by Julien Forest - (no comment)
During the execution of a numerical simulation, it is essential to be able to monitor the evolution of the most relevant physical or numerical parameters. The study of such parameters can provide a detailed view of a simulation state: whether it is converging towards equilibrium or not.Keridwen monitoring functions allow supervising such parameters during the course of a simulation. They provide the user with a graphical representation of the evolution of the parameters in the form of 1D or 2D curves.
Key features
Keridwen monitoring module offers the possibility to:- Visualise a set of curves representing f(x)=y functions and update them as the simulation is executed
- Visualise 2D maps representing f(x,y)=z functions and update them when the simulation computes new time steps
- Regroup monitoring instruments by category and filter the displayed instruments by name
- Display a spectrogram graph that is updated during the simulation course
- Dynamically add/modify/remove those monitoring instruments
- Export all those graphs to images
Embedded technologies
- Penelope
- JFreeChart
- NetCDF
Services and Support
On demand, Artenum can provide a large set of services around these functions, for instance:- Integration with an existing simulation code
- Introduction of live 3D monitoring to visualize the evolution of physical or numerical parameters in three dimensions.
- Tailored monitoring instruments with advanced visualisation capabilities (bar charts, spider or radar charts, etc.)
- Advanced configuration options of the monitoring instruments
- Charts export to advanced formats (Excel, movies, etc.)
- Alert system based on basic or complex convergence criteria
Use case
Keridwen monitoring module has been successfully integrated in the following operational example: During a simulation performed by SPIS, its numerical kernel transmits to Keridwen monitoring module several physical and numerical parameters.For instance, Figure below shows the evolution of the total energy of particles in the simulation versus time and the evolution of the distribution function.By analysing these parameters and others, plasma scientist can determine when the simulation has reached equilibrium. The possibility to see the parameters evolve without having to wait the end of the simulation allows saving time by identifying early in the simulation potential problems or inconsistencies.