AN OPTIMAL CONTROL TOOL FOR COVID-19

• Reference paper.
• Bocop optimal control toolbox.
• Bocop covid-19 package: one control.
• Bocop covid-19 package: two controls.

• Tests 1 to 9 discussed in the above reference paper
• Four first with one control: test1.sol, test2.sol, test3.sol, test4.sol.
• Three next with two controls: test5.sol, test6.sol, test7.sol.
• Two last with one controls: quasi-periodic.sol, long-term.sol.

• In the current version the time step needs to be 1 (day). So, (in the "parameters and dimension" page from the bocop interface) the horizon (final time) and number of steps must be equal.
• About other values of the time step, see the discussion in the appendix of the paper.
• Note the "Hot start" option ("reuse" at bottom left of optimization page of the bocop interface).
• In case of difficulties of convergence, we advise to solve first with a small horizon (e.g. 30 days), and then to increase the horizon, taking advantage of the hot start option.
• More generally for solving a given problem, it is often efficient to solve a sequence of problems, starting from one already solved, with small variations between two successive problems of the sequence (the homotopy approach).
• Due to the large number of states and coefficients, it can be useful to have a tool for generating the data, namely the files problem.def, problems.constants and problem.bounds. See some resources in the python subdirectory of the bocop root.