README.md

Zuckerli baseline

1) Generate .el file for each dataset

Run generate_graph_edelists.py to generate an .el file for some dataset and index string. This script takes 3 arguments:

• Dataset index, specifying which dataset to generate the .el for. See the AVAILABLE_DATASETS variable.
• Max degree parameter of NSG and HNSW.
• Path to the fb_ssnpp/ directory (only used if dataset index is set to 3).

To generate .el files for all datasets in the paper, run the following

fb_ssnpp_dir=...
for dataset_idx in 0 1 2 3; do
    for max_degree in 16 32 64 128 256; do
        ./generate_graph_edgelists_sbatch.sh $dataset_idx $max_degree $fb_ssnpp_dir
    done
done

.el files will be saved in graphs/.

2) Install veluca93/graph_utils and generate Zuckerli graph binaries

Compile the gutil following instructions in https://github.com/veluca93/graph_utils/tree/a9521a943f67466e0e1badaf10876e82c2fbef2a

Create a .bin file for each graph

# Convert graphs to zuckerli binary, skip if exists
for x in graphs/*.el; do
    if [[ ! -f "$x.bin" ]]
    then
        echo "converting $x"
        ./graph_utils/bin/gutil convert -i "$x" -F bin -o "$x.bin"
    fi
done

3) Compile Zuckerli and compress

Compile Zuckerli from https://github.com/google/zuckerli/tree/874ac40705d1e67d2ee177865af4a41b5bc2b250

Assuming you cloned zuckerli in the local directory, run the following to compress each graph.

for x in graphs/*.el.bin; do
    zuckerli/build/encoder --input_path "$x" --output_path "$x.comp"
done

This will generate logs similar to https://gist.github.com/dsevero/ac356c1c1cdf4aac17eee34387a5a4b2