In[]:=
The intersection points correspond to equilibria, and we now know how to find them efficiently. But what of all the different regions? A moment's thought will convince you that within each region the compass direction of the underlying vector field is one of NE, NW, SE, or SW. Thus these regions can be a valuable tool in understanding and illustrating the orbit behavior.In[]:=
Certainly the image becomes much more informative when we superimpose some orbits. We show how to do that using NDSolve (the full VisualDSolve package allows a one-line treatment of such combined images).In[]:=
From now on we will generally suppress the Mathematica code. The preceding example should provide enough information for the reader to generate these as will as his or her own examples.In[]:=
Here is a view of some orbits, colored to represent increasing t (from red to blue).And here is an image of the CODEE teddy bears (cover of the Spring 1995 CODEE newsletter; also [BCB, p. 304]). This example shows that the algorithm is capable of dealing with a very complicated contour plot. This takes a bit of time (and memory) even on a fast computer! We are rewarded with a detailed view of the bear's phase plane and an unexpected glimpse of a latent bear: the contours at the bottom of the image form a bear's head!
In[]:=
Getting the shaded regions was not easy. Here is a bare outline of our algorithm. The main idea is to take the all the curves in the nullcline data produced by the contour plots and subdivide them at certain "critical" x-values. An x-value is critical if it is