Monte Carlo Improvements in Pathfinder 2020.5

Since Pathfinder 2018.3, users have been able to take advantage of Monte Carlo simulations through a command line interface. In modeling, the Monte Carlo method is used to randomize inputs across many simulation instances. With a statistical analysis applied, users can create probabilistic interpretations of the resulting data sets. It is often used in research and is also required by regulations such as RiMEA.

How does Pathfinder utilize the Monte Carlo method?

Our Monte Carlo simulation tool currently focuses on varying Occupants' positions, Profiles, and Behaviors to generate a distribution of starting conditions. Once the inputs, including the number of variations to create, are specified, the tool can run all simulations sequentially and produce output files in individual directories. These results are collected and plotted together in a gnuplot script to summarize metrics such as Completion Time and Travel Distance for all Occupants.

Example Gnuplot of completion times for 10 simulation variations.

What's new in Pathfinder 2020.5?

The method used to distribute occupants outside their original room placement has been overhauled. Previously, this was defined by the parameter --any_room but is now defined by --room_rule and includes three options. Occupants can either be distributed to [0] stay in their original rooms, [1] reposition to other rooms containing repositioning occupants, or [2] reposition to any otherwise viable room.

An example model for Monte Carlo in Pathfinder 2020.5

Agent positions for 100 variations using room_rule 0.

Agent positions for 100 variations using room_rule 1.

Agent positions for 100 variations using room_rule 2.

The results above show how the new --room_rule parameter affects the multiple variations and their occupant positional inputs. This grants more control to the user when determining realistic scenarios of occupant placement.

Additionally, a new --help command shown below has been implemented to provide more instruction to the user.

The Pathfinder Monte Carlo CLI documentation.

How is the Pathfinder Monte Carlo tool used?

Access to the Monte Carlo tool is initiated from the Windows Command Prompt rather than from within Pathfinder itself. When Pathfinder is installed, a separate Monte Carlo utility is installed in the same directory. Users run this file from the Command Prompt to create the model variations and to initiate batch simulation. For more information, view the Pathfinder Monte Carlo User Manual.

Conclusion

Pathfinder 2020.5 is now available with the new Monte Carlo features. We anticipate additional Monte Carlo improvements in 2021 and look forward to announcing them here.

Subscribe through the button in the footer of this page if you would like to stay updated.