[Xiaosong]

Hi all,

I am doing pressure inverse design using SU2. It will speed up the computation if adjoint could be added. I know SU2 can handle adjoint for Cd and Cl, and also find that objective function could be set as inverse_design_pressure in SU2. So, I am just wondering whether I can do adjoint for this, as well, just like that for Cd and Cl.

Thanks for the patience! Any comments are very welcome!

Thanks,
Xiaosong

[hlk]

Originally Posted by Xiaosong

Hi all,

I am doing pressure inverse design using SU2. It will speed up the computation if adjoint could be added. I know SU2 can handle adjoint for Cd and Cl, and also find that objective function could be set as inverse_design_pressure in SU2. So, I am just wondering whether I can do adjoint for this, as well, just like that for Cd and Cl.

Thanks for the patience! Any comments are very welcome!

Thanks,
Xiaosong

Yes, it can.
Set the appropriate objective function, and run the adjoint as you would for Cd or Cl.
See example in TestCases/optimization_euler/steady_inverse_design/ - note that config file and mesh/target cp files are in the su2code/SU2 and su2code/TestCases repositories respectively:

https://github.com/su2code/SU2/tree/...inverse_design

https://github.com/su2code/TestCases...inverse_design

[Xiaosong]

Originally Posted by hlk

Yes, it can.
Set the appropriate objective function, and run the adjoint as you would for Cd or Cl.
See example in TestCases/optimization_euler/steady_inverse_design/ - note that config file and mesh/target cp files are in the su2code/SU2 and su2code/TestCases repositories respectively:

https://github.com/su2code/SU2/tree/...inverse_design

https://github.com/su2code/TestCases...inverse_design

OK! I see! Thank you very much for your reply!
Actually, I have already tried that test case, and started building up my own cases.

For now, I get one more question:
in source code, the objective function = SUM{area * (cp - cp_target) * (cp - cp_target)}, where area is obtained from modulus of normal vectors of each small segment on airfoil. But each node will have two lines sharing it, which arc should we take into the formula, please?

Thanks,
Xiaosong