Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations Chriss Miller on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
-
Is my formula wrong or the intrinsic function "abs" doesn't work?
Hi, maki1337. I can see you sorted out whatever problem you had. But I'm just curious about why you should specify the roots of your quadratic as "integer" (instead of "real"). I don't think you can always know before hand if you'll get integer roots or otherwise. Unless you're only interested... -
Lagrange interpolation over a larger set of data
Hi mikrom, Thanks for the modifications and the extra index-finding functions you introduced. I did the adaptations to my problem and everything works perfectly well! Thank you so much again. -
Lagrange interpolation over a larger set of data
Hi mikrom, I have tried the idea you suggested above. But your post has highlighted the mistake I made of not using the appropriate indexing of the input data and the point of interest. I'll follow your suggestions and try to make my procedure work. -
Lagrange interpolation over a larger set of data
Hi mikrom, I am not able to attach the files themselves. So I'll just paste the data below. The file.dat.400 on is the higher resolution which is the main input file (I call the y-column there Y_hi). The file.dat.200 is the lower resolution file from which I only need the X's. The aim is to... -
Lagrange interpolation over a larger set of data
...! Performing the interpolation do i = 1,n do j = 1, n if (i /= j) then L(i) = ( (z - xvect(j)) / (xvect(i) - xvect(j)) )* L(i) ! part of L(i) end if end do Pz = Pz + L(i)*yvect(i) ! update Pz ~ f(z) end do write(*,*) "The approximate value of... -
solving linear system with zero rows(picking rows and columns)
...I am getting impression that you are confusing variable values with variable coefficients." Oh, I agree salgerman. My language use was very sloppy (or even inaccurate) and hence the confusion between matrix coefficients and variable. Also, the coefficient matrix I printed out in my original... -
solving linear system with zero rows(picking rows and columns)
Yes, looking at the coefficient matrix, there equations are intriguing for sure. Just some brief background of the problem. This is part of an evolution problem which uses implicit time-stepping (integrator). So this involves a Newton-Raphson step at some point. It is in this step that we then... -
solving linear system with zero rows(picking rows and columns)
mikrom,I'm sorry the example program you gave me (http://www.tek-tips.com/viewthread.cfm?qid=1678628 ) couldn't work because I had failed to properly install LAPACK on my system. These were the error messages from the two options you gave me: (1) /tmp/cczJi5YX.o: In function `MAIN__'... -
solving linear system with zero rows(picking rows and columns)
Yes gummibaer, I have tried a countless number of times to get LAPACK onto my system but failed. My PC is running on Linux Ubuntu 11.10 & uses the Gnome desktop (&terminal). My compiler is "gfortran". Even though my professors don not fully approve of using "black-box" solvers, I always wish to... -
solving linear system with zero rows(picking rows and columns)
Those numbers are practically zeros, even when printed to 16 decimal places. They are just floating point artifacts. -
solving linear system with zero rows(picking rows and columns)
...real(sp), allocatable :: a_prime(:, :) , b_prime(:) ! matrix A open(unit=7, file='matrix.dat', status = 'old', action='read') read(7,*) ((a(i,j), j=1,ncols), i=1,mrows) close(7) ! matrix b open(unit=11, file='bb.dat', status = 'old', action='read') read(11,*) (b(i)... -
Using LAPACK library in fortran codes
Has anyone ever used LAPACK in their fortran(90) codes? I believe this can help me compare the results from my hand-coded linear-algebra codes - or even replace these in case LAPACK gives far better performance. Small pieces of example code might be helpful. I've also tried installing this... -
Appropriate use of allocatable arrays
Here are the type declarations in "sf_1d_mod" below. The numerical values of a, b, x, y, p and q are computed in some other subroutine before they are "pointed" to in this "backEuler_time_step" one. Even though I haven't done a specific check, I believe they do find their ways into the arrays... -
Appropriate use of allocatable arrays
...also compute iteration matrix (a 3D-array) coeff_matrix = iter_matrix ! 3D-arrays dimension(nvars,nvars,0:n) rhs_vector_tmp = dt*rhs_fun - u ! 2D-arrays, dimension(nvars,0:n) rhs_vector = rhs_vector_tmp + u0 ! Full step t --> t+dt. this is "b" in "Ax=b" do i = 0, n... -
3D array data storage
Thank you for the assurance and extra tip, gummibaer. When I said "retrieve", I actually meant to say "access". I need to access the 2D arrays (6x6 matrices) for each i, and use it as the matrix "A" in Ax=b and solve a system of equations. In the point you made concerning this, would I be... -
3D array data storage
Hi. I'm working with 1D and 2D f90 arrays. Actually, for a fixed value of a looping/iteration index, I have to form a 2D (dimension(6,6) ) array that holds some Jacobian matrix. Calling this iteration counter i, where i ranges from 0 to n (say n=5000), this means I need to compute the 6x6... -
Finite difference approximation of Jacobian
Oops, the central difference estimation is amazing! Thanks indeed, mikrom!!! -
Finite difference approximation of Jacobian
...) / eps du/dx_2 = ( u(x_1, x_2 + eps) - u(x_1, x_2) ) / eps Central diff.: du/dx_1 = ( u(x_1 + eps, x_2) - u(x_1 - eps, x_2) ) / 2 * eps du/dx_2 = ( u(x_1, x_2 + eps) - u(x_1, x_2 - eps) ) / 2 * eps Thank you in advance! Fortran code... -
Linking Fortran 90 code with 'gfortran' compiler on linux ubuntu 11.10
That's the point. But I had one. Eventually, I have somehow managed to reproduce the error message. I noticed that if at some point you mess up slightly in the linking sequence e.g. using the " -o " option without supplying the name of the output executable, the main program object file gets... -
Linking Fortran 90 code with 'gfortran' compiler on linux ubuntu 11.10
This was the error message: /usr/lib/gcc/i686-linux-gnu/4.6.1/../../../i386-linux-gnu/crt1.o: In function `_start': (.text+0x18): undefined reference to `main' collect2: ld returned 1 exit status I'm sorry for not including it in my initial post. I only wrote it down somewhere. I believe it...
