Chill: Runtime Reconfiguration of Windowed Stream Aggregates with Semantic Guarantees This is a minimal readme which explains how to recreate graphs in the paper. Notice you might need to install dependencies (liebre, kafka) and also install python packages to create plots. Input files available here: Linear Road: https://drive.google.com/drive/folders/11GumJqI7dgngr7tQvX0uGT47GLyke2GS?usp=sharing Synthetic: Generate tuples live during execution Get the metrics' values Figure 3-4: Go into the `chill` folder, and run `./scripts/start_pane.sh`. Each command runs one experimental configuration and takes approximately 4.5 hours to complete. Therefore, running all eight commands sequentially takes approximately 36 hours in total. Figure 5-6: Go into the `chill` folder, and run `./scripts/start_example.sh` Each command runs one experimental configuration and takes approximately 1.5 hours to complete. Therefore, running all eight commands sequentially takes approximately 12.2 hours in total. Figure 7-8: Liebre baseline and $A^*$ variants: go into the `chill` folder, and run `./scripts/start_baseline.sh` Flink baselines: go into the `chill_flink` folder, and run `./scripts/start_flink_baseline.sh` Each command runs one experimental configuration and takes approximately 0.5 hours to complete. Plot the plots shown in the paper Figures 3-4: jupyter/plot1.ipynb Figures 5-6: jupyter/plot2.ipynb Figures 7-8: jupyter/plot3.ipynb