• Major: Chemical

Engineering CHAPTER 5 (Part 2)

• Subject: Chemical
Engineering
Mathematics 2 THE SOLUTION
METHODS for
• Author:
• Prof. Andri Cahyo
Kumoro,PhD
SIMULTANEOUS NON
LINEAR EQUATIONS
• Dept. of Chemical
Engineering
• Diponegoro University
• 2013
 Systems of Nonlinear Equations
• Locate the roots of a set of simultaneous nonlinear equations:

f1 ( x1 , x2 , x3 ,  , xn )  0
f 2 ( x1 , x2 , x3 , , xn )  0

f n ( x1 , x2 , x3 , , xn )  0
Example :
x 2  xy  10  f ( x, y )  x 2  xy  10  0
y  3xy2  57  f ( x, y )  y  3 xy2  57  0
2
 Newton-Raphson Method
• One nonlinear equation (Ch.3)

f ( x i  1 )  f ( x i )  ( x i  1  x i ) f ( x i )
f ( xi )
xi1  xi 
f ( x i )

• Two nonlinear equations (Taylor-series)

f1,i f1,i
f1,i 1  f1,i  ( x1,i 1  x1,i )  ( x2,i 1  x2,i )
x1 x2
f 2,i f 2,i
f 2,i 1  f 2,i  ( x1,i 1  x1,i )  ( x2,i 1  x2,i )
x1 x2
 Solution: Intersection of Two Curves
• Two roots: f1(x1,x2) = 0 , f2 (x1,x2) = 0

 f1,i f1,i f1,i f1,i

 x 1, i  1  x 2 , i  1   f 1, i  x 1 , i  x2 ,i
 f 1, i  1  0  x1 x 2 x1 x 2
  
 f 2 ,i 1  0  f 2 ,i f 2 ,i f 2 ,i f 2 ,i
• Alternatively x  x   f 2 , i  x 1, i  x2 ,i
 x1 1,i  1 x 2 2 ,i 1 x1 x 2

 f 1 , i f 1 , i 
 x x 2   x 1 , i  1  x 1 , i   f 1, i 
 1    
 f 2 , i f 2 , i  x 2 , i  1  x 2 , i   f 2. i 
 
 x 1 x 2 

Jacobian matrix
[J]{x} = { f }
 NONLINEAR SYSTEMS
• Simultaneous nonlinear equations
f1 ( x1 , x 2 , x 3 ,  , xn )  0
f 2 ( x1 , x 2 , x 3 ,  , xn )  0
f 3 ( x1 , x 2 , x 3 ,  , xn )  0

• Example fn ( x1 , x 2 , x 3 ,  , xn )  0

x12  2 x 22  4 x 3  7 f 1( x1, x 2)  x12  2 x22  4 x3  7  0

2 x1 x 2  5 x 2  7 x 2 x 32  8 f 2( x1, x 2)  2 x1 x2  5 x2  7 x2 x32  8  0
5 x1  2 x 2  x 32  4 f 3( x1, x 2)  5 x1  2 x2  x32  4  0
 Three or more nonlinear equations (Taylor-series)
f1 ( x1 , x 2 , x 3 ,  , xn )  0
f 2 ( x1 , x 2 , x 3 ,  , xn )  0
f 3 ( x1 , x 2 , x 3 ,  , xn )  0

fn ( x1 , x 2 , x 3 ,  , xn )  0

 f1 f1 f1 

f1 ( x1 , x2 , x3 , , xn )   1
f f1 f1 
 x   x1     x1 
x2 xn     
 x x2 
xn  
 1   1 
 f 2 f 2 f 2      f f 2 f 2   
  x2   f 2 ( x1 , x2 , x3 , , xn )  2   x2 
 x1 x2 xn           x1 x2 
xn   
               
 f f n f n         f f n f n    
 n        n
  x 
xn  i  n i 1  n 1 2 3 n i xn  i  n i
x f ( x , x , x , , x )
 x1 x2  x1 x2

[ J i ]{ X i 1}  {Fi }  [ J i ]{ X i }
6
Solve this set of linear equations at each iteration
Three or more nonlinear equations (Taylor-series)

 f1 f1 f1  

 x    x1   x1    f1 ( x1 , x2 , x3 , , xn ) 
x2 xn      
 1  
 f 2 f 2 f 2       
 
 2 x   x  
 2 1 2 3
f ( x , x , x ,  , x n 
)
 x1 x2 xn      2
 
  
         
 f f n f n        
 n     xn 
 
 xn    f n ( x1 , x2 , x3 , , xn )
 x1 x2 xn  i  i 1 i   i i

[J] {x} { f }
Jacobian matrix

[J] {x} = { f }
7
 Solution of Nonlinear Systems of Equations

Solve this set of linear equations at each iteration: [ J i ]{ X i 1}  {Fi }  [ J i ]{ X i }

Rearrange: [ J i ]{ X i 1  X i }  {Fi }
[ J i ]{X i 1}  {Fi }

 f1 f1 f1 

 x    x1   f1 ( x1 , x2 , x3 ,, xn ) 
x2 xn    
 1 
 f 2 f 2 f 2     
 x2   f 2 ( x1 , x2 , x3 ,, xn )
 x1 x2 xn        
        
 f f n f n       
 n   xn   f n ( x1 , x2 , x3 ,, xn )
 x1 x2 xn  i i 1 ii

[J]{x} = { f }
8
Example

For the case of more

than two unknown
variables

The general equations to be

applied in iterations is:
Now consider that we have n equations so that j = 1 to n.

Just as in the single variable case, we want our next

iteration to take us to the root so we assume that

is the residual factor at the kth iteration and is defined as:

is the Jacobian matrix at the kth iteration and is defined as:

With :

New guess for x is:

SOLUTION PROCEDURES:
 Example 1

Step One. Make an initial guess. One of the solutions looks to be at

x1= 1.0 and x2 = 2.0 .

Step Two. Using that initial guess, calculate the residual and the
Jacobian.

Step Three. Solve

 x(2) = x(1)+ x(1)

0.8895
Example 2
 Important Notes on the Solution of
Nonlinear Systems of Equations
• n non-linear equations with n unknowns to be solved:
•
• Need to solve this set of Linear Equations [ J i ]{X i 1}  {Fi }

in each and every iterative solution of the original Non-Linear

system of equations.

• For n unknowns, the size of the Jacobian matrix is n2 and the time it
takes to solve the linear system of equations (one iteration of the
non-linear system) is proportional to O(n3)

• There is no guarantee for the convergence of the non-linear

system. There may also be slow convergence.

• In summary, finding the solution set for the Non-Linear system of

equations is an extremely compute-intensive task.
15
 Two Nonlinear Functions

Circle x2 + y2 = r2

 f(x,y)=0
g(x,y)=0 Solution
depends on
initial
guesses

Ellipse (x/a)2 + (y/b)2 = 1

function [x1,y1,x2,y2]=curves(r,a,b,theta) » r=2; a=3; b=1; theta=0:0.02*pi:2*pi;
% Circle with radius A » [x1,y1,x2,y2]=curves(2,3,1,theta);
x1=r*cos(theta); y1=r*sin(theta); » H=plot(x1,y1,'r',x2,y2,'b');
% Ellipse with major and minor axes (a,b) » set(H,'LineWidth',3.0); axis equal;
x2=a*cos(theta); y2=b*sin(theta);
 Intersection of Two Curves

• Intersection of a circle and a parabola

 f ( x, y)  x 2  y 2  1  0  f1 ( x1 , x 2 )  x12  x 22  1  0
 or 
 g ( x , y )  x 2
 y  0 
 2 1 2
f ( x , x )  x 1  x2  0
2
 Intersection of Two Curves

 f1 f1 
 f1 ( x1 , x 2 )  x12  x 22  1  x x 2   2 x1 2 x2 
  1 
 f 2 ( x1 , x 2 )  x12  x 2  f 2 f 2   2 x1  1 
 x1 x 2 

 2 x 1, i 2 x2 ,i   x1,i    f1,i 
[ J ]{ x}      
 2 x 1, i  1 x2 ,i   f 2 ,i 
Solve for xnew

 x1,i   x 1 , i  1  x 1, i 
x    x x 
new old

x2 ,i   y2 , i  1  x 2 , i 
function [x,y,z]=ellipse(a,b,c,imax,jmax)
for i=1:imax+1 » a=3; b=2; c=1; imax=50; jmax=50;
theta=2*pi*(i-1)/imax; » [x,y,z]=ellipse(a,b,c,imax,jmax);
for j=1:jmax+1
phi=2*pi*(j-1)/jmax; » surf(x,y,z); axis equal;
x(i,j)=a*cos(theta); » print -djpeg100 ellipse.jpg
y(i,j)=b*sin(theta)*cos(phi);
z(i,j)=c*sin(theta)*sin(phi);
end
end
function
[x,y,z]=parabola(a,b,c,imax,jmax)
% z=a*x^2+b*y^2+c
for i=1:imax+1 Parabolic or
xi=-2+4*(i-1)/imax; Hyperbolic Surfaces
for j=1:jmax
eta=-2+4*(j-1)/jmax;
x(i,j)=xi;
y(i,j)=eta;
z(i,j)=a*x(i,j)^2+b*y(i,j)^2+c;
end
end

» a=0.5; b=-1; c=1; imax=50; jmax=50;

» [x2,y2,z2]=parabola(a,b,c,imax,jmax);
» surf(x2,y2,z2); axis equal;
» a=3; b=2; c=1; imax=50; jmax=50;
» [x1,y1,z1]=ellipse(a,b,c,imax,jmax);
» surf(x1,y1,z1); hold on; surf(x2,y2,z2); axis equal;
» print -djpeg100 parabola.jpg
Intersection of two nonlinear surfaces

Infinite many
solutions
 Intersection of three nonlinear surfaces
» [x1,y1,z1]=ellipse(2,2,2,50,50);
» [x2,y2,z2]=ellipse(3,2,1,50,50);
» [x3,y3,z3]=parabola(-0.5,0.5,-0.5,30,30);
» surf(x1,y1,z1); hold on; surf(x2,y2,z2); surf(x3,y3,z3);
» axis equal; view (-30,60);
» print -djpeg100 surf3.jpg
function x = Newton_sys(F, JF, x0, tol, maxit)

% Solve the nonlinear system F(x) = 0 using Newton's method

% Vectors x and x0 are row vectors (for display purposes)
% function F returns a column vector, [f1(x), ..fn(x)]'
% stop if norm of change in solution vector is less than tol
% solve JF(x) y = - F(x) using Matlab's "backslash operator"
% y = -feval(JF, xold) \ feval(F, xold);
% the next approximate solution is x_new = xold + y';

xold = x0;
disp([0 xold ]);
iter = 1;
while (iter <= maxit)
y= - feval(JF, xold) \ feval(F, xold);
xnew = xold + y';
dif = norm(xnew - xold);
disp([iter xnew dif]);
if dif <= tol
x = xnew;
disp('Newton method has converged')
return;
else
xold = xnew;
end
iter = iter + 1;
end
disp('Newton method did not converge')
x=xnew;
 Intersection of Circle and Parabola
function f = ex11_1(x) function df = ex11_1_j(x)

f = [(x(1)^2 + x(2)^2 -1) df = [2*x(1) 2*x(2)

(x(1)^2 - x(2)) ]; 2*x(1) -1 ] ;

Use x0 = [0.5 0.5] as initial estimates

» x0=[0.5 0.5]; tol=1.e-6; maxit=100;
» x=Newton_sys('ex11_1','ex11_1_j',x0,tol,maxit);

0 0.5000 0.5000
1.0000 0.8750 0.6250 0.3953
2.0000 0.7907 0.6181 0.0846
3.0000 0.7862 0.6180 0.0045
4.0000 0.7862 0.6180 0.0000
5.0000 0.7862 0.6180 0.0000
Newton method has converged
 Intersection of Three Surfaces

• Solve the nonlinear system

 f ( x , y, z )  x 2  y 2  z 2  14  0  f 1 ( x1, x 2 , x 3 )  x 12  x 22  x 32  14  0
 
 g ( x , y , z )  x 2
 2 y 2
 9  0 or  2 1, 2 3
f ( x x , x )  x 2
1  2 x 2 9  0
2

 h( x , y, z )  x  3 y 2  z 2  0 
 
 f 3 ( x 1 , x 2 , x 3 )  x 1  3 x 2
2  x 2
3 0

 f 1 f 1 f 1 
• Jacobian  x x 2 x 3 
 1  2 x1 2 x2 2 x3 
f 2 f 2 f 2  2 x
[J ]    4 x2 0
 x 1 x 2 x 3   1 
 f f 3 f 3   1  6 x 2 2 x 3 
 3 
 x 1 x 2 x 3 
 Newton-Raphson Method

• Solve the nonlinear system

 2 x1 2 x2 2 x 3   x 1   f1 
2 x    
4 x2 0   x 2     f 2 
 1 
 1  6 x 2 2 x 3   x 3  f 
 3
 x new   x old   x 1 
     
 new  
y  
 old   2 
y x
z   z   x 
 new   old   3 
• MATLAB function ( y = J-1F )

y =  feval(JF,xold) \ feval(F, xold)

 Intersection of Three Surfaces
function f = ex11_2(x) function df = ex11_2_j(x)
% Intersection of three surfaces % Jacobian matrix
df = [ 2*x(1) 2*x(2) 2*x(3)
f = [ (x(1)^2 + x(2)^2 + x(3)^2 - 14) 2*x(1) 4*x(2) 0
(x(1)^2 + 2*x(2)^2 - 9) 1 -6*x(2) 2*x(3) ] ;
(x(1) - 3*x(2)^2 + x(3)^2) ];

» x0=[1 1 1]; es=1.e-6; maxit=100;

» x=Newton_sys(‘ex11_2',‘ex11_2_j',x0,es,maxit);
0 1 1 1
1.0000 1.6667 2.1667 4.6667 3.9051
2.0000 1.5641 1.8407 3.2207 1.4858
3.0000 1.5616 1.8115 2.8959 0.3261
4.0000 1.5616 1.8113 2.8777 0.0182
5.0000 1.5616 1.8113 2.8776 0.0001
6.0000 1.5616 1.8113 2.8776 0.0000
Newton method has converged
 Fixed-Point Iteration

• No need to compute partial derivatives

 f 1 ( x 1 , x 2 , x 3 ,...., x n )  0  x1  g 1 ( x 1 , x 2 , x 3 ,...., x n )
 
f (
 2 1 2 3x , x , x ,...., x n )  0  x2  g 2 ( x 1 , x 2 , x 3 ,...., x n )
 
 f 3 ( x 1 , x 2 , x 3 ,...., x n )  0   x 3  g 3 ( x 1 , x 2 , x 3 ,...., x n )
   
 
 f n ( x 1 , x 2 , x 3 ,...., x n )  0  x n  g n ( x 1 , x 2 , x 3 ,...., x n )

Convergence g i K
Theorem  ; K 1
x j n
 Example1: Fixed-Point

• Solve the nonlinear system

 f 1 ( x 1 , x 2 , x 3 )  x 12  50 x 1  x 22  x 32  200  0

 2 1 2 3
f ( x , x , x )  x 2
1  20 x 2  x 3  50  0
2


f ( x
 3 1 2 3 , x , x )   x 2
1  x 2  40 x 3  75  0
2

• Rearrange (initial guess: x = y = z = 2)

 x 1  g 1 ( x 1 , x 2 , x 3 )  ( 200  x 12  x 22  x 32 ) /50

 2
x  g 2 ( x 1 , x 2 , x 3 )  ( 50  x 2
1  x 2
3 ) /20

x
 3  g 3 ( x 1 , x 2 , x 3 )  ( x 2
1  x 2  75 ) /40
2
function x = fixed_pt_sys(G, x0, es, maxit)
% Solve the nonlinear system x = G(x) using fixed-point method
% Vectors x and x0 are row vectors (for display purposes)
% function G returns a column vector, [g1(x), ..gn(x)]'
% stop if norm of change in solution vector is less than es
% y = feval(G,xold); the next approximate solution is xnew = y';

disp([0 x0 ]); %display initial estimate

xold = x0;
iter = 1;
while (iter <= maxit)
y = feval(G, xold);
xnew = y';
dif = norm(xnew - xold);
disp([iter xnew dif]);
if dif <= es
x = xnew;
disp('Fixed-point iteration has converged')
return;
else
xold = xnew;
end
iter = iter + 1;
end
disp('Fixed-point iteration did not converge')
x=xnew;
 function G = example1(x)
G = [ (-0.02*x(1)^2 - 0.02*x(2)^2 - 0.02*x(3)^2 + 4)
(-0.05*x(1)^2 - 0.05*x(3)^2 + 2.5)
(0.025*x(1)^2 + 0.025*x(2)^2 -1.875) ];

» x0=[0 0 0]; es=1.e-6; maxit=100;

» x=fixed_pt_sys('example1', x0, es, maxit);
0 0 0 0
1.0000 4.0000 2.5000 -1.8750 5.0760
2.0000 3.4847 1.5242 -1.3188 1.2358
3.0000 3.6759 1.8059 -1.5133 0.3921
4.0000 3.6187 1.7099 -1.4557 0.1257
5.0000 3.6372 1.7393 -1.4745 0.0395
6.0000 3.6314 1.7298 -1.4686 0.0126
7.0000 3.6333 1.7328 -1.4705 0.0040
8.0000 3.6327 1.7318 -1.4699 0.0013
9.0000 3.6329 1.7321 -1.4701 0.0004
10.0000 3.6328 1.7321 -1.4700 0.0001
11.0000 3.6328 1.7321 -1.4701 0.0000
12.0000 3.6328 1.7321 -1.4701 0.0000
13.0000 3.6328 1.7321 -1.4701 0.0000
14.0000 3.6328 1.7321 -1.4701 0.0000
15.0000 3.6328 1.7321 -1.4701 0.0000
Fixed-point iteration has converged
 Example 2: Fixed-Point

• Solve the nonlinear system

 f1 ( x1 , x 2 , x 3 )  x12  4 x 22  9 x 32  36  0

 2 1 2 3
f ( x , x , x )  x 2
1  9 x 2  47  0
2


 3 1 2 3
f ( x , x , x )  x 1 x 3  11  0
2

• Rearrange (initial guess: x = 0, y = 0, z > 0)

 x1  g 1 ( x1 , x 2 , x 3 )  11 / x 3


 2
x  g 2 ( x 1 , x 2 , x 3 )  47  x 2
1 /3

 3
x  g 3 ( x 1 , x 2 , x 3 )  36  x 2
1  4 x 2
2 /3
 function G = example2(x)
G = [ (sqrt(11/x(3)))
(sqrt(47 - x(1)^2) / 3 )
(sqrt(36 - x(1)^2 - 4*x(2)^2) / 3 ) ];

» x0=[0 0 10]; es=0.0001; maxit=500;

» x=fixed_pt_sys(‘example2',x0,es,maxit);
0 0 0 10
1.0000 1.0488 2.2852 2.0000 8.3858
2.0000 2.3452 2.2583 1.2477 1.4991
3.0000 2.9692 2.1473 1.0593 0.6612
4.0000 3.2224 2.0598 0.9854 0.2779
5.0000 3.3411 2.0170 0.9801 0.1263
6.0000 3.3501 1.9955 0.9754 0.0238
7.0000 3.3581 1.9938 0.9916 0.0181
8.0000 3.3307 1.9923 0.9901 0.0275
9.0000 3.3332 1.9974 1.0017 0.0129
10.0000 3.3139 1.9969 0.9962 0.0201
11.0000 3.3230 2.0005 1.0037 0.0124
12.0000 3.3105 1.9988 0.9972 0.0142
13.0000 3.3213 2.0011 1.0033 0.0126
14.0000 3.3112 1.9991 0.9973 0.0120
15.0000 3.3212 2.0010 1.0028 0.0116
16.0000 3.3120 1.9992 0.9974 0.0107
17.0000 3.3209 2.0008 1.0024 0.0103
18.0000 3.3126 1.9992 0.9977 0.0097
19.0000 3.3205 2.0007 1.0022 0.0092
20.0000 3.3130 1.9993 0.9979 0.0087
...
50.0000 3.3159 1.9999 0.9996 0.0017
...
102.0000 3.3166 2.0000 1.0000 0.0001
103.0000 3.3167 2.0000 1.0000 0.0001
Fixed-point iteration has converged
THE END