CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > SU2

Possible bug in SU2_DOT with DISCRETE_ADJOINT

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   April 13, 2016, 13:25
Default Possible bug in SU2_DOT with DISCRETE_ADJOINT
  #1
Member
 
Mandar Kulkarni
Join Date: Nov 2013
Location: Virginia Tech, Blacksburg, VA
Posts: 52
Rep Power: 12
kmandar is on a distinguished road
Hi SU2 developers,

Here is what I have

1. Run SU2_CFD with the quick start NACA0012 example/
2. Run SU2_CFD_AD with MATH_PROBLEM= DISCRETE_ADJOINT
3. Run SU2_DOT for Hicks-Henne design variable -- I get the following error message in the log file (or at the command prompt):
Code:
There is no adjoint restart file!! solution_adj_cl.dat.
BUT if I repeat the last step:

3. Run SU2_DOT with configuration file option MATH_PROBLEM= CONTINUOUS_ADJOINT, the job runs correctly and gives me the the result for the derivative of CL with the design variable.

It seems there is some bug with DISCRETE_ADJOINT option for MATH_PROBLEM if run with SU2_DOT.

-- Mandar

Last edited by kmandar; April 13, 2016 at 18:31. Reason: Step 2 was run with SU2_CFD_AD (not SU2_CFD as written earlier)
kmandar is offline   Reply With Quote

Old   April 13, 2016, 14:31
Default
  #2
Super Moderator
 
Tim Albring
Join Date: Sep 2015
Posts: 195
Rep Power: 10
talbring is on a distinguished road
Hi kmandar,

you have to use SU2_DOT_AD to get the correct projection. But this requires the volume adjoint solution, this is why you need to copy restart_adj_cl.dat to solution_adj_cl.dat first.

Or just run the script discrete_adjoint.py, this performs all necessay steps automatically.

Tim
talbring is offline   Reply With Quote

Old   April 13, 2016, 18:32
Default
  #3
Member
 
Mandar Kulkarni
Join Date: Nov 2013
Location: Virginia Tech, Blacksburg, VA
Posts: 52
Rep Power: 12
kmandar is on a distinguished road
Hi talbring,

That worked well.

The script runs well too.

Thanks,
Mandar
kmandar is offline   Reply With Quote

Old   April 13, 2016, 18:52
Default
  #4
Member
 
Mandar Kulkarni
Join Date: Nov 2013
Location: Virginia Tech, Blacksburg, VA
Posts: 52
Rep Power: 12
kmandar is on a distinguished road
After running the discrete_adjoint.py script, the gradient for lift are stored in the file of_grad_cl.dat and it looks like this:


Code:
VARIABLES="VARIABLE"      , "GRADIENT"      , "FINDIFF_STEP"  
          0               , 2.83854         , 0.001
What is the meaning of the last column (FinDiff_Step) in the context of discrete adjoint method?

Thanks,
Mandar
kmandar is offline   Reply With Quote

Old   April 15, 2016, 05:17
Default
  #5
Super Moderator
 
Tim Albring
Join Date: Sep 2015
Posts: 195
Rep Power: 10
talbring is on a distinguished road
Hi Mandar,

actually this is a leftover of an earlier version where the projection was also in the case of the discrete adjoint done using Finite Differences.

However, its not done anymore so you can ignore this column.

Tim
talbring is offline   Reply With Quote

Old   April 15, 2016, 17:27
Default
  #6
Member
 
Mandar Kulkarni
Join Date: Nov 2013
Location: Virginia Tech, Blacksburg, VA
Posts: 52
Rep Power: 12
kmandar is on a distinguished road
Hi Tim,

Ok, thanks,
Mandar
kmandar is offline   Reply With Quote

Reply

Tags
su2_dot discrete_adjoint

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Bug in Workbench CFX Pierre1 CFX 6 August 2, 2017 01:18
SU2_MSH: Periodic boundary conditions bug Zef SU2 1 February 18, 2015 14:28
dieselEngineFoam bug - OpenFOAM-1.6-ext novakm OpenFOAM Bugs 1 December 5, 2013 14:18
Serious bug in LES interface fs82 OpenFOAM Bugs 21 November 16, 2009 09:15
Forum y2k Bug Jonas Larsson Main CFD Forum 1 January 5, 2000 11:22


All times are GMT -4. The time now is 09:46.