Introduce curvature regulators #55

Open
opened 2025-02-25 19:48:13 +00:00 by Vectornaut · 1 comment
Member

Goal

Introduce a kind of regulator that measures and constrains the curvature of a sphere.

Context

As of pull request #46, the application prototype only supports one kind of regulator, which measures and constrains the inversive distance between spheres. We'll eventually need to support various kinds of regulators which have various numbers of subjects and connect to the engine in various ways. A curvature regulator would be a good first step, because:

  • It differs from the inversive distance regulator in several ways.
    • It has one subject instead of two.
    • It connects to the engine through a frozen representation vector entry rather than a Gram matrix entry.
  • It's needed to solve Irisawa's problem, which we're planning to use in a tech demo.
### Goal Introduce a kind of regulator that measures and constrains the curvature of a sphere. ### Context As of pull request #46, the application prototype only supports one kind of regulator, which measures and constrains the inversive distance between spheres. We'll eventually need to support various kinds of regulators which have various numbers of subjects and connect to the engine in various ways. A curvature regulator would be a good first step, because: - It differs from the inversive distance regulator in several ways. - It has one subject instead of two. - It connects to the engine through a frozen representation vector entry rather than a Gram matrix entry. - It's needed to solve Irisawa's problem, which we're planning to use in a [tech demo](projects/2).
Vectornaut added this to the Demo project 2025-02-25 19:48:13 +00:00
Vectornaut added a new dependency 2025-03-01 00:38:06 +00:00
Owner

Per this comment we definitely plan for the resolution of this to make the subjects of a regulator be an array rather than a tuple, or possibly some sort of set rather than a tuple (if we think it will be the case that the subjects are always intrinsically unordered in the semantics of any constraint, as they are in an inversive distance regulator), but in any case some collection that can be iterated over.

Per [this comment](https://code.studioinfinity.org/glen/dyna3/pulls/48#issuecomment-2188) we definitely plan for the resolution of this to make the subjects of a regulator be an array rather than a tuple, or possibly some sort of set rather than a tuple (if we think it will be the case that the subjects are always intrinsically unordered in the semantics of any constraint, as they are in an inversive distance regulator), but in any case some collection that can be iterated over.
glen added the
enhancement
label 2025-03-12 22:17:29 +00:00
Sign in to join this conversation.
No description provided.