ENGINEERING SOFTWARE

EE 8207
LECTURE 3:

Solving Equations and Curve

Fitting

Instructor: E.Tarimo Thursday, December 22, 2022

2022/2023
 OUTLINE

 Linear Algebra

 Polynomials

 Optimization

Instructor: E.Tarimo Solving Equations 2022/2023

 OUTLINE

Linear Algebra
Polynomials

Optimization

Instructor: E.Tarimo Solving Equations 2022/2023

 SYSTEMS OF LINEAR EQUATIONS
 Given a system of linear equations
x + 2y - 3z = 5
-3x – y + z = -8
x–y+z=0
 Construct matrices so the system is described by Ax=b
» A = [1 2 -3;-3 -1 1;1 -1 1];
» b = [5;-8;0];
 And solve with a single line of code!
» x = A\b;
➢ x is a 3x1 vector containing the values of x, y, and z

Instructor: E.Tarimo Solving Equations 2022/2023

 WORKED EXAMPLE: LINEAR ALGEBRA
 Solve the following systems of equations:

Instructor: E.Tarimo Solving Equations 2022/2023

 MORE LINEAR ALGEBRA
 Given a matrix
» mat = [1 2 -3;-3 -1 1;1 -1 1];
 Calculate the rank of a matrix
» r = rank(mat);
➢the number of linearly independent rows or columns
 Calculate the determinant
» d = det(mat);
➢mat must be square; matrix invertible if det nonzero
 Get the matrix inverse
» E = inv(mat);
➢if an equation is of the form A*x = b with A as square
matrix, x = A\b is (mostly) the same as x = inv(A)*b
 Get the condition number
» c = cond (mat); (or its reciprocal: c = rcond(mat);)
➢if condition number is large, when solving A*x = b,
small errors in b can lead to large errors in x (optimal c==1)
Instructor: E.Tarimo Solving Equations 2022/2023
 OUTLINE

Linear Algebra
Polynomials

Optimization

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS
This part covers the following concepts:

 Polynomials in engineering.

 Matlab numeric tools for polynomials.

 Matlab symbolic tools for polynomial and other

analytic functions.

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS IN ENGINEERING
 Polynomials play an important role in the analysis
and design of engineering systems.

 In various fields, the mathematical model for

electrical, electronics, mechanical, biomedical,
chemical, industrial, civil, etc. systems can be
represented by an 𝑛th order linear differential equation
of the form:

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS IN ENGINEERING

Where
 𝑦(𝑡) represents the response (output) of the system at time
𝑡, and 𝑓(𝑡) is the forcing function (input) that drives the
system.
 The 𝑎𝑖 and 𝑏𝑖 coefficients are real constants.
 The solution for 𝑦(𝑡) is the sum of a natural response (𝑦𝑛𝑎𝑡)
and a forced response (𝑦𝑓)

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS IN ENGINEERING
 Analytical methods are available to systematically solve the
above differential equation.
 The forced response (generally) takes the form of the input
𝑓(𝑡), and its derivatives.
 On the other hand, the natural response has the analytical
form:

 𝑥𝑖 called natural frequency; in electrical engineering texts is

denoted by 𝑠𝑖
 𝑥𝑖 is one of the 𝑛 roots (here, assumed to be distinct) of the
characteristic polynomial.
Instructor: E.Tarimo Solving Equations 2022/2023
 POLYNOMIALS IN ENGINEERING
 The characteristic polynomial can be determined (by
inspection) from the left hand side of the differential
𝑑𝑛 𝑦
equation by substituting 𝑥𝑛 for 𝑑𝑡𝑛
and substituting 1

for 𝑦, as follows:

 The roots of 𝑝(𝑥) are the solutions to the equation

𝑝(𝑥) = 0

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS IN ENGINEERING
 Electric circuit model for a car’s ignition system

 Circuit model:

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS IN ENGINEERING
 Writing KVL around the loop,

di 1
v = iR + L +
dt C
idt 
 Differentiate wrt to time and re-arrange

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS IN ENGINEERING
 Mathematical model of the discharge cycle:

 Where 𝑖𝐿 is the circuit current.

 The characteristic polynomial is given by,

 The natural frequencies of the circuit are the solutions

to;

 and are given by,

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS
 Many functions can be well described by a high-order
polynomial

 MATLAB represents a polynomials by a vector of

coefficients

 If vector P describes a polynomial

 P is a vector of length N+1 describing an N-th order

polynomial

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIAL OPERATIONS
 To get the roots of a polynomial
» r = roots (P)
➢r is a vector of length N
 Can also get the polynomial from the roots
» P = poly (r)
➢r is a vector length N
 To evaluate a polynomial at a point
» y0 = polyval (P,x0)
➢x0 is a single value; y0 is a single value
 To evaluate a polynomial at many points
» y = polyval(P,x)
➢x is a vector; y is a vector of the same size

Instructor: E.Tarimo Solving Equations 2022/2023

 Polynomials
➢ Polynomials

❖ x2 + 2x – 7

❖ x4 + 3x3 -15x2 - 2x + 9

➢ In MATLAB polynomials are created by row vector i.e.

❖ s4 + 3s3 - 15s2 - 2s + 9

>>p = [ 1 3 -15 -2 9];

❖ 3x3 - 9

>>q = [3 0 0 -9]; %write the coefficients of every term

➢ Polynomial evaluation : polyval (y, x)

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS EVALUATION
Exp: Evaluate the value of polynomial y = 2x2 + 3x + 4 at x =
[1, -3]
>> y = [2 3 4];
>> x = 1;
>> value = polyval (y, x)
>> value = 9
>> x = -3;
>> value = polyval(y, x)
>> value = 13
Similarly
>> x = [1 -3];
>> value = polyval(y, s)
value = 9 13
OR
>> value = polyval(y,[1 -3])
Instructor: E.Tarimo Solving Equations 2022/2023
 ROOTS OF POLYNOMIALS
Roots of polynomials: roots (p)
>> p = [1 3 2]; % p = s2 + 3s + 2
>> r = roots (p)
r = -2 -1

Try this: find the roots of s4 + 3s3 - 15s2 - 2s + 9 = 0

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIALS MATHEMATICS
 Addition
>> a = [0 1 2 1]; % s2 + 2s + 1
>> b = [1 0 1 5]; % s3 + s + 1
>> c = a + b; % s3 + s2 + 3s + 6

 Subtraction
>> a = [3 0 0 2]; % s3 + 2
>> b = [0 0 1 7]; %s+7
>> c = b - a % -s3 + s + 5
c=
-3 0 1 5
Instructor: E.Tarimo Solving Equations 2022/2023
 POLYNOMIALS MATHEMATICS
 Multiplication : Multiplication is done by convolution
operation .
Syntax z = conv(x, y)
>> a = [1 2 ]; %s+2
>> b = [1 4 8 ]; % s2 + 4s + 8
>> c = conv (a, b) % s3 + 6s2 + 16s + 16
c=1 6 16 16
Try this: find the product of (s+3), (s+6) & (s+2). Hint: two at
a time

 Division : Division is done by deconvolution operation.

Syntax [z, r] = deconv(x, y)
Where
x = divident vector y = divisor vector
z = Quotients vector r = remainders vector
Instructor: E.Tarimo Solving Equations 2022/2023
 POLYNOMIALS MATHEMATICS
>> a = [1 6 16 16]; % a = s3 + 6s2 +16s + 16

>> b = [1 4 8]; % b = s2 + 4s + 8

>> [c, r] = deconv (a, b)

c=2

r=0 0 0 0

Try this: divide s2 - 1 by s + 1

Instructor: E.Tarimo Solving Equations 2022/2023

 FORMULATION OF POLYNOMIALS
 Making polynomial from given roots:
>> r = [-1 -2]; % Roots of polynomial are -1 & -2
>> p = poly(r); % p = s2 + 3s + 2
p=3 2

Instructor: E.Tarimo Solving Equations 2022/2023

 POLYNOMIAL FITTING
 MATLAB makes it very easy to fit polynomials to data
 Given data vectors X = [-1 0 2] and Y = [0 -1 3]
» p2 = polyfit(X,Y,2);
 Finds the best (least-squares sense) second-order
polynomial that fits the point(-1,0), (0,-1), and (2,3)
 see help polyfit for more information
» plot(X,Y);
» hold on;
» x = -3 : .01 : 3;
» plot(x, polyval(p2,x));
Instructor: E.Tarimo Solving Equations 2022/2023
 EXERCISE: POLYNOMIAL FITTING

 Evaluate y = x2 for x = -4:0.1:4.

 Add random noise to these samples. Use randn. Plot

the noisy signal with . markers

 Fit a 2nd degree polynomial to the noisy data

 Plot the fitted polynomial on the same plot, using the

same x values and a red line.

Instructor: E.Tarimo Solving Equations 2022/2023

 OUTLINE

Linear Algebra
Polynomials

Optimization

Instructor: E.Tarimo Solving Equations 2022/2023

 NONLINEAR ROOT FINDING
 Many real-world problems require us to solve f(x)=0
 Can use fzero to calculate roots for any arbitrary
function
 fzero needs a function passed to it.

 We will see this more and more as we delve into solving

equations.

Instructor: E.Tarimo Solving Equations 2022/2023

 MINIMIZING A FUNCTION
fminbnd: minimizing a function over a bounded
interval
» x=fminbnd(fun, x1, x2);
➢fun takes a scalar input and returns a scalar output
➢fun(x) will be the minimum of fun for x1 ≤ x ≤ x2

Instructor: E.Tarimo Solving Equations 2022/2023

 MINIMIZING A FUNCTION
fminsearch: unconstrained interval
» x=fminsearch(fun, x0)
➢finds the local minimum of fun starting at x = x0

 Maximize g(x) by minimizing f(x) = -g(x)

 Solutions may be local!

Instructor: E.Tarimo Solving Equations 2022/2023

 OPTIMIZATION TOOLBOX
 If you are familiar with optimization methods, use the
optimization toolbox
 Useful for larger, more structured optimization
problems
 Sample functions (see help for more info)

» linprog
➢linear programming using interior point methods
» quadprog
➢quadratic programming solver
» fmincon
➢constrained nonlinear optimization
Instructor: E.Tarimo Solving Equations 2022/2023
 EXERCISE: MIN-FINDING

 Find the minimum of the function

over the range–πtoπ. Use fminbnd.

 Plot the function on this range to check that this is the

minimum.

Instructor: E.Tarimo Solving Equations 2022/2023

 A WORD OF CAUTION
 MATLAB knows no fear!
 Give it a function, it optimizes/differentiates/
integrates
➢That’s great! It’s so powerful!
 Numerical techniques are powerful but not magic

 Beware of over trusting the solution!

➢You will get an answer, but it may not be what

you want
 Analytical forms may give more intuition

➢Symbolic Math Toolbox

Instructor: E.Tarimo Solving Equations 2022/2023