Notes from my latest install IBAMR 0.15.0 (on Jan 15, 2025)

[Adnan Morshed]

Hello all,

After some (actually lots of) trial and error, I think I got IBAMR running on my local machine. I would like to share what didn't work which might help some others with their installs. Before diving in, here's the configuration of my test machine:

I tried to follow the installation/build instructions (https://ibamr.github.io/linux) as closely as possible.

1. Chose to install boost 1.60.0 following the instructions, but found boost 1.66.0 later in the instruction at the IBAMR CMAKE section ("-DBOOST_ROOT=$HOME/sfw/linux/boost/1.66.0 \" ). Wasn't sure if IBAMR-0.15 needs boost 1.66.0 or not so removed boost 1.60.0 and installed 1.66.0 instead in sfw. Jumping ahead, during IBAMR build saw the message: 'Found Boost: /home/user/sfw/linux/boost/1.66.0/include (found suitable version "1.66.0", minimum required is "1.66") '. Please update the instructions for Boost.

2. Downloaded and installed Eigen 3.2.5. After extraction, they provide an INSTALL instruction file. Followed that and saw the headers getting installed to /usr/include/eigen3/

3. Tried installing HDF5 with PETSc using '--download-hdf5=1', which worked, but installing SAMRAI failed when it did not find HDF5. I checked and found both hypre and hdf5 inside the '....../sfw/petsc/3.17.5/linux-opt/externalpackages' but maybe I wasn't giving the correct path to hdf5 to SAMRAI. So, downloaded HDF5 1.14.5 from their website which now requires a sign in, and gives deb/rpm files. Installed the deb file and found it installed at root (wow!) '/HDF_Group/HDF5/1.14.5'.

4. Installed Silo 4.11 in sfw

5. Installed OpenMPI 5.0.6 in sfw

6. Installed PETSc 3.17.5 with hypre in sfw

7. Patched and Installed SAMRAI 2.4.4 in sfw

8. Installed libmesh with '--with-hdf5=/path/to/hdf5' and '--with-boost=/path/to/boost' added to the options

9a. First attempt at CMAKE IBAMR succeeded but said 'numdiff' not found but is optional. So went back and installed numdiff (in /usr/bin/numdiff) and tried again. This time no complaints (Found external Eigen3, Found muParser, Found HDF5). So went ahead and tried to make examples. FAILED. A bunch of muparser errors. Turns out I had libmuparser.so already in my system and IBAMR is finding that during configuration, but that native install doesn't seem to work with building examples.

9b. Uninstalled and purged my system muparser. This time, during the configuration, saw the bundled muparser getting installed ('Setting up muParser as a bundled dependency'). The examples were built, no problems. Tried to run the ex0 (following the instructions) in 'cd examples/IB/ex0'. No such files.

9c. Scrolled down to the autotools section and realized the directory should be 'examples/IB/explicit/ex0'. Found all the examples in the explicit directory and ex0 ran fine. So, 'explicit' is missing in the cmake section instructions. Please update/correct it to 'cd examples/IB/explicit/ex0'.

Aplogies for the long message, but hopefully some of the instructions on the website will get updated.

Thanks,

Adnan Morshed

[Boyce Griffith]

Adnan —

Sorry for the hassle in getting all of this stuff built.

David is actually in the middle of revamping these instructions now. (I am not sure if they have been deployed yet.) We are recommending that most people use autoibamr (https://github.com/IBAMR/autoibamr) to install IBAMR and its dependencies.

— Boyce

On Jan 16, 2025, at 3:35 PM, Adnan Morshed <adnan....@gmail.com> wrote:

Hello all,

After some (actually lots of) trial and error, I think I got IBAMR running on my local machine. I would like to share what didn't work which might help some others with their installs. Before diving in, here's the configuration of my test machine:

--
You received this message because you are subscribed to the Google Groups "IBAMR Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibamr-dev+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ibamr-dev/a073bb06-0581-48e7-a7e5-52df19ebbafdn%40googlegroups.com.
<Screenshot from 2025-01-16 12-05-13.png>

[Adnan Morshed]

Dr. Griffith,

Thanks. I see the new instruction set up on the website. Autoibamr sounds like a great idea. Will check it out for the Mac. I am still worried (based on the muparser experience) that some of the dependencies installed using the system package manager (homebrew/apt etc.) updating by themselves might still cause issues.

Thanks for the quick reply.

Adnan Morshed

[Boyce Griffith]

On Jan 17, 2025, at 12:42 PM, Adnan Morshed <adnan....@gmail.com> wrote:

Dr. Griffith,

Thanks. I see the new instruction set up on the website. Autoibamr sounds like a great idea. Will check it out for the Mac. I am still worried (based on the muparser experience) that some of the dependencies installed using the system package manager (homebrew/apt etc.) updating by themselves might still cause issues.

I think that autoibamr is able to avoid collisions with system packages — but if not, please let us know and we can try to fix.

— Boyce

To view this discussion visit https://groups.google.com/d/msgid/ibamr-dev/51ddd15b-573f-4e51-936a-41c7e0a8e883n%40googlegroups.com.