Dyncurv

An implementation of curvature sets over dynamic metric spaces

Usage

First, build dyncurv executable (which supports computation of supports and various distances):

cd cpp_impl && make -j8

Then load the required dependencies (optional if installed):

conda env update -f environment.yml
conda activate dyncurv_venv

Then cd into working directory:

cd python_impl

and generate the boids:

python3 experiment.py --no-analysis --no_dist_mat --num_flocks=10 --num_boids=50 --time_steps 600 --scale 2.5 --write_nth 5)

and then run the k=0 and k=1 stages:

python3 experiment.py --no_analysis --no_boids --num_flocks=10 --k 1 --num_target_samples 15000 --num_max_samples 2000
mv ../data/dist_mat.dat ../data/dist_mat_1.dat
python3 experiment.py --no_analysis --no_boids --num_flocks=10 --k 0 --num_target_samples 7000 --num_max_samples 1000
mv ../data/dist_mat.dat ../data/dist_mat_0.dat

Then leave working directory:

cd ..

Finally, visualize the output:

cd data
python3 build_mats.py # Type dE when prompted: d2 is for testing purposes. Type yes for rescaling when prompted
cd ..
cd python_impl 
python3 experiment.py --no_boids --no_dist_mat