The current state of EDO does not allow for multiple types of the same distribution. For instance, when trying to find a dataset that resembles the unit circle using polar coordinates, both the angle and radius can (should) be modelled using the uniform distribution. However, the radius would aim to focus around 1 with low variance, whilst the angle should have wide variance over the interval [0, 2*pi]. Hence, there is a problem.
The proposed solution is to allow different versions of the uniform distribution to be created. These are then used to create columns in an individual, and are updated to include only those present in the parents (#111).
The current state of EDO does not allow for multiple types of the same distribution. For instance, when trying to find a dataset that resembles the unit circle using polar coordinates, both the angle and radius can (should) be modelled using the uniform distribution. However, the radius would aim to focus around 1 with low variance, whilst the angle should have wide variance over the interval [0, 2*pi]. Hence, there is a problem.
The proposed solution is to allow different versions of the uniform distribution to be created. These are then used to create columns in an individual, and are updated to include only those present in the parents (#111).