Time Evolution Travel Algorithm for Population-Based Optimization
Summary
The Time Evolution Travel Algorithm (TETA) is a population-based optimizer that represents candidate solutions as separate “universes,” each with coordinates for the variables being optimized. It ranks candidates by a fitness function, retains personal and global best states, and updates each candidate by interacting with another selected candidate. The selection probability and update size are linked through a random value: better candidates are favored for selection and receive smaller changes, while poorer candidates can undergo larger changes or copy values from a better candidate. Self-interaction triggers a local Gaussian search, and proposed values are constrained to the allowed variable ranges.
The article presents an implementation and reports comparisons against other optimization algorithms on test functions. It describes TETA as having population size as its only external parameter and notes fast operation and balanced results across problem dimensions among its claimed strengths. A stated weakness is variability on low-dimensional discrete problems. The comparisons are experimental optimization results, not evidence of trading performance; they do not establish that the algorithm will improve a market strategy.
Key ideas
- TETA models candidate solutions as a population of vectors and evaluates them with a fitness function.
- Candidate interaction uses a shared random value to influence both partner selection and the scale of updates.
- The method combines movement toward other candidates with a local Gaussian search during self-interaction.
- The article reports benchmark comparisons and claims balanced results across dimensions, while noting variability on low-dimensional discrete tasks.
- Benchmark performance does not demonstrate that TETA improves trading outcomes.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.