Skip to main content
3 votes

Reasonable implementation of a MILP scheduling problem

Your formulation looks correct, but the objective is nonlinear because it includes a $\max$ function. You can linearize by introducing nonnegative variables $z_j$, imposing linear constraints $z_j \ge ...
RobPratt's user avatar
  • 33.6k
3 votes
Accepted

Column generation for bin packing problem with variable size minimizing the largest cost of any class

Yes, you can. Apply Dantzig-Wolfe decomposition to your compact formulation, with the second and third sets of constraints as your subproblem, which then decomposes into one block per bin $b$.
RobPratt's user avatar
  • 33.6k
3 votes

Is there any way to use Lazy constraints in Pyomo?

Regarding HiGHS, it does not support lazy constraints at the moment, but this is work in progress. They will be available in one of the next releases. Once they are available, you will be able to use ...
filikat's user avatar
  • 76
3 votes
Accepted

Implication linking triplets of non-negative real or integer variables

Part i) was already answered here: https://or.stackexchange.com/a/12403/500 For part ii), let $\bar{x}$, $\bar{y}$, and $\bar{z}$ be constant upper bounds on $x,y,z$, respectively. Let $\epsilon$ be ...
RobPratt's user avatar
  • 33.6k
3 votes

Deviation from the optimal solution for Solomon instances of CVRPTW

Here is how you should compute the distance d between node i and node j: ...
fontanf's user avatar
  • 2,673
2 votes
Accepted

Relation linking pairs of non-negative continuous or integer variables

Let $\bar{x}$ and $\bar{y}$ be constant upper bounds on $x$ and $y$, respectively. Let $\epsilon$ be a small positive constant. Introduce binary variables $z_i$, and impose \begin{align} z_1 + z_2 + ...
RobPratt's user avatar
  • 33.6k
2 votes
Accepted

Are there ways to assist Gurobi in finding good solutions?

You can use a callback to get each new Gurobi solution $x$ and tell Gurobi about your improved solution $g(x)$. This article illustrates it in the context of the Python API. (It might be a bit, um, ...
prubin's user avatar
  • 40.1k
2 votes
Accepted

Modeling the relation between pairs of Non-negative Continuous or Integer Variables

Assume $x \le M$ and $y \le M$, and let $\epsilon>0$ be a small constant tolerance. For i), introduce binary decision variables $u,v,w$, and impose \begin{align} x &\le M u \tag1\label1 \\ y &...
RobPratt's user avatar
  • 33.6k
2 votes
Accepted

Casting the truth value of an inequality to a boolean

Equivalently, you want to enforce the contrapositives \begin{align} b=0 &\implies d<0 \\ b=1 &\implies d\ge 0 \end{align} For $d<0$, rewrite as $d\le -\epsilon$ for some positive ...
RobPratt's user avatar
  • 33.6k
1 vote

How to Implement a Piecewise Function Using SOS2 in GAMS Without Binary Variables in an MILP Model?

This is not a piecewise linear function, as expected by a SOS2 approach. SOS2 assumes that breakpoints are fixed and that there is some interpolation to do. You have a piecewise constant function ...
Erwin Kalvelagen's user avatar
1 vote

How to Implement a Piecewise Function Using SOS2 in GAMS Without Binary Variables in an MILP Model?

If your $h0()$ function only takes two possible values (0.001 or 0.3), an SOS2 constraint would be inappropriate. You would need to use a binary variable. In any case, a solver that allows SOS ...
prubin's user avatar
  • 40.1k
1 vote

Are there ways to assist Gurobi in finding good solutions?

I would suggest watching Faster MIPs Using Custom Heuristics by Gurobi's Greg Glockner.
Riley's user avatar
  • 175
1 vote
Accepted

Infeasibility of subproblem in Combinatorial Benders Decomposition

If imposing $u^T \bar{b}=1$ makes the dual problem infeasible, it means that $u=0$ is the only feasible solution to the original dual and there is no certificate of infeasibility for the primal. That ...
RobPratt's user avatar
  • 33.6k
1 vote
Accepted

Modeling the implication between pairs of Non-negative Continuous or Integer Variables

The first disjunct, $(x \leq y \implies a \leq b)$, would be linearized as the following form: \begin{align*} \ x-y \leq M(1-z_1) \\ \ a-b \leq M(1-z_2) \\ \ z_1 + z_2 = 1 \\ \ Lb \leq x, y, a, b \leq ...
A.Omidi's user avatar
  • 9,248
1 vote

Reference for column generation applications

Another recently published reference by Eduardo Uchoa, Artur Pessoa, and Lorenza Moreno is Optimizing with Column Generation. We are excited to present the early release of Part I of our book “...
A.Omidi's user avatar
  • 9,248

Only top scored, non community-wiki answers of a minimum length are eligible