Use the Cluster Sweep (
) study to solve several models in parallel where each model has a different set of Parameters. For example, find the solution to a sequence of stationary or time-dependent simulations that arise when you vary some parameters of interest. The cluster sweep can include multiple independent parameters directly for a full multiparameter sweep (solve for the first value of the first parameter combined with all values of the second parameter, then the second value of the first parameter combined with all values of the second parameter, and so on, or use a specified combination of parameter values).
You can also add another sweep, such as a Material Sweep node, to create nested parametric sweeps. The COMSOL Multiphysics software then treats the sweeps as a “nested for-loop” and indicates the nested structure using indentations of the sweep nodes’ names. The Cluster Sweep is always the outermost sweep. It starts multiple COMSOL Multiphysics batch processes, one for each parameter set in the
Cluster Sweep node, which solve the current study on your cluster given the parameter set by submitting COMSOL Multiphysics batch jobs to a job scheduler that in turn runs the batch job on a second computer or cluster.
For more information, see Cluster Computing (Job Configurations) and its related functionality. If you used a distributed computing command to launch COMSOL Desktop, you do not need to create a
Cluster Sweep study step. See also
Batch Sweeps vs. Cluster Sweeps.
Right-click the Study node to add a
Cluster Sweep study node to run batch jobs on a cluster. Once you have specified the settings, right-click the main
Study node and select
Compute in order to start a COMSOL Multiphysics batch process that solves the current study for the given parameter sets.
When the cluster sweep process starts, the COMSOL Desktop follows the progress in The External Process Window. When the process finishes (or you click the
Detach Job button to no longer follow the process), an
External Process node, one for each parameter, is added under a
Batch Data node as in
Figure 20-5. The External Process node represents the current running process.
Click the Save as Default button (
) in the
Settings window toolbar to save the current setting as default. If you are running COMSOL Multiphysics in distributed mode, the model runs in the current process.
Use the Probes list to select probes to update during the batch sweep. The default is
All, which selects all probes for plotting and tabulation of probe data. Select
Manual to open a list with all available probes. Use the
Move Up (
),
Move Down (
),
Delete (
), and
Add (
) buttons to make the list contain the probes that you want to see results from while solving. Select
None to disable probe updating for batch sweep.
Select the Accumulated probe table check box to activate the accumulation of probe updates for both the variation on the solver level (time, frequency, and so forth) and on the batch sweep level. Use the
Output table to select where to put the data. Select the
Use all probes check box if all the model probes should be accumulated in the table. If not checked, the probes selected from the
Probes list are used.
Under Before sweep, the
Clear meshes check box is selected by default to clear the meshes before running the batch sweep. The
Clear solutions check box is selected by default to clear the solutions before running the batch sweep. Click to clear one or both of the check boxes as needed.
Under During sweep, click to select the
Synchronize solutions check box to synchronize the solutions computed by the batch processes with the model. This allows additional analysis to be performed after the sweep has finished.
The Synchronize accumulated probe table check box is selected by default to synchronize the accumulated probes computed by the batch processes with the model (collecting this information requires much less memory and time compared to the full solution information). Click to clear the check box if required.
Under After sweep, select the
Output model to file check box to enable that all batch processes save the models to file. Selecting this check box ensures that the MPH-files that are automatically saved contain the solution (for each parameter). In most cases, the solution synchronization and probe synchronization functionality should be used instead because otherwise the data ends up in one file for each process and cannot be postprocessed efficiently.