Good Morning,
I’m new at Palabos. I read the manual and did tutorials. I used aneurysm as the starting point to create my geometry. I created my stl…everything its ok (I’ve even checked the goodness of the geometry with the tool stlcheck). Now I’ve the problem when the simulation start: I’m going to copy and paste the message from terminal:
simone85@ubuntu:~/Downloads/palabos-v2.1r0/examples/isokin(copy4test)$ ./isokin param_isokin.xml
Reynolds number, based on provided inlet diameter: 1.08e+06
Running new simulation at level 0
uLB=837.67
nuLB=0.0062
tau=0.5186
dx=25.02
dt=1.04793e+06
*[ubuntu:12749] *** Process received signal ****
[ubuntu:12749] Signal: Segmentation fault (11)
[ubuntu:12749] Signal code: Address not mapped (1)
[ubuntu:12749] Failing at address: 0x10
[ubuntu:12749] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x153c0)[0x7f428eb6a3c0]
[ubuntu:12749] [ 1] ./isokin(+0x39c61)[0x5567f8ed8c61]
[ubuntu:12749] [ 2] ./isokin(+0x3cdee)[0x5567f8edbdee]
[ubuntu:12749] [ 3] ./isokin(+0x35db3)[0x5567f8ed4db3]
[ubuntu:12749] [ 4] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3)[0x7f428e9880b3]
[ubuntu:12749] [ 5] ./isokin(+0x37b0e)[0x5567f8ed6b0e]
*[ubuntu:12749] *** End of error message ****
Segmentation fault (core dumped)
Anyone can help me to understand what is wrong with MPI?
it’s very difficult to tell from this output. But what i’m seeing is scaring me a bit:
Typically uLB must be <= 0.1.
Could be an allocation problem (you are trying to allocate too much space).
Intersting info. Nowaday the simulation works, even if I’m not still able to define two different outlet with my goemetry. I’ll take a look following you suggestion.
Here below the geometry I’m trying to simulate.
What is the boundary conditions you are using? Could try to use gdb to locate the place where the crash happens?
Following the Aneurysm model SW is able to indentify automatically inlet and outlet. In my case seems that system is not able to see the probe inlet and outlet (while for the aneurysm it does). I tried to mix boundary conditions modyfing the script cpp as follow [see "ADDICTION OF INTERNAL BC]:
// The Guo off lattice boundary condition is set up.
GuoOffLatticeModel3D<T,DESCRIPTOR>* model =
new GuoOffLatticeModel3D<T,DESCRIPTOR> (
new TriangleFlowShape3D<T,Array<T,3> > (
voxelizedDomain.getBoundary(), profiles),
flowType, useAllDirections );
model->setVelIsJ(useIncompressible); // When the incompressible BGK model is used, velocity equals momentum.
model->selectUseRegularizedModel(useRegularizedWall);
model->selectComputeStat(false);
OffLatticeBoundaryCondition3D<T,DESCRIPTOR,Velocity> boundaryCondition (
model, voxelizedDomain, lattice);
boundaryCondition.insert(rhoBarJarg);
//ADDICTION OF INTERNAL Boundary CONDITIONS
** OnLatticeBoundaryCondition3D<T,DESCRIPTOR> boundaryConditions = createLocalBoundaryCondition3D<T,DESCRIPTOR>();
** Box3D inlet (0, 20, 0, 20, 0, 20);**
** Box3D outlet (0, 20, -102.5, 20, 50, 20);**
** // Box3D inlet (100, 20, 100, 20, 125, 20);**
** // Box3D outlet (100, 20, 0, 20, 180, 20);**
** boundaryConditions->addVelocityBoundary2P (inlet, lattice, boundary::freeslip);*
** boundaryConditions->addVelocityBoundary1P (outlet, lattice, boundary::outflow);*