 |
Octopus
|
|
Octopus
|
!brief The oct_t datatype stores the basic information about how the OCT run is done. More...
!brief The oct_t datatype stores the basic information about how the OCT run is done.
This includes, which algorithm, how the control funtion is stored, should the intermediate results be stored for debugging, etc.
Definition at line 142 of file opt_control_global.F90.
Private Attributes | |
| integer | algorithm |
| The algorithm to optimize depends on whether the control function is represented in real time, or is parametrized. Filled by the OCTScheme input variable. More... | |
| logical | mode_fixed_fluence |
| Whether or not the optimization is performed in the subspace of external fields of equal fluence. This is filled by the controlfunction_mod_init subroutine. More... | |
| real(real64) | eta |
| real(real64) | delta |
| "Technical" variables, that complete the definition of some algorithms. More... | |
| real(real64) | direct_step |
| The "initial step" of the optimization search, used by some algorithms. Filled by the OCTDirectStep input variable. More... | |
| logical | oct_double_check |
| At the end of the optimization, a final run can be performed in order to make sure that, indeed, the optimized field produces the optimal value. More... | |
| real(real64) | check_gradient |
| If using the conjugate gradients algorithm, one may make sure that the forward-backward propagation is indeed computing the gradient of the functional, by computing this gradient numerically. This is sent by the OCTCheckGradient input variable. More... | |
| integer | number_checkpoints |
| When propagating backwards, the code may check that the evolution is preserving time-reversal symmetry by checking that the state is equal to a number of previously stored "check-points", saved during the forward propagation. More... | |
| logical | random_initial_guess |
| Can be used only with some algorithms; instead of using the field described in the input file as initial guess, the code may generate a random field. More... | |
|
private |
The algorithm to optimize depends on whether the control function is represented in real time, or is parametrized. Filled by the OCTScheme input variable.
Definition at line 144 of file opt_control_global.F90.
|
private |
Whether or not the optimization is performed in the subspace of external fields of equal fluence. This is filled by the controlfunction_mod_init subroutine.
Definition at line 146 of file opt_control_global.F90.
|
private |
Definition at line 148 of file opt_control_global.F90.
|
private |
"Technical" variables, that complete the definition of some algorithms.
Definition at line 148 of file opt_control_global.F90.
|
private |
The "initial step" of the optimization search, used by some algorithms. Filled by the OCTDirectStep input variable.
Definition at line 149 of file opt_control_global.F90.
|
private |
At the end of the optimization, a final run can be performed in order to make sure that, indeed, the optimized field produces the optimal value.
Definition at line 151 of file opt_control_global.F90.
|
private |
If using the conjugate gradients algorithm, one may make sure that the forward-backward propagation is indeed computing the gradient of the functional, by computing this gradient numerically. This is sent by the OCTCheckGradient input variable.
Definition at line 153 of file opt_control_global.F90.
|
private |
When propagating backwards, the code may check that the evolution is preserving time-reversal symmetry by checking that the state is equal to a number of previously stored "check-points", saved during the forward propagation.
Definition at line 156 of file opt_control_global.F90.
|
private |
Can be used only with some algorithms; instead of using the field described in the input file as initial guess, the code may generate a random field.
Definition at line 159 of file opt_control_global.F90.
1.9.4