1---
2title: Commented TODO list for Python-MIP
3---
4
5Stopping criterion
6------------------
7
8Iterations/time without improving feasible solution
9similar to #40
10
11Numpy and Scipy support
12-----------------------
13
14Improvement #12
15
16Some problems are more easily specified in matrix notation.
17Input and output of problem models using numpy would allow an easier integration with numba, for instance, which would be great for maximum performance.
18
19Column generation
20-----------------
21
22Add example(s) with column generation
23
24Quadratic programming
25---------------------
26
27It would be easy to add the possibility of modelling x1*x2 if both are binary variables since they can be easily internally replaced by an auxiliary variable y that represents the linearization of this expression.
28
29Jupyter Notebooks
30-----------------
31
32Add some Jupyter notebooks with visualization of solutions (routes). Jupyter notebooks also allow a better output of tables for solution visualization. It seems that nbsphinx and jupyter-sphinx are good tools.
33
34Method to generate cutting planes
35---------------------------------
36
37It would be nice to be able to call the CBC procedures for generating specific types of cuts. This would ease the comparison of new cuts with the existing ones and also ease the development of cutting plane algorithms where one wants to evaluate the performance of a new type of cut when included in an existing cutting plane framework. One drawback is that it would be a CBC only feature since gurobi does not allows access to the generated cuts.
38
39Rapsberry pi
40------------
41
42Support this platform - right now no developer access to this hardware
43
44More CBC tests (in CBC)
45-----------------------
46
47The stability of Python-MIP depends directly on the stability of the COIN-OR CBC solver. Thus, it would be good to improve the automated tests in CBC, including modern CBC features such as lazy constraints in additional tests.
48
49- AVX optional in CBC
50- feasopt
51- cg examples
52- option to strengthen formulation with cuts
53- cplex support
54
55
56