Turtle Swarm Optimization

Novel, optimization aglorithm based on the biomimicry of Testudines (:turtle:).

Overview

The turtle swarm optimizer (TSO) is similar to the famous particle swarm optimizer (PSO) in its operation. In both optimization algorithms many small agents work toward a common goal of finding the local minimum of a function. The major advantages that TSO offers over the PSO algorthm can be summarized as the following:

• Turtle velocity is based on the smallest reportable 64 bit floating point number.
  • This ensures that every nook and cranny of the objective function is inspected 🔍.
• The social and personal motivation terms are not configurable.
  • We don't know exactly what motivates turtles to do what they do, so its honest to leave these terms unweighted and equal components of the turtles velocity. Staying true to the 🐢 is very important.
  • Although this new algorithm wasn't written for performance, this design decision greatly speeds up the velocity calculations by dropping slow calls to random number generators 🚀🚀🚀.
• The TSO does not allow a user to bail out of optimization early. No. The user must wait until their presdescribed goal is obtained.
  • This is done so that end-users get the feeling of what its like to rely on turtles to do the heavy lifting 💪.

Why is this unpublished?

The work is too important and publishing would only cause delays. It must be shared with the open source community.

Collaborators Welcome

There are a few rules to collaborating on this repository.

1. You cannot remove any of the advantages of the TSO algorithm.
2. You cannot change the TURTLE_VELOCITY constant unless it is discussed with the original author first.
3. You cannot introduce a performance regression ⌚.
4. You cannot introduce any logging or sense of progress to maintain the most important advantage that TSO has over PSO 🔒.
5. You must be reasonable and polite in all of your interactions on this project 💕.
6. Stay true to the 🐢