About Smaug
The Computational Physics group operates a high-performance computing cluster named Smaug, hosted at the University of Oslo. The cluster supports large-scale simulations across our research areas — from quantum mechanical systems and statistical mechanics to dense-matter physics and porous-media flows.
Smaug provides the computational backbone for many of the group's projects, enabling researchers and students to run simulations that would be impractical on conventional workstations.
Accessing the Cluster
Access to Smaug is available to group members and affiliated students. The cluster is reached via SSH using your University of Oslo credentials. Once connected, you can submit jobs to the queue and manage simulation workflows.
A typical connection command looks like:
ssh username@smaug.comp-phys.netNote: You must be on the university network or connected via VPN to reach the cluster. Contact a group administrator if you need access credentials or assistance with initial setup.
The Smaug computing cluster serves as the central workhorse for the Computational Physics group's most demanding research projects. Named after the legendary dragon, this system provides the raw processing power necessary to simulate physical phenomena across multiple scales. Researchers regularly use Smaug to model quantum mechanical systems where traditional analytical methods fall short, and to run statistical mechanics calculations that explore the collective behavior of many interacting particles. The cluster enables simulations that bridge the gap between theoretical predictions and experimental observations, allowing the group to investigate questions that would otherwise remain beyond reach. Students also gain valuable hands-on experience with parallel computing through coursework and thesis projects.
Access to the Smaug cluster is managed through a structured system that prioritizes both research productivity and educational development. Group members and affiliated students can submit computational jobs that draw upon the cluster's substantial resources. The scheduling system ensures fair distribution of computing time across different projects, from large-scale production runs to smaller test calculations. Users interact with the cluster through command-line interfaces and job submission scripts, learning essential skills in high-performance computing along the way. Documentation and guidelines help users understand best practices for efficient simulation design, resource allocation, and data management, fostering a culture of responsible and effective cluster usage.
The hardware architecture of Smaug is designed to tackle the diverse computational challenges that arise in modern physics research. The cluster combines multiple computing nodes with high-speed interconnects, enabling parallel processing across many processors simultaneously. This configuration is particularly well suited for problems in dense-matter physics, where simulating neutron star interiors requires enormous computational effort. Similarly, studies of fluid flow through porous media benefit from the cluster's ability to handle complex geometries and boundary conditions. The system's storage infrastructure supports the large datasets generated by simulations, ensuring that results can be analyzed and shared among collaborators without bottlenecks.
Beyond its role in research, the Smaug cluster plays a vital part in the educational mission of the Computational Physics group. Students enrolled in advanced courses learn to write parallel code, optimize simulation performance, and interpret computational results using the cluster. These practical skills complement theoretical coursework and prepare students for careers in both academia and industry. The cluster also supports interdisciplinary collaborations, bringing together researchers from physics, mathematics, and computer science to tackle shared problems. Workshops and training sessions are organized periodically to introduce new users to the cluster's capabilities, ensuring that the next generation of computational scientists can make full use of this powerful resource.
Running Simulations
Smaug uses a job-scheduling system to manage computational workloads. Users submit batch scripts that specify the resources needed — such as CPU cores, memory, and wall-clock time — and the scheduler allocates nodes accordingly.
Key points for effective use:
- Prepare your simulation code and input files before logging in.
- Use the provided job-submission templates to structure your batch scripts.
- Monitor job status with the queue-management commands.
- Retrieve output data promptly and clean up temporary files to keep storage available for all users.
Further Resources
Detailed documentation on coding standards, best practices, and cluster policies is available on the Internals page. Group members are encouraged to review these guidelines before running production simulations on Smaug.
For troubleshooting or questions about specific software environments, consult the Software page or reach out to the group's technical contacts.