MixedIntegerLinearProgram method .get_values() returns "IndexError: list index out of range"
I'm writing an MLP to solve a scheduling problem for a set of jobs. I have several variable types, including family 's_ij', which is a boolean indicating if time i occurs before or after time j. My program works, and returns feasible solutions. I'm trying to see the assigned values for each variable, and while I can see the start time for each job, and some other families of booleans, when I try to display the values for the 's_ij' variables, I get an "IndexError". This is weird, since I'm letting the MILP package do all the work here. Here are relevant sections of the code, as well as the output:
from itertools import product ... J = range(len(JOBS)) I = range(2*len(JOBS)) IXJ_copy = product(I, J) ... for pair in IXJ_copy: i = pair j = pair if i == len(I) - 1: scheduling_mlp.add_constraint(s[(i, j)] <= 0) else: scheduling_mlp.add_constraint(M*(1-s[(i, j)]) + b[i+1] - t[j] - EPSILON >= 0) scheduling_mlp.add_constraint(M*s[(i, j)] + t[j] - b[i+1] >= 0) scheduling_mlp.add_constraint(M*e[(i, j)] + b[i] - t[j] - T(j) >= 0) scheduling_mlp.add_constraint(M*(1-e[(i, j)]) + t[j] + T(j) - b[i] - EPSILON >= 0) scheduling_mlp.add_constraint(M*(1-u[(i, j)]) + s[(i, j)] + e[(i, j)] >= 2) scheduling_mlp.add_constraint(-M*(u[(i, j)]) + s[(i, j)] + e[(i, j)] <= 1) ... print("s:") start = scheduling_mlp.get_values(s) print(start)
Traceback (most recent call last): File "lp_solver.py", line 186, in <module> start = scheduling_mlp.get_values(s) File "mip.pyx", line 1161, in sage.numerical.mip.MixedIntegerLinearProgram.get_values (sage/numerical/mip.c:7098) IndexError: list index out of range
As I said, when I run the code, I get feasible solutions back. I have the case handling in place while defining my constraints to prevent b[i+1] from going out of range. I'm able to get_values and print all other variables. Any ideas what's going wrong here?