search

s-leroux / fin

Set of tools for personal investment
MIT License
1 stars 0 forks source link

Auto-normalize solver's constraints #34

Closed s-leroux closed 7 months ago

s-leroux commented 7 months ago

Both with RandomSolver and ParticleSwarmSolver, when an equilibrium function gains significantly more weight than the other, the constraint solver fails to find an optimal solution.

We should provide a solution to roughly normalize the different constraints to evaluate close to the [-1, +1] range.

s-leroux commented 7 months ago

Being a 1-pass solver, I can't see how to fix RandomSolver for that limitation without creating a bias between the different points.

For a multi-pass solver like ParticleSwarmSolver, we can apply the same correction to all points of the same generation and then adjust from one generation to the next.

s-leroux commented 7 months ago

Some work was done in ff3cb0b05298402256c18eef2fcef209926304cb, but the preliminary results are not quite convincing: we lose precision in the general case without improvement in the difficult cases.

s-leroux commented 7 months ago

Both with RandomSolver and ParticleSwarmSolver, when an equilibrium function gains significantly more weight than the other, the constraint solver fails to find an optimal solution.

This statement is questionable and requires further investigation. Closing for now.