User Guide

Setup

  1. Download and install julia.
  2. Add the XPORTA.jl package.
julia> using Pkg; Pkg.add("XPORTA")

Simple Example

3-Simplex: Vertex Representation -> Halfspace Representation

Given a set of vertices, PORTA can find the linear equalities and inequalities bounding the convex hull of vertices.

Consider the vertices of the following 3-simplex (equilateral triangle).

\[\begin{matrix} \begin{matrix} v_1 = (1, 0, 0) \\ v_2 = (0, 1, 0) \\ v_3 = (0, 0, 1) \\ \end{matrix} & \rightarrow & \begin{bmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{bmatrix} \end{matrix}\]

The vertices are cartesian coordinates, $v_i = (x_i, y_i, z_i)$, and the right-hand-side matrix is constructed by stacking each vertex as a row in the matrix.

This code block demonstrates how to use XPORTA.jl to compute the halfspace representation of the 3-simplex.

using XPORTA

# Construct the vertex representation (POI) of the 3-simplex.
simplex_poi = POI(vertices = [1 0 0;0 1 0;0 0 1])

# Compute the halfspace representation (IEQ) with traf().
simplex_ieq = traf(simplex_poi)

# Print out the bounding linear equalities and inequalites.
println("Simplex Equalities: ", simplex_ieq.equalities)
println("Simplex Inequalities: ", simplex_ieq.inequalities)
Simplex Equalities: Rational{Int64}[1//1 1//1 1//1 1//1]
Simplex Inequalities: Rational{Int64}[0//1 -1//1 0//1 0//1; 0//1 0//1 -1//1 0//1; 0//1 1//1 1//1 1//1]

The equality represents a normalization constraint

\[x + y + z = 1.\]

The inequalities represent positivity constraints on $x$, $y$, and $z$.

\[\begin{matrix} -y \leq 0 & & y \geq 0 \\ -z \leq 0 & \rightarrow & z \geq 0 \\ y + z \leq 1 & & x \geq 0\\ \end{matrix}\]

The right-hand-side is realized by applying the the normalization constraint and performing some algebra.

PORTA is a Rational Solver

Methods accept matrices of type Int or Rational{Int}. All other types will result in a TypeError when constructing a POI or IEQ.