Releases: mdolab/dafoam
v2.0.4
This release migrates more solvers and functionalities from v1.
-
Added DATurbFoam for turbomachinery optimization
-
Added DASolidDisplacement for structural optimization
-
Added DASimpleTFoam for incompressible flow with heat transfer
-
Added new objective functions: wallHeatFlux, totalPressure, and vonMisesStressKS
v2.0.3
This release improves user interfaces and facilitates development.
-
The
patch
keys inprimalBC
,alpha
, andBC
are changed topatches
to enable defining multiple patches. (da6a7b2) -
Implemented lower bounds for dRdWT and dRdWTPC to reduce memory. (15c4c57)
-
Added the capability to automatically write decompseParDict file. (7eb9a4a)
-
Added python scripts to facilitate the reading and comparing Petsc vectors and matrices. (8010363)
-
Fixed an issue in Travis such that the test for pyDAFoam.py runs in serial and parallel. (f64a2ba)
v2.0.2
v2.0.1
v2.0.0
DAFoam v2.0 is a major update that introduces multiple features for a more efficient interface, better performance, and more modulated code structures.
-
Rewrite the classes and member functions. Now each partial derivative is implemented in a child class of DAPartDeriv. Similarly, each objective function is implemented in DAObjFunc. The C++ classes now receive parameters directly from the Python layer through DAOption class. The primal and adjoint calls use Petsc vectors as input and output.
-
All the primal and adjoint solvers are now compiled as libraries, and Cython is used to wrap all libraries such that everything happens in memory; there is no file IO interaction between the C++ and Python layers.
-
Optimize the screen output for primal and adjoint solutions. Now all the optimization process will be printed to screen, instead of separate files. Users can set
printInterval
to specify how frequently the primal and adjoint solution progress is printed. -
Improve the performance of the file output of intermediate shapes. Facilitate the visualization of optimization results by renaming the intermediate shape and flow fields into different time folders.
-
Use setup.py to enable
pip install .
. -
The
pyDAFoam.py
will no longer write configuration files for OpenFOAM. Users need to provide a working OpenFOAM configuration to start the optimization. -
Multipoint optimization no longer needs multiple folders.