[Core Use Case #1] Monte Carlo Random Walk Research

[dominikusbrian]

In this use case, we ought to develop ChatRnD teams of AI-Agents capable of developing and presenting novel use case of Monte Carlo Algorithm.

As part of this use case the ChatRnD AI-Agents should cover the literature to first introduce existing state of Monte Carlo Algorithm research. Make the basic derivations and cover basic knowledge on the topic related to its application. The team should then pick a topic of interest to then focus developing improved version that can be tested and utilized. Finally a report on the development should be made by the team.

The first version out of the box based on ChatDev basic framework has been capable of doing the following:

In which a simple model box for performing monte carlo random walk has been performed.

Improved version of the ChatRnD should be capable of performing statistical analysis and more.

[dominikusbrian]

After adjusting for base chat_env and modifying the level of details for the task our ChatRnD team can now produce the following, in which user can specify a specific file, to then save the result for each experiment.

It also generate a bit of overview on the theory such as:

Next, we want to see this be self-iterated by the team and expecting to see them started plotting and performing analysis on the data, maybe also improve on the nature of trajectories noise variation and more.

[dominikusbrian]

Another extra iteration for today's test. Now it can generate specific number of trajectories, then save .csv and also plot the trajectories

The review part also is improving with a bit more equation on it.

This is getting exciting. Will try to reformulate the role recruiting and let them running with iteration when it comes to analyzing the data for more rigor.