Most simulation systems require execution to stop before changes can be applied. This research examines runtime architectures that allow simulations to remain active while inputs, parameters, and conditions are modified in real time.
When simulations can be edited during execution, several practical capabilities emerge:
Rather than treating simulation as a sequence of discrete runs, analysis becomes a continuous process where understanding evolves alongside the system being modeled.
As simulations move closer to operations, the cost of interruption increases. Restarting a study is not just a computational expense—it is a cognitive one. Each interruption breaks continuity, discourages deeper exploration, and narrows the range of scenarios that can be evaluated.
Architectures that support live editing remove this barrier. They allow simulation to keep pace with human reasoning and real-world change, enabling analysis to remain aligned with evolving conditions rather than lagging behind them.
Over time, this capability reshapes how simulations are used. They become tools for active reasoning rather than retrospective analysis, supporting faster decisions, more confident interventions, and a deeper understanding of complex systems as they change.
Simulation workflows have traditionally been built around a clear separation between setup and execution. Inputs are defined, models are initialized, and only then does computation begin. Any meaningful change to the system—whether a boundary condition, operating parameter, or environmental assumption—requires the simulation to be paused, reset, or restarted entirely.
This approach reflects historical constraints rather than present-day needs.
As simulations are increasingly used in interactive training, operational planning, and live decision support, the inability to modify a study mid-execution introduces friction. Each restart disrupts analytical flow, delays insight, and limits the scope of questions that can be explored within a given timeframe. In environments where conditions evolve continuously, this stop–start model becomes a fundamental limitation.
This research exists to address that limitation directly by questioning the assumption that simulation inputs must remain static once execution begins.
A STRUCTURAL RETHINK OF SIMULATION EXECUTION
Most attempts to enable interactivity within simulations focus on surface-level controls—adjusting visualization parameters or pre-defined toggles that do not alter the underlying computation. While these features improve usability, they do not change the execution model itself.
The core insight of this research is that change does not need to be treated as an exception.
By designing simulation runtimes that continuously evaluate system state, parameter changes can be incorporated as part of normal execution rather than as a trigger for restart. Inputs become dynamic variables rather than fixed initial conditions. The simulation adapts as conditions change, maintaining continuity while recalculating only what is necessary.
This requires an execution model that is resilient to modification—one that preserves numerical stability, physical consistency, and performance even as the system evolves. Achieving this is not a user-interface problem; it is an architectural one.