Main Functionality
- All the copyrights have been changed.
ODYSSEE Quasar – Solver
- Integrated new function loadCsvTags() to load csv file containing some qualitative data (for example tags).
- For maintenance purposes, ROM solvers POD ARBF, POD RBF, POD Kriging, and POD InvD are merged under pod_all (the interpolation method being specified further in the external function’s arguments list). POD ARBF, POD RBF, POD Kriging, and POD InvD will still work but not be maintained.
- Merged direct interpolators ARBF, RBF, Kriging, and InvD under “forecast” for simplicity of usage.
- The interpolation method being defined further in the external function’s arguments list.
- ARBF, RBF, Kriging, and InvD will still work but not be maintained.
- Implemented a new library called UsualFunction.qsr in Quasar.
- All functions described in this library can be used in the main quasar script thanks to the command import UsualFunction.qsr.
- Currently, only one useful function is scripting: addLabelsToMatrixAndSave().
- This function allows to save a matrix with labels, that saveCsv does not.
- Updated the Quasar documentation.
- Bug Fixes
- Quasar error message when the full database contains the same constant curves (view in CFD). It was fixed with the pod_all operator.
- Quasar Embedded (Solver):
- Ensured that the processes did not close after running.
- Reduced the CPU computational resources it occupied when it is used.
ODYSSEE Lunar – GUI
- Added a new splash screen for the Lunar interface.
- Convergence plot: A viewer has been added to display the convergence plot during an optimization. Lunar can now display the evolution of the evaluation function during the convergence loops. The same applies to the constraint functions.
- Updated the buttons for the configuration and access to the log files in the horizontal top menu with new icons. For the “5-Optimization” step, a third icon is available to access the convergence plots.
- Added a Pareto Front analysis.
- In the optimization dialog box, the user defines a constraint function for the optimization problem (for example f(x1,x2)<H), and specifies a list of N values for constraint functions H (or G) to test.
- N optimizations will be run.
- The convergence plot window will open after the first optimization has converged and will appear on the plot one by one, the following optimal points found.
- In the optimization configuration dialog box, in the third tab 3-configure optimizer, 4 parameters for convergence (Variation, EPS1, EPS2, EPS3) can now be edited.
- Variation is for the variation increment for VARIABLES.
- EPS1 is the variation increment for the gradient computation (∆x).
- EPS2 is a small value used for the machine precision test.
- EPS3 is a small value used for the convergence test. Default values are applied as before.
- Added an Export button in the sensitivity tab to export the regression coefficients in a csv file depending on the settings applied.
For example: at the cursor position, a scalar is saved; at each point, curves are saved.
- Adams Results (Advanced extraction): Added a Python tool to the parser tools list (Tools Top-menu > Parsers).
- This new parser can extract scalar or temporal (curves) values from Adams results files (*.res).
- Currently this parser is only available if Python is already installed on the computer.
- In the Adams Car parser tool, in the project configuration part, the Adams installation directory, default Adams Car Database, and default Python executable in the Adams installation directory are saved and automatically prompted at the opening of the tool. If a path is not valid, it will be written in red.
- In the Adams Car parser tool, in the extraction step, it is possible to select with ticks for which Events the output will be extracted. An error message appears if there is not at least one Event per output selected.
- A new output was added in the LS-DYNA parser tool for shell elements (ELOUT).
- Updated the Lunar documentation for:
- the pareto configuration
- the Adams Results (Advanced extraction) tool
- Added the training example (door_example) in the installer as Example5- door_example.
- Bug Fixes
- The names of the prediction curves for the validation points remained displayed in the legend even after changing the XN file in the project configuration window.
- It is fixed. Now, when the XN file is changed, the rest of the names in the legend is applied.
- For multiple datasets, there was a display problem of curves after selection/deselection/selection of the curves in Graph settings dialog box.
- If the DOE contains values that have more than 8 digits, the Nastran Parser tool truncates the value after 8 digits, therefore the resulting Nastran model is not properly written.
- The format must remain with 8 digits, therefore a large number will be approximated.
For example: 1.05292e+11 becomes 1.05e+11
- In the sensitivity tab, if the bar-plot of regression is shown in the view mode “at each point” and we entered an X-curve abscissa in graph settings then the abscissa was not updated for this graph.
- In DOE tool, when you select the Monte Carlo algorithm, a random seed is automatically used in the function. A fix has been done on the random value.
ODYSSEE Nova – Optimizer
- Bug Fix
Annealing Simulated algorithm has been rewritten to be compiled in Intel version.