Linear Algebra, math homework help

User Generated

urael

Mathematics

Description

10 questions in total

Unformatted Attachment Preview

c Vikram Krishnamurthy 2017 1 Assignment 1: Due May 27 Please email your solutions (as a pdf file) to the TA Sujay Bhatt at sh2376@cornell.edu Make sure your email contains your name and student number. Notation: Below x0 denotes transpose of a matrix or vector x. 1. Define what is meant by a permutation matrix. What is their purpose? Show that the product of permutation matrices is a permutation matrix. Show that the inverse of a permutation matrix is the transpose of the matrix. 2. Markowitz Portfolio Optimization: This exercise requires some elementary optimization (from your calculus course) together with basic linear algebra. Consider the following optimization problem: min x0 Cx x∈Rn 0 such that g x = r Here x is a portfolio allocation vector, C is an n × n covariance matrix, g denotes the rate of returns of individual components and the scalar r denotes the return of the entire portfolio. It is assumed that g, r, C are known. Find the optimal x. 3. Consider the equation Ax = 0 where A is an m × n matrix and 0 denotes the vector with elements 0. For the cases m < n, m = n and m > n, discuss the solutions of this equation. (Disregard the trivial solution x = 0). 4. In abstract terms, explain how the LU decomposition of a matrix works. That is: starting with a matrix A and applying successive row-wise transformations using lower triangular matrices L1 , L2 , . . . show that one ends up with upper triangular matrix U . This abstraction is at the core of solving linear systems of equations by Gaussian elimination. 5. Is a linear system of equations well posed? This is crucial - because otherwise you can get nonsensical answers. Consider solving the system Ax = b where A is an m × n matrix. Show that: (a) Either a solution x exists (b) OR there exists a vector y such that A0 y = 0, b0 y 6= 0. That is either (a) holds or (b) holds; but both (a) and (b) cannot hold. A more general statement (which arises in linear programming optimization) is: Consider solving the system Ax = b where A is an m × n matrix and x is a non-negative vector. Show that: c Vikram Krishnamurthy 2017 2 (a) Either a solution x exists (b) OR there exists a vector y such that A0 y ≥ 0, b0 y < 0. That is either (a) holds or (b) holds; but both (a) and (b) cannot hold. 6. This exercise is meant to give you some familiarity of pivoting and the so called echelon form of matrices. Compute the rank of the following matrices (justify your method)     2 0 −1 3 6 2 4 1 1 2 2 ,  2 0 2 2 0 −1 1 −1 −1 2 7. The equation of an n-dimensional plane is c0 x = a, where c is the normal to the plane and a is a scalar. Determine the point in the plane nearest the origin and find the distance of this point to the origin. 8. What is the equation of a line in n-dimensional space? 9. Show that the following matrix rotates vectors counter-clockwise (for example in computer graphics)   cos θ − sin θ 0 R =  sin θ cos θ 0 0 0 1
Purchase answer to see full attachment
User generated content is uploaded by users for the purposes of learning and should be used following Studypool's honor code & terms of service.

Explanation & Answer

Hey there, the work took longer than expected but I have completed it. Please find the solution in a Word document attached. I also converted the same to PDF. If you need me to work on it further please let me know.

SOLUTIONS:
1. A permutation matrix is a square matrix resulting from permuting the rows of a square identity
matrix. Each row and column has exactly a single non-zero entry, which is 1, and zeros as the
remaining entries.
Use: The transpose of a permutation matrix can be used to factor tensors.
Proof that the product of permutation matrices is a permutation matrix.
Suppose you have two permutation matrices XX and YY.
If the non-zero entry in the first row of XX is in the ii-th column, then the first row of the
matrix X⋅YX⋅Y will be:
(∑nj=1X1, jYj, 1, ∑nj=1X1, jYj, 2… ∑nj=1X1, jYj, n)(∑j=1nX1, jYj, 1, ∑j=1nX1, jYj, 2… ∑j=1nX1, jYj,
n)
We know that:
X1, j=δi, jX1, j=δi, j
(Where δi, jδi, j is the Kronecker delta taking the value 11 when i=ji=j and 00 otherwise).
Hence the first row of X⋅YX⋅Y will be:
(X1, iYi, 1, X1, iYi, 2… X1,iYi,n)(X1,iYi,1,X1,iYi,2,…,X1,iYi,n)
Which since X1, i=1X1, i=1 will be:
(Yi,1,Yi,2,…,Yi,n).(Yi,1,Yi,2,…,Yi,n)
Therefore, first row of X⋅YX⋅Y corresponds to the ii-th row of YY. As we move on, you’ll get to see
that the mm-th row of X⋅YX⋅Y will be the ll-th row of YY, where the 11 in the mm-th row
of XX appears in the ll-th position.

Multiply two permutation matrices, therefore, is basically just rearranging the rows of one of
the matrices resulting in also in a permutation matrix.

Proof that the inverse of a permutation matrix is the transpose of the matrix.
A permutation matrix is the product of a sequence of interchange elementary matrices.
Supposing, P = E1E2 · · · En,
Where P denotes a permutation matrix,
Each Ei interchanges some two rows of the identity matrix. It’s obvious that Ei is symmetric,
So:
ETi = Ei. Also, we have E2i = I because applying the same interchange thrice returns to the
identity.

Therefore,
PPT = (E1E2 · · · En) (E1E2 · · · En) T = (E1E2 · · · En) (ETn ETn−1 · · · ET1)
= (E1E2 · · · En) (EnEn−1 · · · E1) = E1 · · · En−1E2nEn−1 · · · E1
= E1 · · · En−1En−1· · · E1 = · · · = I,
Which implies P−1 = PT

2. Step 1 Define Objectives
Minimize x’Cx x∈Rn such that g’x = r

x  x x

x ]T

1 2
n
The asset weight vector is given as
with
portfolio.
Let g be represented by p for individual asset return

xi as the weight of asset i in the

The expected return for each asset in the portfolio is expressed in the vector form

p  [ p1 p2

pn ]T with pi as the mean return of the asset i . The portfolio expected return is the
n

weighted average of individual asset return
given as a covariance matrix; C =

  11

V 

 n1

x p  pT x   xi pi
i 1

. The covariance of an asset is

 1n 



 nn 


, where i ,i is the variance of asset i and i , j is the covariance between
n

asset i and asset j . The portfolio variance is

n

 p  x Vx   xi x j i , j
2

T

i 1 j 1

Step 2
Use multi-objective optimization to solve for x:
The Lagrangian multiplier can be used to solve for optimization by:

L( x )   pT x   xTVx   (1T x  1)
Set

L
 0 , it follows that
x
x

To solve, substitute as follows:

1
(V 1 )( p   1)
2



1T V 1 p
2
 T 1
T
1
1 V 1 1 V 1

Let a1  1T V 1 1 and a2  1T V 1 p , both of which are scalars, equation 1.3 can be written as:



a2 2

a1 a1

The optimized solution for the portfolio weight vector x is

x* 

1 1
V 1 a2 2
V p
(  )1
2
2 a1 a1

3. Discuss possible solutions to Ax=0
1. A homogeneous system of m linear equations in n unknowns always has a non–trivial solution if
m < n. Suppose that m < n and that the coefficient matrix of the system is row–equivalent to B, a
matrix in reduced row–echelon form. Let r be the number of non–zero rows in B. Then r ≤ m < n
and hence n − r > 0 and so the number n − r of arbitrary unknowns is in fact positive. Taking one
of these unknowns to be 1 gives a non–trivial solution.
2. The amount of non-zero conditions in the echelon type of the framework is equivalent to the
quantity of leading entries.
3. The number of leading variables is ≤ min (m, n).
4. The quantity of free variables including the amount of leading variables = n which is the size of
columns of A.
5. The homogenous system Ax = 0 has non-trivial solutions if and only if there are free variables.
6. If there are more unknowns than equations, the homogeneous system always has nontrivial
solutions. Why? This is one of only few cases in which we can ...


Anonymous
I was having a hard time with this subject, and this was a great help.

Studypool
4.7
Trustpilot
4.5
Sitejabber
4.4

Related Tags