A repository with the code to reproduce the results in the paper entitled "Model Learning with Guaranteed Performance Through Embdedded Matrix submanifolds"
The results presented in each figure can be reproduced by picking any value between braces as the parameter for the corresponding table or figure.
To reproduce Table 1:
python main.py --alpha {0.5, 1, 2, 3}
To reproduce Table 2:
python main.py --lin hinfnn -g {0.5, 0.75}
python main.py --lin hinfnn -g {0.5, 1, 1.5} -e 3500
To reproduce table 3 :
python main.py --lin h2nn --mu {1, 0.1, 1e-2, 1e-3}
To reproduce table and figure 2 results :
python main.py -d pendulum -m flnssm --lmi hinf -g 1e-3 -e 10000
python main.py -d pendulum -m flnssm --lin hinfnn -g 1e-3 -e 15000 --lr 5e-3
Feel free to contact me : alexandre dot hache at imt-tlantique dot fr