JACOBIAN MATRIX AND DETERMINANT

Revision C

By Tom Irvine
Email: [email protected]

May 26, 2012

Introduction

The Jacobian of a function describes the orientation of a tangent plane to the function at a
given point.
Consider a function F given by m real-valued component functions:
y1(x1, ... , xn), ... , ym(x1 , ... , xn)

The Jacobian matrix J of F is

 y1 y1 
 x 
x n 
 1 
J     (1)
 y m 
y m 
 x1 x n 
 

Example

Consider a single-degree-of-freedom system with the following differential equation.

m x  c x  k x  F( t ) (2)

x  (c / m) x  (k / m) x  (F / m) (3)

x  (c / m) x  (k / m) x  (F / m) (4)

Let
y  x

y  x

1
The second-order equation is thus transformed into two first-order equations.

 y  0 1  x 
 y    k / m  c / m x  (5)
    

The Jacobian is

 0 1 
J  (6)
 k / m  c / m

Other Applications

The Jacobian is also used in geometrical transformation applications, as shown in

Appendix A

Reference

1. T. Irvine, The State Space Method for Solving Shock and Vibration Problems,
Revision A, Vibrationdata, 2005.

2
 APPENDIX A

Four-Node, Two-Dimensional Isoparametric Plate Element

s
Node 1
s=1
y Node 2

Node 3

Node 4

r = -1 s = -1
r=1

z x

Figure A-1.

Note that

-1 < r < +1

-1 < s < +1

3
The coordinate interpolation is

x
1
1  r 1  sx1  1 1  r 1  sx 2  1 1  r 1  sx3  1 1  r 1  sx 4
4 4 4 4
(A-1)

y
1
1  r 1  sy1  1 1  r 1  sy2  1 1  r 1  sy3  1 1  r 1  sy4
4 4 4 4
(A-2)

x 1
 1  s x1  1  s x 2  1  s x3  1  s x 4
1 1 1
(A-3)
r 4 4 4 4

x 1
 1  r x1  1  r x 2  1  r x 3  1  r x 4
1 1 1
(A-4)
s 4 4 4 4

y 1
 1  s y1  1  s y 2  1  s y 3  1  s y 4
1 1 1
(A-5)
r 4 4 4 4

y 1
 1  r x1  1  r x 2  1  r x 3  1  r x 4
1 1 1
(A-6)
s 4 4 4 4

The transformation equation is

    x y  
 r   r  
r   x 
   (A-7)
     
    x y  
 r   s s   y 

4
The Jacobian matrix J is

 x y 
 r r 
J  (A-8)
 
 x y 
 s s 

By substitution,

J

[(1+ s)x1 - (1 + s)x2 - (1 - s)x3 + (1 - s)x4 ] / 4 [(1+ s)y1 - (1 + s)y2 - (1 - s)y3 + (1 - s)y4 ] / 4
 
 
[(1+ r)x1 + (1 - r)x2 - (1 - r)x3 - (1 + r)x4 ] / 4 [(1+ r)y1 + (1 - r)y2 - (1 - r)y3 - (1 + r)y4 ] / 4 

(A-9)

5
The inverse of the Jacobian is evaluated symbolically using the software tool wxMaxima 0.8.7.

Let

Ĵ  J 1 (A-10)

Ĵ 

 (2r + 2)y4 + (2 - 2r)y3 + (2r - 2)y2 + (-2r - 2)y1 - (2s - 2)y4 + (2 - 2s)y3 + (2s + 2)y2 + (-2s - 2)y1 
1  
den  
- (2s - 2)x 4 + (2 - 2s)x3 + (2s + 2)x 2 + (-2s - 2)x1  (2s - 2)x 4 + (2 - 2s)x3 + (2s + 2)x 2 + (-2s - 2)x1 

(A-11)

den = ((s-1) x3+(-s-r) x2+(r+1) x1) y4+((1-s) x4+(r-1) x2+(s-r) x1) y3

+((s+r) x4+(1-r) x3+(-s-1) x1) y2+((-r-1) x4+(r-s) x3+(s+1) x2) y1

(A-12)

6
The wxMaxima format is

Ĵ11 

((2*r+2)*y4+(2-2*r)*y3+(2*r-2)*y2+(-2*r-2)*y1)/(((s-1)*x3+(-s-r)*x2+(r+1)*x1)*y4+((1-s)*x4+(r-1)*x2+(s-
r)*x1)*y3+((s+r)*x4+(1-r)*x3+(-s-1)*x1)*y2+((-r-1)*x4+(r-s)*x3+(s+1)*x2)*y1)
(A-13)

Ĵ12 

-((2*s-2)*y4+(2-2*s)*y3+(2*s+2)*y2+(-2*s-2)*y1)/(((s-1)*x3+(-s-r)*x2+(r+1)*x1)*y4+((1-s)*x4+(r-1)*x2+(s-
r)*x1)*y3+((s+r)*x4+(1-r)*x3+(-s-1)*x1)*y2+((-r-1)*x4+(r-s)*x3+(s+1)*x2)*y1)
(A-14)

Ĵ 21 

-((2*r+2)*x4+(2-2*r)*x3+(2*r-2)*x2+(-2*r-2)*x1)/(((s-1)*x3+(-s-r)*x2+(r+1)*x1)*y4+((1-s)*x4+(r-1)*x2+(s-
r)*x1)*y3+((s+r)*x4+(1-r)*x3+(-s-1)*x1)*y2+((-r-1)*x4+(r-s)*x3+(s+1)*x2)*y1)
(A-15)

Ĵ 22 

((2*s-2)*x4+(2-2*s)*x3+(2*s+2)*x2+(-2*s-2)*x1)/(((s-1)*x3+(-s-r)*x2+(r+1)*x1)*y4+((1-s)*x4+(r-1)*x2+(s-
r)*x1)*y3+((s+r)*x4+(1-r)*x3+(-s-1)*x1)*y2+((-r-1)*x4+(r-s)*x3+(s+1)*x2)*y1)
(A-16)

7
The inverse Jacobian at each of the four nodes is

Ĵ 1 

- (2y4 - 2y1 ) 2y 2 - 2y1 

1  
(x 2 - x1 )y 4  (x1 - x 4 )y 2  (x 4 - x 2 )y1  
 2x 4 - 2x1 - (2x 2 - 2x1 )

(A-17)

Ĵ 2 

- (2y4 - 2y1 ) 2y 2 - 2y1 

1  
(x 2 - x1 )y3  (x1 - x 3 )y 2  (x 3 - x 2 )y1  
 2x 4 - 2x1 - (2x 2 - 2x1 )

(A-18)

Ĵ 3 

- (2y3 - 2y 2 ) - (2y4 - 2y 3 )
1  
(x 3 - x 2 )y 4  (x 2 - x 4 )y3  (x 4 - x 3 )y 2  
 2x 3 - 2x 2 2x 4 - 2x 3 

(A-19)

Ĵ 4 

- (2y4 - 2y1 ) - (2y4 - 2y 3 )

1  
(x 3 - x1 )y 4  (x1 - x 4 )y3  (x 4 - x 3 )y1  
 2x 4 - 2x1 2x 4 - 2x 3 

(A-20)

8
Note that the Jacobian and its inverse should be evaluated at each node. Furthermore, at
node i,

 
   r 
 x  1  
   J i   , at r  r i and s  s j
 
 y   s 
(A-21)
The determinant of the Jacobian is found via wxMaxima.

a(r,s):=((1+s)*x1-(1+s)*x2+(-(1-s))*x3+(1-s)*x4)/4;
b(r,s):=((1+s)*y1-(1+s)*y2+(-(1-s))*y3+(1-s)*y4)/4;
c(r,s):=((1+r)*x1+(1-r)*x2+(-(1-r))*x3+(-(1+r))*x4)/4;
d(r,s):=((1+r)*y1+(1-r)*y2+(-(1-r))*y3+(-(1+r))*y4)/4;
ratsimp(determinant(matrix([a(r,s),b(r,s)],[c(r,s),d(r,s)])));

(A-22)

det J =

-(((s-1)*x3+(-s-r)*x2+(r+1)*x1)*y4+
((1-s)*x4+(r-1)*x2+(s-r)*x1)*y3+((s+r)*x4+(1-r)*x3+(-s-1)*x1)*y2+
((-r-1)*x4+(r-s)*x3+(s+1)*x2)*y1)/8
(A-23)

Equation (A-23) is left in wxMaxima format because it is convenient to copy and paste
into Matlab.

Other useful wxMaxima commands:

J(r,s):=matrix([a(r,s),b(r,s)],[c(r,s),d(r,s)]);

detJ(r,s):=ratsimp(determinant(matrix([a(r,s),b(r,s)],[c(r,s),d(r,s)])));

Jinv(r,s):=invert(matrix([a(r,s),b(r,s)],[c(r,s),d(r,s)]));