 3<br />

1 3<br />

3 3<br />

3 −1<br />

−1<br />

1<br />

−3<br />

1<br />

⎞ ⎛<br />

2 1 0 0 −6<br />

−5⎟<br />

⎜<br />

⎟ ⎜0<br />

1 0 4<br />

9 ⎟ ∼ B = ⎜<br />

0 ⎠ ⎝<br />

75<br />

4<br />

−<br />

−1 0 7 −1 −3<br />

47<br />

0 0 1 −1<br />

0 0 0 0<br />

⎞<br />

⎟<br />

4 ⎟<br />

9 ⎟<br />

4 ⎟<br />

0 ⎠<br />

0 0 0 0 0<br />

In answering the following questions, provide at least a short, decisive argument.<br />

1. What is the rank of A?<br />

2. Find linearly independent rows of A.<br />

3. Find linearly independent columns of A.<br />

4. Find a basis for the orthogonal complement of the row space of A.<br />

5. Delete some rows and columns until you find a non<strong>–</strong>singular square<br />

matrix of maximal size. Justify that your minor is non<strong>–</strong>singular and of<br />

maximal size.<br />

6. How does the null space {x ∈ R 5 | Ax = 0} and the nullity relate to<br />

any of the above?<br />

Solution: 1. The rank of A is 3, as one may see from the number of<br />

leading 1’s in B.<br />

2. We need to find three linearly independent rows, as the dimension of<br />

the row<strong>–</strong>space is 3. We may start with the first two rows because they are<br />

linearly independent, none is a multiple of the other one. You may see that<br />

the third row is twice the first minus the second row. We cannot use the first<br />

three rows. We try to use the first, second, and fourth row, and to check<br />

linear independence, we look at the first three coordinates (if you want, you<br />

may also use all coordinates):<br />

⎛ ⎞ ⎛ ⎞ ⎛ ⎞<br />

1 2 3 1 2 3 1 2 3<br />

⎝0<br />

1 3 ⎠ ∼ ⎝0<br />

1 3 ⎠ ∼ ⎝0<br />

1 3 ⎠<br />

2 3 −1 0 −1 −7 0 0 −4<br />

1


From the last matrix in the computation we see that the matrix is nonsingular,<br />

and we conclude that the first, second, and fourth row are linearly<br />

independent.<br />

3. The first three columns of A are linearly independent because that is<br />

where B has the leading 1’s.<br />

4. The orthogonal complement to the row<strong>–</strong>space is the null space of the<br />

matrix. The last two ccordinates are free. We obtain a basis by setting either<br />

x4 = 1 and x5 = 0 or x4 = 0 and x5 = 1. With these choices we get the basis<br />

⎧⎛<br />

⎞<br />

6<br />

⎪⎨ ⎜<br />

⎜−4<br />

⎟<br />

⎜ 1 ⎟<br />

⎝<br />

⎪⎩<br />

1 ⎠<br />

0<br />

,<br />

⎛ ⎞⎫<br />

−75<br />

⎜ 47 ⎟⎪⎬<br />

⎟<br />

⎜ −9 ⎟<br />

⎝ 0 ⎠<br />

⎪⎭<br />

4<br />

5. We solved part 5. as part 2. of the problem. Delete the last two<br />

columns of the matrix, as well as row three and five. The result is a nonsingular<br />

matrix. There is not larger nonsingular minor, because is size will be<br />

the rank of the matrix, and that is 3.<br />

6. As stated in 4. the nullspace of the matrix is the orthogonal complement<br />

of its rowspace. The nullity is its dimension, and it is the number of<br />

columns minus the rank of the matrix. The nullity is 5 − 3 = 2.<br />

Problem 2. [10 points] Consider the vector space P2 of polynomials of degree<br />

at most 2 with the basis consisting of the vectors v1 = 1, v2 = 1 + t, and<br />

v3 = 1 + t + t 2 . Differentiation defines a linear map D : P2 → P2. Find the<br />

matrix of D with respect to the given basis.<br />

Solution: We apply D to the basis elements, making sure that the results<br />

are given in terms of the basis again. We find: D(v1) = 0, D(v2) = 1 = v1,<br />

and D(v2) = 1 + 2t = −1 + 2(1 + t) = −v1 + 2v2. Accordingly, the matrix<br />

for D is ⎛ ⎞<br />

0 1 −1<br />

⎝0<br />

0 2 ⎠<br />

0 0 0<br />

Problem 3. [20 points] Let P2 be as in Problem 2. Set<br />

S = {1 − t 2 } and T = {1 + t, 1 − t 2 , t + t 2 , 1 + 2t, 1 + t + t 2 }<br />

2


1. Is it possible to add elements to S until you have a basis of P2? Explain<br />

why! If possible, do it.<br />

2. Is it possible to omit elements of T until you have a basis of P2? Explain<br />

why! If possible, do it.<br />

Solution: Observe that P2 is of dimension 3, so we are looking for bases<br />

consisting of three elements.<br />

1. Because S is linearly independent, we can add elements, keeping the<br />

set linearly independent, until the set spans and we have a basis. Here is a<br />

set that works, obviously:<br />

{1 − t 2 , 1, t}<br />

2. Note that T spans P2 and contains a basis because<br />

1 = (1 + t + t 2 ) − (t + t 2 )<br />

t = (1 + 2t) − (1 + t)<br />

t 2 = (1 + t + t 2 ) − (1 + t)<br />

To cut down T to a basis, we use the first element, it is nonzero. We keep<br />

the second element, because the first two elements are linearly independent,<br />

none of the elements is a multiple of the other one. Note that the third<br />

element is the difference of the first two, so discard it. It is easy to see that<br />

the fourth element is independent of the first two, so we keep it. By now we<br />

have three linearly independent elements selected from T , and they form a<br />

basis:<br />

{1 + t, 1 − t 2 , 1 + 2t, }<br />

Problem 4. [15 points] Find the determinants of the matrices<br />

⎛ ⎞<br />

⎛ ⎞<br />

1 2 3 1<br />

3 0 1<br />

⎜<br />

A = ⎝1<br />

0 4⎠<br />

and B = ⎜0<br />

1 2 5 ⎟<br />

⎝1<br />

2 1 1⎠<br />

2 1 5<br />

0 7 1 2<br />

Solution: Using the standard formula for 3 × 3 matrices, we find<br />

det A = 1 − 12 = −11.<br />

We may apply elementary row operations (recalling how they affect the determinate)<br />

to bring the matrix into triangular form, and then get the determinante<br />

as the product of the diagonal elements. Here is one way, without<br />

3


additional explanations for the steps. We use the notation | · | to indicate<br />

that we find the determinant of the enclosed expression.<br />

<br />

<br />

1<br />

2 3 1<br />

<br />

1<br />

0 3 1<br />

<br />

0<br />

0 2 0<br />

<br />

<br />

<br />

det B = 0<br />

1 2 5<br />

<br />

<br />

<br />

1<br />

2 1 1<br />

= 0<br />

1 2 5<br />

<br />

<br />

<br />

1<br />

0 1 1<br />

= 0<br />

1 2 5<br />

0<br />

2 0<br />

<br />

<br />

<br />

1<br />

0 1 1<br />

= <br />

1<br />

2 5<br />

= 66<br />

<br />

0<br />

7 1 2<br />

0<br />

7 1 2<br />

0<br />

7 1 2<br />

7 1 2<br />

<br />

6 3<br />

Problem 5. [15 points] Let A = .<br />

−2 1<br />

1. Find the eigen vectors and values for A.<br />

2. If possible, find a matrix P so that P −1 AP is diagonal. If A cannot be<br />

diagonalized, explain why.<br />

Solution: We calculate:<br />

<br />

6 − λ 3<br />

det<br />

= (6 − λ)(1 − λ) + 6 = λ<br />

−2 1 − λ<br />

2 − 7λ + 12 = (λ − 3)(λ − 4).<br />

<br />

<br />

1<br />

3<br />

An eigen vector for λ = 3 is , and an eigen vector for λ = 4 is .<br />

−1<br />

−2<br />

As matrix P we use the matrix with the eigen vectors as its columns:<br />

<br />

1 3<br />

P = .<br />

−1 −2<br />

Then<br />

P −1 AP =<br />

−2 −3<br />

1 1<br />

provides the diagonaliztion.<br />

6 3<br />

−2 1<br />

<br />

1 3<br />

=<br />

−1 −2<br />

<br />

3 0<br />

0 4<br />

Problem 6. [25 points] Let V be a vector space, and 〈·, ·〉 an inner product<br />

on V . The inner product defines a norm on V .<br />

1. Define the concept of a norm on a vector space.<br />

2. Tell, by which formula an inner product defines a norm.<br />

3. Prove that the formula that you gave in the previous item does indeed<br />

give you a norm.<br />

4


Solution: A norm on a vector space is a map || · || : V → R, so that<br />

1. ||x|| ≥ 0 for all x ∈ V , and ||x|| = 0 if and only if x = 0.<br />

2. ||ax|| = |a| · ||x|| for all x ∈ V and a ∈ R.<br />

3. ||x + y|| ≤ ||x|| + ||y|| for all x and y in V<br />

An inner product defines a norm by setting<br />

||x|| = 〈x, x〉<br />

We verify that this expression defines a norm. By definition of the square<br />

root, ||x|| = 〈x, x〉 ≥ 0. From the positive definiteness of the inner product<br />

we deduce that<br />

||x|| = 0 ⇐⇒ 〈x, x〉 = 0 ⇐⇒ 〈x, x〉 = 0 ⇐⇒ x = 0.<br />

Using the linearity properties of the inner product and basic rules of algebra<br />

we find:<br />

<strong>Final</strong>ly,<br />

||ax|| = 〈ax, ax〉 = a 2 〈x, x〉 = |a| 〈x, x〉 = |a| · ||x||.<br />

||x + y|| 2 = 〈x + y, x + y〉 = 〈x, x〉 + 2〈x, y〉 + 〈y, y〉<br />

≤ ||x|| 2 + 2||x|| · ||y|| + ||y|| 2 = (||x|| + ||y||) 2<br />

The inequality is a direct consequence of Cauchy<strong>–</strong>Schwarz’ inequality. The<br />

triangle inequality is obtained by taking square roots of the outermost terms.<br />

Problem 7. [20 points] Consider a vector space V of dimension 2 with basis<br />

{x, y}. On V , define an inner product 〈·, ·〉 by setting 〈x, x〉 = 〈y, y〉 = 3 and<br />

〈x, y〉 = 1. We denote the associated norm by || · ||.<br />

1. Find ||x − y|| and the cosine of the angle between x and y.<br />

2. Orthonormalize the basis.<br />

Solution: We calculate:<br />

||x − y|| = 〈x − y, x − y〉 = √ 3 + 3 − 2 = 2.<br />

5


Let α denote the angle between x and y, then<br />

cosα =<br />

〈x, y〉 1<br />

=<br />

||x|| · ||y|| 3 .<br />

Let us first orthogonalize the vectors, resulting in<br />

v1 = x and v2 = y −<br />

〈y, x〉 1<br />

x = y −<br />

〈x, x〉 3 x.<br />

The norms of these vectors are<br />

||v1|| = √ <br />

3 and ||v2|| = 〈y − 1<br />

<br />

1<br />

x, y − x〉 = 3 +<br />

3 3 1<br />

<br />

1 2<br />

− 2 · = 2<br />

3 3 3 .<br />

Normalizing the vectors v1 and v2 we find the orthonormal basis consisting<br />

of<br />

u1 = 1<br />

√ x and u2 =<br />

3 1<br />

<br />

3<br />

y −<br />

2 2<br />

1<br />

3 x<br />

<br />

.<br />

Problem 8. [15 points] Show, if the columns of C are linearly independent,<br />

then C t C is invertible.<br />

Solution: Suppose the size of C is m × n. Let x ∈ R n . Then Cx is a<br />

linear combination of the columns of C, and because the columns of C are<br />

linearly independent, the only solution of Cx = 0 is the trivial solution x = 0.<br />

Suppose C t C is singular, then there exists some non<strong>–</strong>zero vector x ∈ R n so<br />

that C t Cx = 0. It follows that x t C t Cx = (Cx) t Cx = 0, and this means that<br />

Cx = 0. This is a contradiction, and it follows that C t C is non<strong>–</strong>singular,<br />

hence invertible.<br />

6

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!