mLR

Memoization-based Scalable Laminography Reconstruction.

Install and setup

We use conda to manage the environment

GPU: NVIDIA A100 GPU with 40GB memory $\times 4$ CPU: 32-core AMD EPYC 7543P 32-Core Processor

cd /OpenLR/
conda env create --name mLR -f install.yml
pip install .

Similarity profiling

To check the similarity between chunks over iteration at a same location.

cd OpenLB/Tests
python test_similarity.py

CNN encoder

To train the CNN encoder, please use the dataset in https:

    cd OpenLB/encoder/
    python encoder_train.py
    python encoder_test.py

Training the CNN encoder from scratch: Approximately 200 minutes

Run reconstruction

    cd OpenLB/Tests/
    python test_admm_brain.py