Examples of gradients calculations using the adjoint state method with OwlDE
Currently, we have two examples: one solves an intial value problem and the other learns the parameters of a dynamical system as well as its initial condition.
Example runs of the two examples, including the invocation commands can be found below:
$ mkdir results
$ dune exec examples/initial_value_problem.exe --profile=release
iter 203 | loss 0.001 | pct change 0.03960
x1: actual and target
C0
R0 0.970162
R1 0.00882855
C0
R0 1
R1 0 $ mkdir results
$ dune exec examples/params_adjoint.exe --profile=release
step 2520 | loss 0.01008 | pct change 0.00131