Data Analysis
The code includes several capabilities for analyzing simulation data on the fly. The main features include:
Time statistics: Calculation of averages and correlations.
Probes: Tools to study the evolution of specific points or regions over time.
Time statistics
The code can store the evolution of certain properties over time, storing mean and mean square
Root mean squre (rms) can be easily obtained a posteriori by:
This can be directly extracted in Visit, by loading the visitexpressions.xml file in ./Analysis
Set-up
To alocate space to store the statistics, in prob.h a new index template has to be used, indicies_stat_t
This increases the storage data to allocate space for statistical variables. Two input to the template are required, record_velocity and record_PTrho , which determine if statistics are to be stored for velocities (includes all correlations) and pressure, temperatrue and density
Additionaly, the flag cns.record_stats = 1 in the input must be set to 1 to enable the recording of statistics. If not, space for the statistics will be allocated, but no data will be written.
New variables will appear in the plot files. Note that storing statistical data will increase greatly the size of plotting and checkpoint files.
user-specific statistics in progress
Probes
Probes record time-evolution of quantities . They are handled through the input files
In the example above, a file named time_probe.log is defined to store data, including four quantities: two derived ones—energy and enstrophy—and two direct ones—density and pressure. The first two are averaged over a box of size (0, 6.28), while the latter correspond to a much smaller box (a "point")
The cns.record_probe = 1 is needed to store the data. An example can be found in the constant volume reactor where temperature is tracked over time. Results can then be plot with conventional python scripts.
Last updated