In choosing a simulation framework, it can be useful to understand where a particular choice may not be viable. Sandia National Laboratories is also sensitive to potential observation or concern that Umbra can do anything. While it is flexible, there are applications that are not suited to Umbra, either because of technical issues or efficiency. These contrary applications are not specific, but depend on the environment and application constraints.
The following are examples in which Umbra may not be suitable:
- Replacing an existing and sufficient simulation environment with Umbra.
- Creating simulations that require only high-performance parallel computing (e.g., High Performance Computing [HPC], with supercomputers).
- Requiring a narrow field of simulation that might only focus on specific attributes of only one category: cyber, physical, or behavior.
- Creating simulations where a space-time component is not valuable to the result.
- Requiring a pure light-weight agent-based simulation that is effective for HPC.
Umbra also is not a discrete event simulator, where the simulation clock advances arbitrarily to the time of the next event on the event queue. In addition, Umbra drives simulation either strictly by time-step increment, or by event occurrences, or a hybrid of both. However, if a time-stepped simulation engine with event processing capability is required, especially for embodied agent (i.e., heavyweight agent) simulation, Umbra may be a good choice.
Sandia National Laboratories is only interested in helping others apply Umbra where it can have the most benefit. Contact Sandia if you have questions.