matrix finite difference method

[ptc-3682623]

Hi,
please try to find solusion of the problem in attachament. Please solve it for coefficients A B C f v = 1. I tried write the matrix but i've could not. Please help write correct matrix for my problem and then solve linear it. The matrix i the importantiest

[Fred_Kohlhepp]

Couple things:

Mathcad considers that array and matrix indexes start at 0 by default.  You change this with the system variable ORIGIN.

The order you do things makes a difference.  I've put in a small example (3 by 3 instead of 440 by 440) to demonstrate:

[terryhendicott]

Hi,

I think for matrix K you intend to assemble as one large matrix, but you have made a nested matrix.

This shows how you assemble a larger matrix from four smaller ones.

Cheers

Terry

[-MFra-]

Hi,

By chance, is the K11 matrix tridiagonal? that is:

[ptc-3682623 · Accepted solution]

Thanks for help.It was wery helpfull for me. But this is only 1 problem. We need it resume with the neuton-rapson method. I need only sample for it.

[-MFra-]

Are you looking for Newton-Raphson in several variables algorithm? see attached.

[ptc-3682623]

Thanks for this informaction we will back to this nxt time.I ve done it but i am not sure if i did it correctly, would you like looking it in attachment?

[-MFra-]

Hi,

I made some changes ....

[ptc-3682623]

Great job. Thanks. How we can write separately u and v? then see a graph in 2d for i and j parameters?

[-MFra-]

I do not know the various steps you did to get the matrix that I renamed MK, so I do not know how to proceed. u and v?

[ptc-3682623]

Look at the equation. we have u and v dimensions. Results is for both dimensions. We need to divide it on down matrix u and v. Then we need to show results on graphs.

[-MFra-]

I could hazard this solution ..... but it is completely heuristic ....

[ptc-3682623]

Hello, wow, great job. Would you compare solution in two case. For me is not the same. Is possibility draw  a graph u,v in 3d?

[-MFra-]

Hi,

[ptc-3682623]

Hej. Thanks for solusions. Would you like preparining a situaction when we have orgin=1. This will be the best situaction. I going to prepare real seample and then I will show you it.

[-MFra-]

Hi,

[ptc-3682623]

Hi, I am from Poland, I've been writing PhD and I have some problems with the finite.... I need to take your solisions and think how it write with real coefficients. I going to write it and we will think about it.ok?

[-MFra-]

Ok!

[ptc-3682623]

Hi. I could not have solved my problem. In attachment you will find a sample of shield. I discretizationed it and for it we have model of finite difference method the same like last time. for point 3 - u,v=0 for point 28 - u,v=0. For point 5,10,15,20,25 - right site of equation is 10. Please help.

[-MFra-]

Please be clearer. You should consider the fact that I am not aware of the physical system related to the differential equations that you have published, so I do not know how to help you.

[ptc-3682623]

ok. i will write some. I need only help in mathcad I can write what we need to do in Word.

[ptc-3682623]

I ve done some. I do not know why i have 74 x74=K we need 72x72 and order are the same like in sample. positions 36 of v are ok. this means 36 u and 36 v from beginning 1,i and i,1

[LucMeekes]

You've set N:=72, and you let j run from Origin to N+Origin (that's up to N+Origin=72+1=73).

Then you index MK up to j+1. That makes 73+1 =74 maximum.

Success!
Luc

[-MFra-]

Hi,

[ptc-3682623]

Hi. I ve done some but I need to do stresses and I do not know how to take into account boundary conditions. Solution is without it.