UNIT – V

SOLUTION OF PARTIAL DIFFERENTIAL EQUATION

1. State the conditions for the equation. Auxx + Buxy + Cuyy +Dux +Euy + Fu =G where A, B, C,
D, E, F, G are function of x and y to be elliptic (ii) parabolic (iii) hyperbolic.
Solution:
 B2- 4AC < 0 , then it is Elliptic equation.

 B2- 4AC = 0 , then it is Parabolic equation.

 B2- 4AC > 0 , then it is Hyperbolic equation.

2. What is the classification of fxx-2fyy=0?,x>0,y>0.

Solution:
Here A = 1, B = - 2, C = 0.

B2- 4AC = 4 - 0 = + ve.

This equation is of Hyperbolic type.

3. Give an example of a parabolic equation?

Solution:
u  2u
The one dimensional heat flow equation  2 2 is an important example of the Parabolic
t x
equation.

U xx  U yy  0
4. State standard Five points finite difference formula for solving .
Solution:
The standard five point formula is

1
ui , j  [ui 1, j  ui 1, j  ui , j 1  ui , j 1 ]
4
5. Write down the finite difference scheme for solving the Poisson’s equation.
Solution:
The finite difference scheme for Poisson’s equation is

Ui 1, j  Ui 1, j  Ui, j 1  Ui, j 1  4Ui, j  h2 f (ih, jh)

6. Write down one dimensional wave equation and its boundary conditions.
Solution:

 2u  2u
One Dimensional wave equation is a   0 with
2

x 2 t 2
 u(0, t )  0, u(l , t )  0, u( x,0)  f ( x), ut ( x,0)  0 .

7. State the explicit formula for the one dimensional wave equation
Solution:
The explicit scheme for one dimensional wave equation is

ui , j 1  ui 1, j  ui 1, j  ui , j 1

8. Write down the Diagonal Five Point formula in Laplace Equation.

Solution:
The Diagonal Five Point formula in Laplace Equation is

1
ui , j  ui 1, j 1  ui 1, j 1  ui 1, j 1  ui 1, j 1 
4
9. State schmidt’s explicit formula for solving heat flow equation.

Solution:

The schmidt’s explicit formula for solving heat flow equation is

ui , j 1 
1
2
 ui1, j  ui1, j 
10. Write an explicit formula to solve numerically the heat equation uxx  aut  0
Solution:

ui , j 1  ui 1, j  (1  2 )ui , j  ui 1, j

k
Where   , h is the step for x and k is the step for time.
h2a

11. Write the Crank Nicholson difference scheme to solve u xx  aut with u(0, t )  T0 , u (l , t )  Tl , and
the initial condition as u ( x, 0)  f ( x) .
Solution:

The scheme is
12. Why is Crank Nicholson’s scheme is called an implicit scheme?
Solution:
The schematic representation of Crank Nicholson method is shown below

13. Write the difference scheme for solving the Laplace’s equation.
Solution:
The difference scheme for solving the Laplace’s equation is

1
ui , j  [ui 1, j  ui 1, j  ui , j 1  ui , j 1 ]
4
14. State the method for solving the parabolic equation.
Solution:
The parabolic equation can be solved by Bender-Schmidt recurrence equation.

u i , j 1 
1
2

u i 1, j  u i 1, j . 
15. Mention the one dimensional wave equation.
Solution:
 2u  2u
The well-known hyperbolic equation , a 2   0 is called the one dimensional Wave
x 2 t 2
equation.

16. State the explicit scheme for the solution for the wave equation.
Solution:

ui , j 1  2(1  2 a 2 )ui , j  2 a 2 (ui 1, j  ui 1, j )  ui , j 1

1 h
If 2  2
or k  , then we get u i , j 1  u i 1, j  u i 1, j  u i , j 1
a a
This scheme is called an explicit scheme for the solution of the Wave equation.
17. State the condition for stability.
Solution:
ka
 If  1, The finite difference approximation is unstable.
h

ka
 If  1 , The finite difference approximation is stable.
h
ka
 If  1 , The finite difference approximation is stable but the accuracy of the solution decreases
h
ka
with the value of .
h
PART-B
The most general solution of partial differential equation of second order can be written as
 2u  2u  2u u u
A  B  C D  E  Fu  0
x 2
xy y 2
x y
Where A, B, C, D, E, F are all functions of x and y
The solution of the equation is elliptic or parabolic or hyperbolic
1. If the equation is elliptic then, B 2  4 AC  0
2. If the equation is parabolic then, B 2  4 AC  0
3. If the equation is hyperbolic then, B 2  4 AC  0

Examples:
I. Elliptic Type(Laplace Equation)
 2u  2u
1.   0 or uxx  u yy  0
x 2 y 2
i.e 2u  0
 2u  2u
2.   0  Poison equation
x 2 y 2
i.e 2u  f  x, y 
II. Parabolic Type:
 2u 1 u
  one dim ensional heat equation 
x 2  2 t
III. Hyperbolic Type:

 2u 1  2u
  one dim ensional wave equation 
x 2  2 t 2
 SOLUTION OF LAPLACE EQUATIONS BY LIEBMANN’S ITERATION PROCESS

To solve the Laplace equation u xx  u yy  0 in a bounded square region ‘R’ with a boundary ‘c’, when
the boundary value of ‘u’ are given on the boundary or atleast at the grid points in the boundary.
Let us divide the square region in to a network of sub – squares of side ‘h’

The boundary values of ‘u’ at the grid points are given and noted by b1 , b2 ,...........b16 . The values of
‘u’ at the interior lattice or grid points are assumed to be u1 , u2 ,...........u9
To start the iteration process, initially we find the rough values at interior points and then we
improve them by iterative process mostly using standard five point formula (SFPF).
Find u5 first: (By using standard 5 point formula)
1
u5 b3  b11  b15  b7 
4
Knowing u5 , we find u1 , u3 , u7 , u9 by using diagonal five point formula (DFPF)
1
u1  b1  b3  u5  b15 
4
1
u3  b3  b5  u5  b7 
4
1
u7  b15  u5  b11  b13 
4
1
u9  u5  b7  b9  b11 
4
The remaining ‘4’ values u2 , u4 , u6 , u8 can be got by using SFPF.
1
u2  b3  u3  u5  u1 
4
1
u4  u1  u5  u7  b15 
4
1
u6  u3  b7  u9  u5 
4
1
u8  u5  u9  b11  u7 
4
 Now we know all the boundary values of ‘u’ and rough values of ‘u’, at every grid point in the
interior region ‘R’.
Now we iterate the process and improve the values of ‘u’ with accuracy. Start with u5 and proceed to
get the values of u1 , u2 , u3 , u4 , u5 , u6 , u7 , u8 , u9 by using SFPF.
Continue the process we get the same values up to u1 , u2 , u3 , u4 , u5 , u6 , u7 , u8 , u9 .

Problems
1. Find by the Liebmann’s method, the value at the interior lattice points of the square region of the
harmonic function u whose boundary values are as shown in the following diagram:

Solution:
 2u  2u
Since u is the harmonic function it satisfies Laplace function (i.e.) 2  2  0
x y
To find the rough values:
*u5  SFPF 
1
u5  17.0  12.1  0  21.9
4
u5  12.525
*u1 , u3 , u7 , u9  DFPF 
1
u1  0  u5  17.0  0
4
u1  7.3812
1
u3  17.0  21.0  18.6  u5 
4
u3  17.2812
1
u7  0  12.1  u5  0
4
u7  6.1562
 1
u9  u5  9.0  21.0  12.1
4
u9  13.6562
*u2 , u4 , u6 , u8  SFPF 
1
u2  17.0  u5  12.1  0  u1  u3  12.9
4
1
u2  17.0  12.525  12.1  7.3812  17.2812  21.9
4
u2  22.0468
1
u4  19.7  u3  u9  0  u5  21
4
u4  16.112
1
u6   0  u5  u1  u7 
4
u6  6.518
1
u8  u5  12.1  u7  u9 
4
u8  11.105
Now we have got the rough values at all the interior points and already we possesses boundary
values at the lattice points.
Now we will improve the values by using SFPF
I – Iteration:
1
u11   0  u2  11.1  u6   7.787
4
1
u12  17.0  12.5  7.8  17.3  13.64
4
1
u31  13.64  21.9  19.7  u4   17.8
4
u14  16.25 , u51  11.87 , u61  6.45 , u71  6.565 , u81  11.047 , u91  14.27
II – Interation:
1
u12  0  u12  11.1  u61   7.797
4
1
u22  17.0  u51  u12  u31   13.616
4
1
u32  19.7  u14  u22  21.9   17.87
4
1
u42  u32  u91  u51  21.0   16.21
4
1
u52  u22  u81  u61  u14   11.83
4
1
u62  0  u51  u12  u71   6.548
4
 1 2
u72  u6  8.7  0  u81   6.573
4
1
u82  u52  12.1  u72  u91   11.2325
4
1
u92  u42  12.8  u82  17.0   14.31
4
III – Iteration:
1
u13  0  u22  u62  11.1
4
1
u23  u13  u32  19.0  u52 
4
1
u33  u23  21.9  19.7  u42 
4
1
u43  u33  u92  u52  21.0 
4
1
u53  u23  u82  u43  u62 
4
1
u63  0  u53  u13  u72 
4
1
u73  u63  8.7  0  u82 
4
1
u83  u53  12.1  u73  u92 
4
1
u93  12.8  u43  u83  17.0 
4
By Liebmann’s Iteration Process,
u1  7.8, u2  13.6, u3  17.8, u4  6.5, u5  11.9, u6  16.3, u7  6.6, u8  11.2, u9  14.3
 Hence the solution is
u1  7.8, u2  13.6, u3  17.8, u4  6.5, u5  11.9, u6  16.3, u7  6.6, u8  11.2, u9  14.3

2. Solve the equation  2u  0 for the following mesh, with boundary values as shown, using
Liebmann’s iteration procedure.

Solution;
Take the central horizontal and vertical lines as AB and CD.
 Let u1 , u2 , u3 , u4 , u5 , u6 , u7 , u8 , u9 be the values of u at the interior grid points of the mesh.
The values of u on the boundary are symmetrical with respect to the lines AB and CD.
Hence the values of u inside the mesh will also be symmetrical about AB and CD.
u1  u3  u7  u9 ; u2  u8 ; u4  u6 and u5 is not equal to any value.
 It is enough to find  u1 , u2 , u4 and u5
Rough values of u ' s
1
u5   2000  2000  1000  1000   1500  SFPF 
4
1
u1   0  1500  1000  2000   1125  DFPF 
4
1
u2  1000  1500  1125  1125   1187.5  SFPF 
4
1 1
u4   u1  u5  u7  2000   1225  1125  1500  2000   1437.5  SFPF 
4 4
1
u5   2u2  2u4   1312.5
4
Hereafter we can use only SFPF.
First Iteration:

Hence the solution is

u1  937.6, u2  1000.1, u4  1250.1, u5  1125.1
Solution:
Rough values of u ' s
Let us assume that u4  0
Now

1
u1  1000  1000  2000  0   1000  DFPF 
4
1
u2  1000  1000  500  0   625  SFPF 
4
1
u3   2000  1000  500  0   875  SFPF 
4
1
u4   625  875  0   375  SFPF 
4
First Iteration:

u1 u2
1000 625
1125 750
1187.5 781.2
1203.1 789.1
1208 791

u3 u4
875 375
1000 437.5
1031.2 453.11
1039 458
1041

Hence the solution is

u1  1208.3, u2  1791.7, u3  1041.7, u4  458.4
3. Evaluate the function u(x,y) satisfying  2u  0 at the lattice points given the boundary values as
follows.

Solution:
Rough values of u ' s
Let us assume that u4  0
Now

1
u1   50  100  0  0   37.5  DFPF 
4
1
u2  100  37.5  0  0   34.38  SFPF 
4
1
u3   37.5  0  0  0   9.38  SFPF 
4
1
u4   34.38  9.38  0  0   10.94  SFPF 
4
The iteration table is:
 Hence the solution is
u1  37.4, u2  37.4, u3  12.4, u4  12.4
4. Solve u xx  u yy  0 for the following square mesh with boundary conditions as shown below. Iterate
until the maximum difference between successive values at any grid point is less than 0.001.

Solution:
From the above figure, it is symmetrical about the diagonals AC and BD.
Let u1 , u2 , u3 , u4 be the values at the interior grid points.
By symmetric u1  u4 and u2  u3
Therefore, we need to find only two values u1 and u2
Since the corner values are not known, assuming u2 , we will get u1 . But assume u1 judiciously
seeing the values of u in the vertical line through u2 .
1
Let u2  1.6 [since u2 is distance of the side length from the value 2]
3
1
i.e) u 
3
1
u2  2 
3
u2  1.6
Rough values estimation:
u2  1.6
1
u1  1  1  1.6  1.6  1.3
4
1
u2   2  2  1.3  1.3  1.65
4
First iteration:
1
u1  1  1  1.65  1.65  1.33
4
1
u2   2  2  1.33  1.33  1.66
4
Second iteration:
1
u1  1  1  1.66  1.66  1.33
4
1
u2   2  2  1.33  1.33  1.66
4
The iteration table is:
 Hence the solution is
u1  1.33, u2  1.66, u3  1.33, u4  1.66

5. Solve u xx  u yy  0 for the following square mesh with boundary conditions as shown below.

Solution:
From the above figure, it is symmetrical about the diagonals AC but not about BD.
Let u1 , u2 , u3 , u4 be the values at the interior grid points.
By symmetric u1  u4 and u2  u3
Therefore, we need to find only two values u1 and u2
By symmetry, u2  u3 ; u1  u4
we need to find u1 , u2 , u4 only.
1
Assume u2  3 ( since u2 is at distance from the value u  2 )
3
1
We assume roughly u2  2   5  2   3
3
Rough values estimation:
1
u1  1  1  2u2   2
4
u2  3
1
u4  5  5  2u2   4
4
1
u2   u2  u4  2  4   3
4
First iteration:
1
u1  1  u2   2
2
1 1
u2   6  u1  u4    6  2  4  3
4 4
1
u4   5  u2   4
2
The iteration table is:

u1 u2
2 3
2 3
2 3

u3 u4
3 4
3 4
 3 4

Hence the solution is

u1  1.33, u2  1.66, u3  1.33, u4  1.66

6. Solve u xx  u yy  0 over the square of side 4 units satisfying the following conditions.
i) u  0, y   0 for 0  y  4
ii) u  4, y   12  y for 0  y  4
iii) u  x,0  3x for 0  x  4
iv) u  x, 4  x2 for 0  x  4
Solution:
We divide the square mesh in to 16 sub squares of side one unit and calculate the numerical values
of u on the boundary using given analytical expressions.
y

y=4
A0 1 4 9 16 B

0 u1 u2 u3 15

0 u4 u5 u6 14
x =0 x =4

0 u7 u8 u9 13

D 12 x
0 3 6 9 C
y=0
Let the internal grid points be u1 , u2 , u3 , u4 , u5 , u6 , u7 , u8 , u9
Rough values estimation:
1
u5   4  6  0  14   6  SFPF 
4
1
u1   0  6  4  0   2.5  DFPF 
4
1
u3  16  6  14  4   10  SFPF 
4
 1
u7  0  6  0  6  3  SFPF 
4
1
u9   6  14  6  12   9.5  DFPF 
4
We use SFPF to get the other values of u.
1
u2   4  6  2.5  10   5.625
4
1
u4   0  6  2.5  3  3.125
4
1
u6   6  14  10  9.5   9.875
4
1
u8   6  6  3  9.5   6.125
4
First iteration:
The iteration table is:
0 1 4 9 16

0 15
u1 u2 u3

2.5 5.625 10
2.4375 5.6094 9.8711
2.3672 5.5888 9.8652
2.37 5.59 9.876

0 14
u4 u5 u6

3.125 6 9.875
2.8594 6.1172 9.8721
2.8698 6.1209 9.8731
2.88 6.13 9.88

0 13
u7 u8 u9

3 6.125 9.5
2.9948 6.153 9.5063
3.0057 6.1582 9.5078
3.01 6.16 9.51

12
0 3 6 9
POISSON EQUATION:

An equation is of the form 2u  f ( x, y )

 2u  2u
 2u 
  f ( x, y )
x 2 y 2
is called as Poisson equation. Where f(x, y) is a function of x and y.
1. Solve  2u  10  x 2  y 2  10  over the square mesh with sides x  0, y  0, x  3, y  3 with u  0 on
the boundaries and the mesh length is 1 unit.
Solution:

 2u  10  x 2  y 2  10 

W.K.T  2u  f  x, y   10  x 2  y 2  10 
Here x  ih, y  jh, h  1
f  ih, jh   10  i 2  j 2  10 
By Poisson Equation
ui 1, j  ui 1, j  ui , j 1  ui , j 1  4ui , j  h2 f  x, y 


ui 1, j  ui 1, j  ui , j 1  ui , j 1  4ui , j  1 10  i 2  j 2  10   --------------- (1)
To find u1 , u2 , u3 , u4 :
At the point u1 , D 1, 2 i  1, j  2


u0,2  u2,2  u1,1  u1,3  4u1,2  10 12  22  10  
0  u2  u3  0  4u1  10 1  4  10 
u2  u3  4u1  150
-------------------- (2)
At the point u2 , E  2, 2 i  2, j  2
u1,2  u3,2  u2,1  u2,3  4u2,2  10  4  4  10
u1  0  u4  0  4u2  180
u1  4u2  u4  180 -------------------- (3)
At the point u3 , F 1,1 i  1, j  1
u0,1  u2,1  u1,0  u1,2  4u1,1  10 1  1  10
0  u4  0  u1  4u3  120
u1  4u3  u4  120 -------------------- (4)
At the point u4 , G  2,1 i  2, j  1
u1,1  u3,1  u2,0  u2,2  4u2,1  10  4  1  10
u3  0  0  u2  4u4  150
u2  u3  4u4  150 -------------------- (5)
3  u4  180  u1  4u2
5  u2  u3  4  180  u1  4u2   150
 u2  u3  720  4u1  16u2  150
4u1  15u2  u3  870
 2  4u1  u2  u3  150
Add we get 14u2  2u3  1020

7u2  u3  510 -------------------- (6)

 2  u3  150  u2  4u1
 4  u1  4  150  u2  4u1   u4  120
u1  600  4u2  16u1  u4  120
15u1  4u2  u4  720
 3  u1  4u2  u4  180
14u1  2u4  900

7u1  u4  450 ------------------- (7)

From  6 &  7 
u3  510  7u2
u4  450  7u1
 2  u2  510  7u2  4u1  150
4u1  8u2  360
u1  2u2  90 ------------------- (8)
3  u1  4u2  450  7u1  180
 8u1  4u2  270
2u1  u2  67.5 ------------------- (9)
8  u1  2u2  90
9  2  4u1  2u2  135
3u1  225
u1  75
sub u1 in (8)
75  2u2  90
2u2  165  u2  82.5
u3  510  7(82.5)  u3  67.5
u4  450  7(75)  u4  75
Hence u1  75, u2  82.5, u3  67.5, u4  75

2.Solve  2u  8 x 2 y 2 over the square mesh with u  0 on the boundaries Dividing the square inton16
sub squares of the length is 1 unit.
Solution:
 2u  8 x 2 y 2
W.K.T 2u  f  x, y   8x2 y 2
The Poisson differential equation is  2u  8 x 2 y 2 .Take the coordinates system with origin at the
Centre of the square. Since PDE &Boundary conditions are symmetrical about x,y axis.
i.e) x=y.
u1  u3  u7  u9 & u2  u4  u6  u8
Here h=1
By Poisson Equation
ui 1, j  ui 1, j  ui , j 1  ui , j 1  4ui , j  h2 f  x, y 
ui 1, j  ui 1, j  ui , j 1  ui , j 1  4ui , j  1 8 x 2 y 2 
--------------- (1)
To find u1 , u2 , u5 :
At the point u1 , D  1,1 i  1, j  1
u2,1  u0,1  u1,0  u1,2  4u1,1  8(1)(1)
0  u2  u4  0  4u1  8
u2  u4  4u1  8
4u1  u2  u2  8
-------------------- (2)
4u1  2u2  8
At the point u2 , E  0,1 i  0, j  1
u1,1  u1,1  u0,0  u0,2  4u0,1  0
u1  u3  u5  0  4u2  0
u1  u3  u5  4u2  0 (but u3  u1 )
4u2  2u1  u5  0 -------------------- (3)

At the point u5 , F  0,0 i  0, j  0

u1,1  u1,0  u0,1  u0,1  4u0,0  8(0)
u4  u6  u8  u2  4u5  0
4u2  4u5  0
-------------------- (4)
 4u1  2u2  8 -----------------(2)

4u2  2u1  u5  0 ----------(3)

4u2  4u5  0 --------------(3)

u1  3
Solve eqn (2), (3) and (4) We get u 2  2
u5   2

BENDER-SCHMIDT METHOD

 2u u
1. Solve 2  0 given that u  0, t   0, u  4, t   0, u  x,0   x(4  x), assuming h  1 .find
x 2
t
the value of u up to t  5
Solution:
 2u u
W.K.T one dimensional heat equation is a
x 2
t
 2u u
Given that 2 0
x 2
t
 2u u
2
x 2
t
On comparing (1) and (2) we get
a2
ah2 2(1)
k  1
2 2
The boundary conditions are u  0, t   0, u  4, t   0, u  x,0   x(4  x),
(i)u  0, t   0 (ii)u  4, t   0 (iii)u  x,0  x(4  x)
u  0,0   0 u  4,1  0 u  0,0   0
u  0,1  0 u  4, 2  0 u 1,0  3
u  0, 2  0 u  4,3  0 u  2,0  4
u  0,3  0 u  4, 4  0 u  3,0  3
u  0, 4  0 u  4,5  0 u  4,0  0
u  0,5  0

0 1 2 3 4

0 0 3 4 3 0

1 0 2 3 2 0
 2 0 1.5 2 1.5 0

3 0 1 1.5 1 0

4 0 0.75 1 1.75 0

5 0 0.5 0.75 0.5 0

ab
c
From diagram: 2

CRANK- NICHOLSON METHOD

1. Using crank Nicholson scheme, uxx  16ut , given that u  x,0  0, u  0, t   0, u 1, t   100t ,
1
0  t  1, t  0 compute u for one step in t taking h  .
4
Solution:
W.K.T Crank- Nicholson Formula is
1
ui , j 1  ui 1, j 1  ui 1, j 1  ui 1, j  ui 1, j 
4
Given uxx  16ut
u xx  aut
a  16
1
h
4
 1
k  ah 2  16    1  k   1
 16 
Given X varies from 0 to 1
Y Varies from 0 to 1
h  0.25
x  0, 0.25, 0.5, 0.75, 1
Boundary conditions are
u  x,0  0, u  0, t   0, u 1, t   100t ,
(i)u  x,0  0 (ii) u  0, t   0 (iii) u 1, t   100t
u  0,0   0 u  0,0   0 u 1,0  0
u  0.25,0  0 u  0,1  0 u 1,1  100
u  0.5,0  0
u  0.75,0  0
u 1,0  0
 x
0 0.25 0.5 0.75 1
t
0 0 0 0 0 0

1 0 u1 u2 u3 100

1 1
u1  0  0  0  u2   u1  u2
4 4
1 1
u2   0  0  u1  u3   u2   u1  u3 
4 4
1 1
u3   0  0  u2  100  u3   u2  100 
4 4
sub u1 & u3 in (2)
1 1 1 
u2   u2   u2  100  
4 4 4 
1 1 1 
  u2  u2  25
4 4 4 
1 2 
  u2  25
4 4 
u2 25
 
8 4
50
u2   u2  7.1429
7
1
u1   7.1429   1.7857
4
1
u3   7.1429  100   26.7857
4
Result:
x
0 0.25 0.5 0.75 1
t
0 0 0 0 0 0

1 0 1.7857 7.1429 26.7857 100

WAVE EQUATION:
1. Solve numerically 16uxx  utt , given that u  0, t   0, u  5, t   0, & u  x,0  x2 5  x  & h  1
Solution:

W.K.T a u xx  utt
2

Given 16uxx  utt

a4
h
k  0.25
a
X varies from 0 to 5
x  0,1, 2,3, 4,5
k  0.25
t  0, 0.25 , 0.5, 0.75, 1

Boundary conditions are u  0, t   0 , u  5, t   0 , u  x,0  x2  5  x 

i)u  0, t   0 ii)u  5, t   0 iii)u  x,0  x2  5  x 

u  0,0   0 u  5,0   0 u 1,0  4
u  0,0.25  0 u  5,0.25  0 u  2,0  12
u  0,0.5  0 u  5,0.5  0 u  3,0  18
u  0,0.75  0 u  5,0.75  0 u  4,0  16
u  0,1  0 u  5,1  0 u  5,0   0

x
0 1 2 3 4 5
t
0 0 4 12 18 16 0
0.25 0 6 11 14 9 0
0.5 0 7 8 2 -2 0
0.75 0 2 -2 -8 -7 0
1 0 -9 -14 -11 -6 0