Skip to main content

Position

caution

Work in progress

The vectors within a population-based optimization algorithm (such as evolutionary and swarm-intelligence algorithms) represent the possible solutions to the current optimization problem. These "candidate solutions" are locations within the problem search space which the optimization problem is currently evaluating. Because the candidate solutions change over time as the optimization proceeds with the problme solution search, the candidate solutions may exist in one of two posible states:

  1. A candidate solution may be a "point" within the search space. For this location within multi-dimensional vector space, no other information is known.
  2. The candidate solution's location within the multi-dimensional vector space is known, but also the "quality" of the candidate solution has been deteremined. The quality of the candidate solution is determined by evaluating the optimization problem with the candidate solution as input.

The data structure Position represents the above description of candidate solutions, and has 2 possible states which cannot be extended:

  1. Point - a candidate solution that maintains the location within the problem search space
  2. Solution - canidate solutions that maintain both the location within the problem search space and the evaluaiton result of the location by the optimization problem

Position values implement a closed algebra of operations. This operations (such as addition and multiplication) are the same operations that are expected to exist when workting with vectors. Notably, when the resulting Position value from an operation is a new location within the problem search space, the Position will be a Point value. Similarly, re-evaluating a Solution value will not actually re-evaulate the candidate solution, as the candidate solution is an immutable value and the already evaluated value can be used as is.

Creating Positions​

Creating a Position value results in an opaque value - it is not a given that the value will be either a Point or a Solution.