OPTimization in PYthon (OPTPY)

This repository compares different alternatives to solve optimization problems in Python

We solve a standard linear programming problem of the form

min  c*x
s.t. A*x <= b
     x >= 0

where x is a vector with all variables and c, A, b are matrices of appropiate dimension whose elements are randomly generated using normal probability distributions.

Installation

git clone https://github.com/salvapineda/optpy.git

cd optpy

Method 1: SCIPY (link)

• Requirements: scipy
• Pros: It only requires importing the scipy package
• Cons: Optimization problem must be transform into matricial form

  # Import lp class
  from optpy import lp

  # Create linear programming probblem with 10 variables and 5 constraints. Further info: help(lp)
  lp1 = lp(nvar=10,ncon=5)

  # Solve linear problem using scipy. Further info: help(lp.solve_scipy)
  obj,x = lp1.solve_scipy(method='simplex')

Method 2: PULP (link)

• Requirements: pulp, glpk, cbc
• Pros: It relies on open-source solvers. Linear and integer problems can be solved. Intuitive syntax.
• Cons: It does not solve non-linear problems.

  # Import lp class
  from optpy import lp

  # Create linear programming probblem with 10 variables and 5 constraints. Further info: help(lp)
  lp1 = lp(nvar=10,ncon=5)

  # Solve linear problem using pulp. Further info: help(lp.solve_pulp)
  obj,x = lp1.solve_pulp()

Method 3: PYOMO Environ (standard) (link)

• Requirements: Pyomo and access to solvers
• Pros: It solves all kind of optimization problems. Intuitive syntax.
• Cons: It needs access to solvers, although neos server can be also used.

  # Import lp class
  from optpy import lp

  # Create linear programming probblem with 10 variables and 5 constraints. Further info: help(lp)
  lp1 = lp(nvar=10,ncon=5)

  # Solve linear problem using pyomo. Further info: help(lp.solve_pyomo_environ)
  obj,x = lp1.solve_pyomo_environ(neos=True,solver='cplex')

Method 4: PYOMO Kernel Library (link)

• Requirements: Pyomo and access to solvers
• Pros: It solves all kind of optimization problems. Lighter and more flexible than Pyomo environ.
• Cons: Less intuitive than Pyomo environ. Connection to neos server under development

  # Import lp class
  from optpy import lp

  # Create linear programming probblem with 10 variables and 5 constraints. Further info: help(lp)
  lp1 = lp(nvar=10,ncon=5)

  # Solve linear problem using pyomo. Further info: help(lp.solve_pyomo_kernel)
  obj,x = lp1.solve_pyomo_kernel(solver='cplex')

Do you want to contribute?

Any feedback is welcome so feel free to ask or comment anything you want via a Pull Request in this repo. If you need extra help, you can ask Salvador Pineda (spinedamorente@gmail.com).

Developed by

• Salvador Pineda - spinedamorente@gmail.com

Contributed by

• Miguel Angel Muñoz Diaz - miguelangeljmd@uma.es