Purpose

---

  This function computes the potential energy of a mass
  distribution. It uses a divide and conquer algorithm based on the
  Barnes-Hut algorithm, and scales as N(log(N)). The poten_slow
  program is more accurate, but scales as N^2. Generally, this
  procedure will calculate energies accurate to 1%

Inputs

---

  pos: A [3, n] array of 3D particle locations
  mass: A n element vector of masses

Keyword Parameters

---

  theta: A precision pramater which controls the algorithm. Higher
  values translate to faster run time and larger errors. A value of 1
  is recommended, and usually achieves 1% accuracy. A value of 1.5
  achieves 1% accuracy for >100 evenly distributed particles. Default
  is 1

Outputs

---

  The potential energy of the system. It is assumed that G=1, so that
  PE = sum_i (sum j > i (m_i * m_j / r_ij) )

Modification History

---

  July 2010: Written by Chris Beaumont.