Installation#

1. New to Python or Anaconda?#

Python is a high-level programming language. To install and use Python on your computer, I highly recommend you install Anaconda, which provides a platform to use Python and managing Python packages. You can download and install Anaconda by following the instructions on https://www.anaconda.com/distribution/.

2. Install FastMBAR with pip#

  • Install FastMBAR without CUDA support

    If you do not have GPUs on your computers, you can still use FastMBAR on CPUs. In this case, the installing FastMBAR is very easy. Open a Terminal and run the following command

    pip install -U FastMBAR

  • Install FastMBAR with CUDA support

    If you have GPUs on your computers and you want to run FastMBAR on GPUs, you need to install FastMBAR with CUDA support. Because FastMBAR uses PyTorch for calculations on GPUs, you will need to install PyTorch with CUDA support before installing FastMBAR:

    1. Install PyTorch with CUDA support.

      Follow the instructions at PyTorch to install PyTorch with CUDA support. The specific command you need could depends on your operation system, Python version, and CUDA version you have on your computers. You can use the following command in a Terminal to check these version information.

      ## check Python version
python --version

## check CUDA version
nvcc --version

      If the command nvcc --version returns the error message nvcc: command not found..., it means the CUDA toolkit is not installed on your computer or the module of CUDA is not loaded in your current environment. You can contact your server administrator to install CUDA toolkit or activate it in your environment.

    2. Test installed PyTorch with CUDA support.

      Run the following command in a Terminal on a computer with GPUs to test if installed PyTorch has CUDA support:

      python -c "import torch;  print(torch.cuda.is_available())"

      If PyTorch is installed correctly and has CUDA support, the above command should print True in the terminal. If not, please go back to step 1 and reinstall PyTorch with CUDA support correctly. For more information on installing PyTorch, please read more detailed instructions on PyTorch.

  1. Install FastMBAR

    After you have successfully installed PyTorch with CUDA support, you can use the following command to install FastMBAR.

    pip install -U FastMBAR

3. Test the Installation of FastMBAR#

Run the following command in a terminal to test if FastMBAR has been installed successfully.

pytest -v --pyargs FastMBAR

If FastMBAR has been successfully installed, it will output the following information:

  • If FastMBAR is installed with CUDA support, then on a computer with GPUs, the above command will print information similar as the following output:

    ========================================================= test session starts ==========================================================
platform linux -- Python 3.11.4, pytest-7.4.0, pluggy-1.2.0 -- /home/xqding/apps/miniconda3/envs/test/bin/python3.11
cachedir: .pytest_cache
rootdir: /home/xqding
collected 12 items

test_FastMBAR.py::test_FastMBAR_cpus[False-Newton] PASSED        [  8%]
test_FastMBAR.py::test_FastMBAR_cpus[False-L-BFGS-B] PASSED      [ 16%]
test_FastMBAR.py::test_FastMBAR_cpus[True-Newton] PASSED         [ 25%]
test_FastMBAR.py::test_FastMBAR_cpus[True-L-BFGS-B] PASSED       [ 33%]
test_FastMBAR.py::test_FastMBAR_gpus[False-False-Newton] PASSED  [ 41%]
test_FastMBAR.py::test_FastMBAR_gpus[False-False-L-BFGS-B] PASSED [ 50%]
test_FastMBAR.py::test_FastMBAR_gpus[False-True-Newton] PASSED   [ 58%]
test_FastMBAR.py::test_FastMBAR_gpus[False-True-L-BFGS-B] PASSED [ 66%]
test_FastMBAR.py::test_FastMBAR_gpus[True-False-Newton] PASSED   [ 75%]
test_FastMBAR.py::test_FastMBAR_gpus[True-False-L-BFGS-B] PASSED [ 83%]
test_FastMBAR.py::test_FastMBAR_gpus[True-True-Newton] PASSED    [ 91%]
test_FastMBAR.py::test_FastMBAR_gpus[True-True-L-BFGS-B] PASSED  [100%]

==================================================== 12 passed in 111.64s (0:01:51) ====================================================

  • If FastMBAR is installed without CUDA support or if FastMBAR is installed with CUDA support but the above command is run on a computer without GPUs, the above command will print information similar as the following output:

    ========================================================= test session starts ==========================================================
platform linux -- Python 3.11.4, pytest-7.4.0, pluggy-1.2.0 -- /home/xqding/apps/miniconda3/envs/test/bin/python3.11
cachedir: .pytest_cache
rootdir: /home/xqding/test
collected 12 items

test_FastMBAR.py::test_FastMBAR_cpus[False-Newton] PASSED                                                                        [  8%]
test_FastMBAR.py::test_FastMBAR_cpus[False-L-BFGS-B] PASSED                                                                      [ 16%]
test_FastMBAR.py::test_FastMBAR_cpus[True-Newton] PASSED                                                                         [ 25%]
test_FastMBAR.py::test_FastMBAR_cpus[True-L-BFGS-B] PASSED                                                                       [ 33%]
test_FastMBAR.py::test_FastMBAR_gpus[False-False-Newton] SKIPPED (CUDA is not avaible)                                           [ 41%]
test_FastMBAR.py::test_FastMBAR_gpus[False-False-L-BFGS-B] SKIPPED (CUDA is not avaible)                                         [ 50%]
test_FastMBAR.py::test_FastMBAR_gpus[False-True-Newton] SKIPPED (CUDA is not avaible)                                            [ 58%]
test_FastMBAR.py::test_FastMBAR_gpus[False-True-L-BFGS-B] SKIPPED (CUDA is not avaible)                                          [ 66%]
test_FastMBAR.py::test_FastMBAR_gpus[True-False-Newton] SKIPPED (CUDA is not avaible)                                            [ 75%]
test_FastMBAR.py::test_FastMBAR_gpus[True-False-L-BFGS-B] SKIPPED (CUDA is not avaible)                                          [ 83%]
test_FastMBAR.py::test_FastMBAR_gpus[True-True-Newton] SKIPPED (CUDA is not avaible)                                             [ 91%]
test_FastMBAR.py::test_FastMBAR_gpus[True-True-L-BFGS-B] SKIPPED (CUDA is not avaible)                                           [100%]

==================================================== 4 passed, 8 skipped in 29.67s =====================================================