[CleverBoy]

Hello Everybody,

Iam trying to use Discrete Adjoint method in SU2 for unsteady simulations. After the simulation solves for DSN001/DIRECT, it goes into DSN001/ADJOINT but it starts the adjoint simulation using 3 restart_iter files at once, as shown in below

Simulation Run using the Single-zone Driver
The simulation will run for 50 time steps.
 Loading flow solution from direct iteration 47 for zone 0.
 Loading flow solution from direct iteration 48 for zone 0.
 Loading flow solution from direct iteration 49 for zone 0.

I have 50 Time steps and since adjoint starts 1st step using 3 restart files at once, at the last steps, it diverges saying

Clearing the computational graph.

-------------------------------------------------------------------------
Storing computational graph wrt MESH COORDINATES.
+-----------------------------------------------------------------------+
|        File Writing Summary       |              Filename             |
+-----------------------------------------------------------------------+
|SU2 binary restart                 |restart_adj_pe_00047.dat           |
|                                   |                     (311.658 MB/s)|
|Paraview                           |adj_vol_solution_00047.vtu         |
|                                   |                     (184.353 MB/s)|
|Paraview surface                   |surface_adjoint_00047.vtu          |
|                                   |                     (30.7611 MB/s)|
+-----------------------------------------------------------------------+
 Setting freestream conditions at direct iteration -1 for zone 0.

-------------------------------------------------------------------------
Clearing the computational graph.


Error in "void CSolver::SetResidual_RMS(const CGeometry*, const CConfig*)": 
-------------------------------------------------------------------------
SU2 has diverged (NaN detected).
------------------------------ Error Exit -------------------------------

Do you know what is the reason for this and how can I solve it?

[CleverBoy]

Okay it was my mistake, adjoint simulation starts using 3 files at the first adjoint iteration and goes to the minus iterations for Unsteady Adjoint. My simulation itself was diverging.

[bigfootedrockmidget]

Thank you for letting us know, yes adjoints can be tricky to set up.