CIF 3

Real-time simulation

The CIF 3 simulator allows for a form of real-time simulation. By enabling real-time simulation, the simulator interprets the model time (the value of variable time) in seconds. The simulator attempts to ensure that the actual simulation time is equal to the model time. This feature is very useful for visualizers, which then visualize their output at a steady rate. Without real-time simulation, the rate at which such visualizers would visualize their output, would primarily be determined by the amount of time it takes for the simulator to compute the output, leading to abrupt slow downs and sudden speed increases.

Frames

If real-time simulation is disabled, visualizations are only updated for the states that are reached. That is, the visualizations are updated for the initial state, and each new state that is reached after taking a transition. The visual representations of the visualizations resulting from these states are called frames.

As time progresses, the values of continuous variables (and algebraic variables for which the value is directly or indirectly derived from a continuous variable, etc) change. If the visualization is only updated before and after time transitions, as is done when real-time simulation is disabled, this leads to ‘jumps’ in the visualization. For real-time simulation, the simulator calculates states that are evenly spaced over time. This ensures that we get regular updates of the visualizations, as time progresses, and the visualizations thus become ‘smooth’ or ‘fluent’. If enough states are visualized per second, a visualization essentially becomes a movie.

The states that result from transitions are always calculated. For real-time simulation, the simulator calculates additional states for the purpose of visualization, as mentioned above. These additional states are called intermediate states, and the corresponding visual representations of the visualizations to which they lead are called intermediate frames.

For real-time simulation, the simulator thus uses intermediate frames that are evenly spaced over time, for smooth visualizations. A consequence of this, is that state changes between those intermediate frames are lost. If the effect is still visible in the next intermediate state, this is not a problem. If however the effect is temporary, and does not last until the next intermediate state, the effect is thus not visualized.

For real-time simulation combined with an interactive input mode, the simulator uses both the state frames and the intermediate frames. This ensures that at the moment an interactive choice has to be made about which transition to take, the visualization reflects the current state of the system. Semi-automatic mode is not taken into account.

Frame rate

Using the Frame rate option (Output category), it is possible to specify how often per second of real time, visualizations should be updated. A higher frame rate leads to more updates per second, and thus to a more fluent visualization. However, since a higher frame rate also leads to more frequent updates, the amount of CPU power it takes to perform these updates increases as well. If the computer can not update the visualization in time for the next update to take place, the visualization may start to ‘stutter’.

The frame rate will thus always be a compromise between fluency and performance. Choosing an appropriate frame rate depends not only on the complexity of the specification being simulated and the visualization that is being shown, but also on the computer on which the simulation is run. Furthermore, it may depend on the other tasks the computer is performing while the simulation is being run.

Frame rates are always positive, and may be fractional. For instance, a frame rate of 7.5 is valid, and will update the visualization 7.5 times per second.

A low frame rate (typically less than 10 frames per second) can have a devastating effect on the responsiveness of SVG input. Similarly, it can hinder termination of the simulation, as the red Terminate button has no effect between intermediate frames. That is, the effect of clicking the button is postponed until the next frame, or until the time transition ends.

Simulation speed

If real-time simulation is enabled, the simulator by default interprets the model time (the value of variable time) in seconds. However, this can be changed using the Simulation speed option (Output category). This option specifies a scaling factor. A scaling factor of 1.0 interprets model time as seconds, 2.0 is twice as fast (1 unit of model time takes 0.5 seconds), etc. The simulation speed must always be positive, and may be fractional. For instance, a simulation speed of 0.5 is valid, and will interpret model time as twice as slow (1 unit of model time takes 2 seconds).

Enabling real-time simulation

By default, real-time simulation is disabled, and the simulator will try to simulate as fast as possible (infinite speed, taking time transitions takes no time). By setting a frame rate and/or a simulation speed, real-time simulation can be enabled. If only one of the options is specified, the other gets a default value. The default frame rate for real-time simulation is 20.0 frames per second. The default simulation speed is 1.0.

Enabling real-time simulation requires a real-time output component (visualizer). Currently, the following real-time output components are available:

If real-time simulation is enabled without using a real-time output component, simulation can’t be started. Closing the visualizer of a real-time simulation during the simulation, if it is the last remaining real-time output component, leads to the simulation being stopped.

All real-time output components can also be used when real-time simulation is disabled, but obviously without any real-time effects.

Model time delta

The simulation speed does not influence the frame rate. That is, both options can be specified orthogonally. The two options together however, do influence the model time delta, the amount of model time between two consecutive updates of the visualizations, i.e. the amount of model time between two consecutive intermediate frames. The model time delta is calculated using the following formula (where r is the frame rate, and s is the simulation speed): 1 / (r / s). Doubling the frame rate results in half the model time delta, while doubling the simulation speed results in double the model time delta.

If the model time delta becomes to big, this may lead to ‘jumps’ in the visualization. In such cases, increasing the frame rate, or reducing the simulation speed causes a smaller model time delta, leading to a more fluent visualization.

The model time delta (and thus the frame rate and simulation speed) also affects the ODE solver fixed output step size.

Test mode

Test mode disables all real-time delays of the real-time simulation mode, but still updates the visualizations for the intermediate frames. With test mode enabled, it is also allowed to perform a real-time simulation without real-time output component.