Run Your First Simulation

Technical Details

Running your DualSPHysics simulation workflows on Inductiva is similar to running them locally, with a few key differences. Instead of executing your DualSPHysics shell script directly, you'll need to pass it to the Inductiva API using the run() method. This will allow your simulation to be executed on a remote resource. If you already have a functioning shell script that orchestrates your DualSPHysics simulation locally, you're almost ready to run it on Inductiva.

Key Adaptations for Inductiva

While the overall process is similar, there are a few minor adjustments that might be needed to accommodate the different environment:

• Path-related changes: The location of the DualSPHysics binaries in the Inductiva infrastructure may differ from your local setup. You may need to modify path variables in your orchestration script.
• Interactive commands: Any commands that require keyboard input will need to be adjusted. You can either remove these commands or set them to run with default parameters.

Location of binaries

All DualSPHysics binaries are located in the directory /DualSPHysics_v5.2/bin/linux/. This path is included in the system's PATH environment variable, so you can execute commands directly without needing to specify the full path.

We've also created symbolic links for easier access. For example, instead of using the full name DualSPHysics5.2CPU_linux64, you can simply call dualsphysics. The naming convention is simple: we remove the _linux64 suffix and convert the name to lowercase.

This renaming allows you to use either the full binary name (e.g., DualSPHysics5.2CPU_linux64) or the simplified name (e.g., dualsphysics), which abstracts away the architecture and simulator version.

Objective

This tutorial will show you how to run DualSPHysics simulations using the Inductiva API.

We will cover a classical CFD case of a flow over a cylinder from the DualSPHysics offical download page to help you get started with simulations.

Prerequisites

Download the required files present in DualSPHysics_v5.2.2.ziphere and place them in a folder called DualSPHysics_v5.2. All your simulation files will be located at DualSPHysics_v5.2/examples/inletoutlet/01_FlowCylinder.

After that you will need to make some changes to the file xCaseFlowCylinder_Re020_linux64_GPU.sh:

• Update the dirbin to export dirbin=/DualSPHysics_v5.2/bin/linux/
• Remove the last line of the script read -n1 -r -p "Press any key to continue..." key

You now have all you need to start your simulation at DualSPHysics_v5.2/examples/inletoutlet/01_FlowCylinder.

Running a DualSPHysics Simulation

Here is the code required to run a DualSPHysics simulation using the Inductiva API:

"""DualSPHysics example."""
import inductiva

# Allocate a machine on Google Cloud Platform
cloud_machine = inductiva.resources.MachineGroup( \
    provider="GCP",
    machine_type="g2-standard-32",
    spot=True)

# Initialize the Simulator
dualsphysics = inductiva.simulators.DualSPHysics( \
    version="5.2.1")

# Run simulation
task = dualsphysics.run( \
    input_dir="/Path/to/01_FlowCylinder",
    shell_script="xCaseFlowCylinder_Re020_linux64_GPU.sh",
    on=cloud_machine)

# Wait for the simulation to finish and download the results
task.wait()
cloud_machine.terminate()

task.download_outputs()

task.print_summary()

In this basic example, we're using a cloud machine (g2-standard-32) equipped with 32 virtual CPUs and an NVIDIA L4 GPU. For larger or more compute-intensive simulations, consider adjusting the machine_type parameter to select a more powerful GPU machine. You can explore the full range of available machines here.

Note: Setting spot=True enables the use of spot machines, which are available at substantial discounts. However, your simulation may be interrupted if the cloud provider reclaims the machine.

To adapt this script for other DualSPHysics simulations, replace input_dir with the path to your DualSPHysics input files and set the shell_script accordingly. Be sure to specify the DualSPHysics version compatible with your input files.

When the simulation is complete, we terminate the machine, download the results and print a summary of the simulation as shown below.

Task status: Success

Timeline:
    Waiting for Input         at 09/04, 10:29:35      0.871 s
    In Queue                  at 09/04, 10:29:36      54.536 s
    Preparing to Compute      at 09/04, 10:30:31      12.672 s
    In Progress               at 09/04, 10:30:44      99.348 s
        └> 99.186 s        bash xCaseFlowCylinder_Re020_linux64_GPU.sh
    Finalizing                at 09/04, 10:32:23      64.657 s
    Success                   at 09/04, 10:33:28

Data:
    Size of zipped output:    1.25 GB
    Size of unzipped output:  2.41 GB
    Number of output files:   2032

Total estimated cost (US$): 0.044 US$
    Estimated computation cost (US$): 0.034 US$
    Task orchestration fee (US$): 0.010 US$

Note: A per-run orchestration fee (0.010 US$) applies to tasks run from 01 Dec 2025, in addition to the computation costs.
Learn more about costs at: https://inductiva.ai/guides/how-it-works/basics/how-much-does-it-cost

As you can see in the "In Progress" line, the part of the timeline that represents the actual execution of the simulation, the core computation time of this simulation was 99.3 seconds (around 1 minute and 39 seconds).