Pareto surface navigator

GUI for interactively navigating high dimensional convex Pareto surfaces

This builds a navigation GUI for navigating n-dimensional convex Pareto surfaces interactively. It requires linprog from the
optimization toolbox, but can probably be replaced easily with a free version. The linear programs solved during navigation are very small and easy. It is assumed that all objectives are 'minimize' objectives. Thus normal navigation mode is to pull sliders downward.

mnav(pSurf) is how you call it. pSurf is an mxn matrix, (the pareto Surface) where m is the numberof points on the surface, and n is the dimension of the space.

mnav sacrifices 'stay exactly on the Pareto surface' for speed by allowing the current location to be anywhere in the convex hull
of the points in pSurf. When a slider request is made, the navigator takes a path toward the pSurf and then bending downwards to the extreme solution for that request. One can adjust the balance between shooting for the Pareto surface and
shooting directly for the anchor solution (the anchor points are the best points in each dimension) by adjusting the parameter g
within calc_j_trajectory.m.

Clamping checkboxes allow any dimension to be locked at its current value. A black marker shows up to the left of the slider when a clamp is on.

This software was developed by David Craft at Mass General Hospital for use in navigating Pareto surfaces for multi-objective intensity modulated radiotherapy (IMRT) treatment planning.