Sensitivity
A Sensitivity solver node () solves a sensitivity analysis problem set up in The Sensitivity Interface.
Also see The Sensitivity Analysis Algorithm. Sensitivity analysis for time-dependent problems is available with the Optimization Module.
General
From the Defined by study step list, choose any applicable study step or choose User defined to define the following two settings in this node.
In the Objective field, specify the objective to use for the sensitivity analysis.
From the Sensitivity method list, choose one of the following options:
Adjoint — The adjoint method solves for the derivatives of a single scalar objective function with respect to any number of sensitivity variables.
Forward — The forward sensitivity method solves for the derivatives of all dependent variables and an optional scalar objective function with respect to a small number of sensitivity variables.
Forward numeric — This method is a variant of the forward sensitivity method where the right-hand side of the sensitivity problem is computed by numerical differentiation.
See also Choosing a Sensitivity Method.
If desired, select the check boxes and modify the values for any of the method-dependent parameters that are available.
From the Store sensitivity states list, choose On (the default), or choose Off if you do not want to store them.
Backward Time Stepping
This section is only available when you have selected Adjoint as the sensitivity method. It includes settings that can make the adjoint sensitivity algorithm more robust.
From the Backward time stepping list, choose one of the following options:
Automatic (the default). This option corresponds to the behavior in versions before version 6.0.
From forward. In this case, time levels taken by solver in the backward time stepping correspond to time levels taken by solver from the forward time stepping. The Reverse time step list check box is selected by default to reverse the list of time steps. From the BDF order list, choose 1 or 2 (the default). Additionally, select the Number of intermediate steps check box to add a number of intermediate steps in every time interval in the associated text field (default: 0).
Manual. With this option you can specify the backward time stepping manually using the following settings:
-
From the BDF order list, choose 1 or 2 (the default).
-
Specify an Initial step fraction (default: 1) and a relative Initial step growth rate (default: 1.5).
-
In the Time step field, enter a time step as a negative number for a backward time stepping (default: 0.01) using the base unit for time.
Advanced
From the Compensate for nojac terms list, choose Automatic (the default), On, or Off.
With Automatic, the software tries to assemble the complete Jacobian if an incomplete Jacobian has been detected. If the assembly of the complete Jacobian fails or in the case of nonconvergence, a warning is written and the incomplete Jacobian is used in the sensitivity analysis for stationary problems. For time-dependent problems, an error is returned.
With On, the software tries to assemble the complete Jacobian if an incomplete Jacobian has been detected. If the assembly of the complete Jacobian fails or in the case of nonconvergence, an error is returned.
If you get a warning about an incomplete Jacobian, you can then avoid that warning choosing Off from this list. With that setting, the software does not attempt to assemble the complete Jacobian (the incomplete Jacobian is used immediately).
From the Adjoint solution choose Initialize with zero solution (the default) or Initialize with forward solution. The latter option can reduce the computational time, if the problem is self-adjoint and an iterative solver is used. This typically happens in linear structural mechanics when the total elastic strain energy variable (Ws_tot) is used as objective. The option only affects the adjoint problem of stationary solvers.