GSoC 2020 Week-6

13 July 2020 - 3 mins read time
Tags: Sympy GSoC

This week, I spent most of my time reading about algorithms to parametrize algebraic curves and surfaces.

Parametrization of algebraic curves and surfaces

In many cases, it is easiar to define a region using an implicit equation over its parametric form. For example, a sphere can be defined with the implict eqution x2 + y2 + z2 = 4. Its parametric equation although easy to determine are tedious to write.

To integrate over implicitly defined regions, it is necessary to determine its parametric representation. I found the report on conversion methods between parametric and implicit curves from Christopher M. Hoffmann very useful. It lists several algorithms for curves and surfaces of different nature.

Every plane parametric curve can be expressed as an implicit curve. Some, but not all implicit curves can be expressed as parametric curves. Similarly, we can state of algebraic surfaces. Every plane parametric surface can be expressed as an implicit surface. Some, but not all implicit surfaces can be expressed as parametric surfaces.

Conic sections

One of the algorithm to parametrize conics is given below:

  1. Fix a point p on the conic. Consider the pencil of lines through p. Formulate the line equations.
  2. Substitute for y in the conic equation, solve for x(t).
  3. Use the fine equations to determine y(t).

The difficult part is to find a point on the conic. For the first implementation, I just iterated over a few points and selected the point which satisfied the equation. I need to implement an algorithm that can find a point on the curve.

The parametric equations determined by the algoithm are large and complex expressions. The vector_integrate function in most cases is taking too much time to calculate the integral. I tried fixing this using simplification functions in SymPy like trigsimp and expand_trig. The problem still persist.

>>> ellipse = ImplicitRegion(x**2/4 + y**2/16 - 1, x, y)
>>> parametric_region_list(ellipse)
[ParametricRegion((8*tan(theta)/(tan(theta)**2 + 4), -4 + 8*tan(theta)**2/(tan(theta)**2 + 4)), (theta, 0, pi))]
>>> vector_integrate(2, ellipse)
### It gets stuck

The algorithm does not work for curves of higher algebraic degree like cubics.

Monoids

A monoid is an algebraic curve of degree n that has a point of multiplicity n -1. All conics and singular cubics are monoids. Parametrizing a monoid is easy if the special point is known.

To parametrize a monoid, we need to determine a singular point of n - 1 multiplicity. The singular points are those points on the curve where both partial derivatives vanish. A singular point (xo, yo) of a Curve C is said to be of multiplicity n if all the partial derivatives off to order n - 1 vanish there. This paper describes an algorithm to calculate such a point but the algorithm is not trivial.

Next week’s goal

My goal is to complete the work on parametrizing conics. If I can find a simple algorithm to determine the singular point of the required multiplicity, I will start working on it.

Comments are closed