IB_Parallel

This code is used to simulate strongly-coupled fluid-structure interaction problems. It has parallel computing functionality in addition to an efficient fluid-structure coupling strategy. Refer to Nair and Goza JCP 2022 for more details.

Building

The following dependencies are required:

• mpich
• fftw
• PETSC with
  • fftw
  • elemental
  • mpich

then, configure PETSC_ARCH and PETSC_DIR in the makefile. Then:

make clean && make ib

Running

The compiled binary is located in ./bin/ib. Running with 16 mpi processes:

mpirun -np 16 ./bin/ib

output is stored in ./output

Apptainer

If you wish to get up and running quickly, a container solution is available with apptainer (previously singularity).

First, you will need to perform a full build. The full build compiles a base image ./base.sif that contains dependencies that will not change between compilations of IB_Parallel. Then, the actual compilation of ./ib_parallel.sif will happen with ./base.sif as the base image.

The first time you are building:

./build-apptainer.sh full

all subsequent compilations can be executed with:

./build-apptainer.sh

Running Apptainer Containers

To run a default case of

mpirun -np 2 ./bin/ib -malloc_dump -ksp_monitor_true_residual -ksp_converged_reason

simply run:

./run-apptainer.sh default

If you wish to change this command, you can edit ./ib_parallel.apptainer and recompile the image. To avoid recompiling the entire IB_Parallel project, you can instead pass a user-defined command. For example, we can print "1234" from inside the container with:

./run-apptainer.sh "echo 1234"

which outputs:

running user-defined string: 'echo 1234'
1234

The enclosing "" quotes around 1234 are important. Applying this methodology to the solver, you can adjust the number of processes for the default case with:

./run-apptainer.sh \
	"mpirun -np 16 ./bin/ib -malloc_dump -ksp_monitor_true_residual -ksp_converged_reason"