Hello,
I’m trying to execute “waveguide.edp” example by FreeFem++ in Frederichecht github. But the solution only appears the image of cube mesh, and with the error: Error umpfack umfpack_zi_numeric status 1
Error umfpack_di_solve status 1.
How can I handle it and get the solution?
Thank you so much for your help!
Minh
Can you post your output? It looked like ir ran for me with or without ff-mpirun .
I did discover though reaidng the code that “m” in the medit windo dissplays
the data which was really cool- I was trying to use plot() lol.
Hello bro,
Attached is the output when I ran the c
ode. It seemingly created the cube mesh, not the solution

at the end this line should execute and put up just a wireframe mesh.
I hit “m” in the window and got what looked like an E field pattern
Click on the wireframe and type “m” maybe.
medit("real",Th,[real(Ex),real(Ey),real(Ez)]); // in the medit window press h=help, m=data!!^M
By the way, could you upload your output for this problem?
Thanks bro. I don’t know why, I only copied the code in github and ran, but I can’t get the result like this, only creating the mesh
Posting text would be easier to read than a screenshot.
here is my text file.
script (20).log (8.5 KB)
Did you read it? The solver ran out of memory. Probably a low process
limit. I set mine lower to avoid hanging everything and set most solvers
to CG
Could you send me the text code? the solution only shows the mesh.
from the linux or cygwin command line, type this and if it still fails
send the output of this command or again try changing the solver.
ulimit -Sv 1000000
ulimit -a
Could I ask you how you have managed the EDP file? I keep getting NaN solutions. I have obtained non NaN solutions only by enlarging the waveguide dimensions, but that led large norms of the errors.
I’m not sure what you are talking about but there was a case IIRC
on the forum of waveguide code being used below cutoff but the
propagation constant was a real. Changing that to complex IIRC
produced the effeanescent or decaying wave.
Mike Marchywka
157 Zachary
Talking Rock GA 30175
was 44 Crosscreek Trail Jasper GA 30143
was 306 Charles Cox Drive Canton, GA 30115
470-758-0799
404-788-1216
Hi, thank you for your response.
I was trying to work with the “waveguide.edp” example, which is the same as the one in this topic. I keep getting NaN results, and I’ve seen that you were able to make it work (see the message from February 15th, 2023).
Did you change any parameters in the file, or did it just run without modifications?
Thank you so much for your help.
Again, I think if you just change the propagation constant to complex it prevents the
NaN and provided some results that IIRC made sense.
Mike Marchywka
157 Zachary
Talking Rock GA 30175
was 44 Crosscreek Trail Jasper GA 30143
was 306 Charles Cox Drive Canton, GA 30115
470-758-0799
404-788-1216
Hello,
When I run the waveguide.edp from
without modifications, it works without any NaN.
I have just to type “m” in the medit window to see the data (instead of only the mesh).
My output is
> FreeFem++ -ne waveguide.edp
-- FreeFem++ v4.14 (jeu. 15 févr. 2024 14:20:39 CET - git no git)
file : waveguide.edp
Load: lg_fem lg_mesh lg_mesh3 eigenvalue
(already loaded: medit)
*** Warning The identifier y0 hide a Global identifier
*** Warning The identifier y1 hide a Global identifier
sizestack + 1024 =15336 ( 14312 )
-- Square mesh : nb vertices =45 , nb triangles = 64 , nb boundary edges 24
-- Build Nodes/DF on mesh : n.v. 1305, n. elmt. 5376, n b. elmt. 1472
nb of Nodes 18904 nb of DoF 37808 DFon=0220
-- FESpace: Nb of Nodes 18904 Nb of DoF 37808
-- Solve :
min (-0.189128,-0.188568) max (0.189024,0.188986)
Hcurl norm of the exact solution = 150.235
L2 of the exact solution = 0.076252
Hcurl norm of the error = 1.03431
L2 norm of the error = 0.000152772
relative Hcurl norm of the error = 0.00688463
relative L2 norm of the error = 0.00200352
version de medit ffmedit -popen -filebin -addsol 1 real
-- Medit, Release 3.0a (Nov. 30, 2007)
Copyright (c) LJLL, 1999-2007.
compiled: jeu. 15 févr. 2024 14:27:31 CET (with ff++ 4.14).
medit with binary version of popen : mesh(es) and solution(s)
mesh_name= real
Loading data file(s)
End of mesh
.sol: Dimension 3 (mesh)3 (lecture)1
Input seconds: 0.00
medit1()
Building scene(s)
Creating scene 1
Loading default options
Scene seconds: 0.10
Rendering scene(s)
Total running seconds: 0.23
Thank you for using Medit.
times: compile 0.01481s, execution 24.706s, mpirank:0
######## We forget of deleting 6 Nb pointer, 0Bytes , mpirank 0, memory leak =11008
CodeAlloc : nb ptr 4938, size :567768 mpirank: 0
Ok: Normal End
Good evening, sir. Thank you so much for your help.
I tried running the code you linked, but I’m still getting the same problem.
This is my output:
104 :
105 : // Plot the real part of the solution
106 : medit("real",Th,[real(Ex),real(Ey),real(Ez)]); // in the medit window press h=help, m=data!!
107 : sizestack + 1024 =14608 ( 13584 )
-- Square mesh : nb vertices =45 , nb triangles = 64 , nb boundary edges 24 rmdup= 0
-- Build Nodes/DF on mesh : n.v. 1305, n. elmt. 5376, n b. elmt. 1472
nb of Nodes 18904 nb of DoF 37808 DFon=0220
-- FESpace: Nb of Nodes 18904 Nb of DoF 37808
1 UMFPACK WARNING singular matrix
1 UMFPACK WARNING singular matrix
-- Solve :
min (0,0) max (0,0)
Hcurl norm of the exact solution = nan
L2 of the exact solution = nan
Hcurl norm of the error = nan
L2 norm of the error = nan
relative Hcurl norm of the error = nan
relative L2 norm of the error = nan
version de medit ffmedit.exe -popen -filebin -addsol 1 real
-- Medit, Release 3.0a (Nov. 30, 2007)
Copyright (c) LJLL, 1999-2007.
compiled: Wed Oct 19 14:42:24 CEST 2022 (with ff++ 4.11).
medit with binary version of popen : mesh(es) and solution(s)
mesh_name= real
Loading data file(s)
End of mesh
.sol: Dimension 3 (mesh)3 (lecture)1
Input seconds: 0.15
medit1()
Building scene(s)
Creating scene 1
Loading default options
Scene seconds: 1.70
Rendering scene(s)
And this is the plot of the solution I’m getting.
I haven’t modified anything in the code.
Thank you again, for your help
Maybe you should try to install a more recent version of Freefem, apparently you have version 4.11.
version 4.15 is available at