Multi-Objective and Operations

If multi-objective optimization is not converging well, is there anything I can do to improve it?

Tunny's NSGAII, TPE, and GP can use initial individuals. So you can first run single-objective optimization for each objective function, then use the best combination of variables found as the initial individuals for multi-objective optimization. This can speed up convergence.

Can I increase the initial generation for optimization, like Galapagos' "INITIAL BOOST"?

Tunny can restart a Study, and you can change the optimization method when you restart. So you can first sample the boost portion with random or QMC, then run NSGAII using that as the initial generation.

Tunny's UI freezes during optimization, and I cannot press the STOP button to stop it.

Create a file named "quit.fishing" in the ".tunny_env" folder (or ".tunny_dev_env" for the development version), where Tunny-related files are located. When Tunny recognizes the file, it deletes the file and stops optimization on the current trial.