shaunharker / chomp Goto Github PK

On Mac OS X, after installing Anaconda based Python my CHomP installation breaks. Many errors of the form

/usr/local/include/CImg.h:7178:31: error: use of undeclared identifier
      'XOpenDisplay'
        Display *const _dpy = XOpenDisplay(0);

are thrown. Googling this we find https://sourceforge.net/p/dclib/discussion/442518/thread/57f9b43f/ which suggests the problem is related to Anaconda. Some testing reveals that indeed the include path being set in CMake X11_X11_INCLUDE_PATH is to the anaconda include directory. This problem halts the installer entirely.

Allow the user to specify periodic boundary conditions in one or more dimensions for a cubical complex.

Compiler error occurs in file included from include/chomp/Ring.h.

As a simple test case, I am trying to compute the homology with generators of the chain complex:

C_0 = { v }
C_1 = { a, b }
C_2 = { c, d }

with boundary:

\partial (c) = b

and all others equal to zero. I have the following .mat file:

0
0 1 0
0 0 0
1
0 1 0
1 0 1
0 0 0
1 1 0

Using the command chomp-matrix myfile.mat -g, I obtain the output:

Betti Numbers: 1 1 1 
------------
| Homology |
------------
  H_0 = ℤ^1
  H_1 = ℤ^1
  H_2 = ℤ^1

--------------
| Generators |
--------------
H_0:
  (0) : 1[0]
H_1:
  (0) : 1[1]
H_2:
  (0) : 1[0]

While the Betti Numbers and vector spaces are as expected, the generators are not. For H_1, the [1] index refers to b, which is in the image of boundary and thus is reducible to the trivial group in homology. Furthermore, the generator for H_2 is given as c, which is not a cycle and so can't be a generator.

I suppose I must be doing something incorrectly, and would appreciate any guidance as to how to correct it.

As a note, when I use similar incidence matrices to generate a CellComplex in Sagemath, and then ask for homology with generators, I obtain a reasonable result and as far as I know, Sagemath is just interfacing back to CHomP (although I think it is using the old version).