Documentation index

Deterministic Simulations

Simulations are fully deterministic.

Given the same configuration and random seed:

  • Episode progression is identical
  • Scores occur at the same episodes
  • Pausing/resuming does not affect outcomes

This enables reliable experimentation and reproducibility.

You can also vary only the random seed to validate that results are not artifacts of a “lucky run.”


Supported Models: Linear, RNN, and GRU

  • Linear Model
  • Recurrent Neural Network (RNN)
  • Gated Recurrent Network (GRU)

Model selection is available directly in the TUI.