wecopttool.core.WEC.solve

WEC.solve(waves, obj_fun, nstate_opt, x_wec_0=None, x_opt_0=None, scale_x_wec=None, scale_x_opt=1.0, scale_obj=1.0, optim_options={}, use_grad=True, maximize=False, bounds_wec=None, bounds_opt=None, callback=None)[source]

Simulate WEC dynamics using a pseudo-spectral solution method and returns the raw results dictionary produced by scipy.optimize.minimize().

Parameters:

  • waves (Dataset) – xarray.Dataset with the structure and elements shown by wecopttool.waves.

  • obj_fun (StateFunction) – Objective function to minimize for pseudo-spectral solution, must have signature fun(wec, x_wec, x_opt, waves) and return a scalar.

  • nstate_opt (int) – Length of the optimization (controls) state vector.

  • x_wec_0 (ndarray | None) – Initial guess for the WEC dynamics state. If None it is randomly initiated.

  • x_opt_0 (ndarray | None) – Initial guess for the optimization (control) state. If None it is randomly initiated.

  • scale_x_wec (list | None) – Factor(s) to scale each DOF in x_wec by, to improve convergence. A single float or an array of size ndof.

  • scale_x_opt (FloatOrArray | None) – Factor(s) to scale x_opt by, to improve convergence. A single float or an array of size nstate_opt.

  • scale_obj (float | None) – Factor to scale obj_fun by, to improve convergence.

  • optim_options (Mapping[str, Any] | None) – Optimization options passed to the optimizer. See scipy.optimize.minimize().

  • use_grad (bool | None) – If True, optimization will utilize autograd for gradients.

  • maximize (bool | None) – Whether to maximize the objective function. The default is to minimize the objective function.

  • bounds_wec (Bounds | None) – Bounds on the WEC components of the decision variable. See scipy.optimize.minimize().

  • bounds_opt (Bounds | None) – Bounds on the optimization (control) components of the decision variable. See scipy.optimize.minimize().

  • callback (StateFunction | None) – Function called after each iteration, must have signature fun(wec, x_wec, x_opt, waves). The default provides status reports at each iteration via logging at the INFO level.

Raises:

  • ValueError – If scale_x_opt is a scalar and nstate_opt is not provided.

  • Exception – If the optimizer fails for any reason other than maximum number of states, i.e. for exit modes other than 0 or 9. See scipy.optimize for exit mode details.

Return type:

list[OptimizeResult]

Examples

The wecopttool.WEC.solve() method only returns the raw results dictionary produced by scipy.optimize.minimize().

>>> res_opt = wec.solve(waves=wave,
                        obj_fun=pto.average_power,
                        nstate_opt=2*nfreq+1)

To get the post-processed results for the wecopttool.WEC and wecopttool.pto.PTO for a single realization, you may call

>>> realization = 0 # realization index
>>> res_wec_fd, res_wec_td = wec.post_process(wec,res_opt,wave,nsubsteps)
>>> res_pto_fd, res_pto_td = pto.post_process(wec,res_opt,wave,nsubsteps)

See also

wecopttool.waves, wecopttool.core.wec.post_process, wecopttool.core.pto.post_process