These page comprises some demos given at a course on Discrete Optimization at TU Darmstadt, Summer 2007. Some things are more closely related to optimization than others.
This page is under construction: It will grow until the end of the term.
This is the first demo whose purpose is to show the principal behavior of polymake.
The following commands produce a 3-dimensional cube (whose vertices have (+/-1)-coordinates). It is visualized, and some combinatorial properties are computed. Having the output sections numbered makes the implicit numbering scheme explicit.
Zonotopes are Minkowski sums of line segments (or, equivalently, affine projections of a regular cube). The section ZONOTOPE_INPUT_VECTORS serves as an input section to define a zonotope. For the sake of the compatibility with other polymake functions each line is a vector in homogeneous coordinates (this time always starting with a non-zero entry).
For instance a file z.poly containing the lines
defines the polytope [0,e1]+[0,e2]+[0,e3]+[0,e1+e2+e3].
How does a random zonotope look alike? Well here is one version. We start with the (convex hull of) random points on the unit sphere. And then we write a short script, say z_script, which takes the VERTICES of the random polytope and take them as ZONOTOPE_INPUT_VECTORS for a new polytope which is then visualized.
The picture to the right is what we obtain if we apply JavaView's function Method->Effect->Unfold Geometry ... to our zonotope. There is also a postscript version for better visibility. | ![]() |