PyEPO: A PyTorch-based End-to-End Predict-then-Optimize Tool

Learning Framework

Publication

This repository is the official implementation of the paper: PyEPO: A PyTorch-based End-to-End Predict-then-Optimize Library for Linear and Integer Programming (Accepted to Mathematical Programming Computation (MPC))

Citation:

@article{tang2024,
  title={PyEPO: a PyTorch-based end-to-end predict-then-optimize library for linear and integer programming},
  author={Tang, Bo and Khalil, Elias B},
  journal={Mathematical Programming Computation},
  issn={1867-2957},
  doi={10.1007/s12532-024-00255-x},
  year={2024},
  month={July},
  publisher={Springer}
}

Introduction

PyEPO (PyTorch-based End-to-End Predict-then-Optimize Tool) is a Python-based, open-source software that supports modeling and solving predict-then-optimize problems with the linear objective function. The core capability of PyEPO is to build optimization models with GurobiPy, Pyomo, or any other solvers and algorithms, then embed the optimization model into an artificial neural network for the end-to-end training. For this purpose, PyEPO implements various methods as PyTorch autograd modules.

Documentation

The official PyEPO docs can be found at https://khalil-research.github.io/PyEPO.

Slides

Our recent tutorial was at the ACC 2024 conference. You can view the talk slides here.

Tutorial

• 01 Optimization Model: Build up optimization solver
• 02 Optimization Dataset: Generate synthetic data and use optDataset
• 03 Training and Testing: Train and test different approaches
• 04 2D knapsack Solution Visualization: Visualize solutions for knapsack problem
• 05 Warcraft Shortest Path: Use the Warcraft terrains dateset to train shortest path
• 06 Real-World Energy Scheduling: Apply PyEPO to real energy data
• 07 kNN Robust Losses: Use the optDatasetKNN for robust losses

Experiments

To reproduce the experiments in the original paper, please use the code and follow the instructions in this branch.

Features

• Implement SPO+ [1], DBB [3], NID [7], DPO [4], PFY [4], NCE [5] and LTR [6], I-MLE [8], AI-MLE [9], and PG [11].
• Support Gurobi, COPT, and Pyomo API
• Support Parallel computing for optimization solver
• Support solution caching [5] to speed up training
• Support kNN robust loss [10] to improve decision quality

Installation

Clone and Install from this Repo

You can download PyEPO from our GitHub repository.

git clone -b main --depth 1 https://github.com/khalil-research/PyEPO.git

And install it.

pip install PyEPO/pkg/.

Pip Install

The package is now available for installation on PyPI. You can easily install PyEPO using pip by running the following command:

pip install pyepo

Conda Install

PyEPO is also available on Anaconda Cloud. If you prefer to use conda for installation, you can install PyEPO with the following command:

conda install -c pyepo pyepo

Dependencies

• NumPy
• SciPy
• Pathos
• tqdm
• Pyomo
• Gurobi
• Scikit Learn
• PyTorch

Sample Code

#!/usr/bin/env python
# coding: utf-8

import gurobipy as gp
from gurobipy import GRB
import numpy as np
import pyepo
from pyepo.model.grb import optGrbModel
import torch
from torch import nn
from torch.utils.data import DataLoader


# optimization model
class myModel(optGrbModel):
    def __init__(self, weights):
        self.weights = np.array(weights)
        self.num_item = len(weights[0])
        super().__init__()

    def _getModel(self):
        # ceate a model
        m = gp.Model()
        # varibles
        x = m.addVars(self.num_item, name="x", vtype=GRB.BINARY)
        # model sense
        m.modelSense = GRB.MAXIMIZE
        # constraints
        m.addConstr(gp.quicksum([self.weights[0,i] * x[i] for i in range(self.num_item)]) <= 7)
        m.addConstr(gp.quicksum([self.weights[1,i] * x[i] for i in range(self.num_item)]) <= 8)
        m.addConstr(gp.quicksum([self.weights[2,i] * x[i] for i in range(self.num_item)]) <= 9)
        return m, x


# prediction model
class LinearRegression(nn.Module):

    def __init__(self):
        super(LinearRegression, self).__init__()
        self.linear = nn.Linear(num_feat, num_item)

    def forward(self, x):
        out = self.linear(x)
        return out


if __name__ == "__main__":

    # generate data
    num_data = 1000 # number of data
    num_feat = 5 # size of feature
    num_item = 10 # number of items
    weights, x, c = pyepo.data.knapsack.genData(num_data, num_feat, num_item,
                                                dim=3, deg=4, noise_width=0.5, seed=135)

    # init optimization model
    optmodel = myModel(weights)

    # init prediction model
    predmodel = LinearRegression()
    # set optimizer
    optimizer = torch.optim.Adam(predmodel.parameters(), lr=1e-2)
    # init SPO+ loss
    spop = pyepo.func.SPOPlus(optmodel, processes=1)

    # build dataset
    dataset = pyepo.data.dataset.optDataset(optmodel, x, c)
    # get data loader
    dataloader = DataLoader(dataset, batch_size=32, shuffle=True)

    # training
    num_epochs = 10
    for epoch in range(num_epochs):
        for data in dataloader:
            x, c, w, z = data
            # forward pass
            cp = predmodel(x)
            loss = spop(cp, c, w, z)
            # backward pass
            optimizer.zero_grad()
            loss.backward()
            optimizer.step()

    # eval
    regret = pyepo.metric.regret(predmodel, optmodel, dataloader)
    print("Regret on Training Set: {:.4f}".format(regret))

Reference

• [1] Elmachtoub, A. N., & Grigas, P. (2021). Smart “predict, then optimize”. Management Science.
• [2] Mandi, J., Stuckey, P. J., & Guns, T. (2020). Smart predict-and-optimize for hard combinatorial optimization problems. In Proceedings of the AAAI Conference on Artificial Intelligence.
• [3] Vlastelica, M., Paulus, A., Musil, V., Martius, G., & Rolínek, M. (2019). Differentiation of blackbox combinatorial solvers. arXiv preprint arXiv:1912.02175.
• [4] Berthet, Q., Blondel, M., Teboul, O., Cuturi, M., Vert, J. P., & Bach, F. (2020). Learning with differentiable pertubed optimizers. Advances in neural information processing systems, 33, 9508-9519.
• [5] Mulamba, M., Mandi, J., Diligenti, M., Lombardi, M., Bucarey, V., & Guns, T. (2021). Contrastive losses and solution caching for predict-and-optimize. Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence.
• [6] Mandi, J., Bucarey, V., Mulamba, M., & Guns, T. (2022). Decision-focused learning: through the lens of learning to rank. Proceedings of the 39th International Conference on Machine Learning.
• [7] Sahoo, S. S., Paulus, A., Vlastelica, M., Musil, V., Kuleshov, V., & Martius, G. (2022). Backpropagation through combinatorial algorithms: Identity with projection works. arXiv preprint arXiv:2205.15213.
• [8] Niepert, M., Minervini, P., & Franceschi, L. (2021). Implicit MLE: backpropagating through discrete exponential family distributions. Advances in Neural Information Processing Systems, 34, 14567-14579.
• [9] Minervini, P., Franceschi, L., & Niepert, M. (2023, June). Adaptive perturbation-based gradient estimation for discrete latent variable models. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 37, No. 8, pp. 9200-9208).
• [10] Schutte, N., Postek, K., & Yorke-Smith, N. (2023). Robust Losses for Decision-Focused Learning. arXiv preprint arXiv:2310.04328.
• [11] Gupta, V., & Huang, M. (2024). Decision-Focused Learning with Directional Gradients. Training, 50(100), 150.