Re: Displaying Shadow Prices using pulp

[Stuart Mitchell]

CBC now produces duals but you do have to keep hold of the constraint

for example:

from pulp import *

prob = LpProblem("test080", LpMinimize)

x = LpVariable("x", 0, 5)

y = LpVariable("y", -1, 1)

z = LpVariable("z", 0)

c1 = x+y <= 5

c2 = x+z >= 10

c3 = -y+z == 7

prob += x + 4*y + 9*z, "obj"

prob += c1, "c1"

prob += c2,"c2"

prob += c3,"c3"

prob.solve()

>>> c1.pi

0.0

>>> c2.pi

1.0

>>> c3.pi

8.0

On Wed, Jul 4, 2012 at 8:07 PM, Nigel Cleland <nigel....@gmail.com> wrote:

I'm currently trying to set up a fairly rudimentary generation dispatch problem using pulp.

I am currently able to solve the solution optimally using both the GLPK and CBC solvers.

I would however like to display the shadow prices/dual values so I can determine nodal pricing.

Any assistance would be highly appreciated

System Particulars:

Ubuntu 12.04

Python 2.7

Pulp 1.5.3

Solvers: GLPK/CBC

Thank you in advance,

Nigel

--
You received this message because you are subscribed to the Google Groups "pulp-or-discuss" group.
To view this discussion on the web visit https://groups.google.com/d/msg/pulp-or-discuss/-/ivjcz8n2FEYJ.
To post to this group, send email to pulp-or...@googlegroups.com.
To unsubscribe from this group, send email to pulp-or-discu...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/pulp-or-discuss?hl=en.

--

Stuart Mitchell

PhD Engineering Science

Extraordinary Freelance Programmer and Optimisation Guru

www.stuartmitchell.com

[Nigel Cleland]

Thank you, this is perfect

To post to this group, send email to pulp-or-discuss@googlegroups.com.
To unsubscribe from this group, send email to pulp-or-discuss+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/pulp-or-discuss?hl=en.

[Stuart Mitchell]

Is your problem contain integer variables? If so dual values are not produced as they do not make sense after branch and bound.

On 1 Jun 2013 19:57, "Sean Whipple" <seda...@gmail.com> wrote:

I am having the same issue and found this post and found it quite useful (one difference is I'm using CPLEX to solve).  I tested this solution and it works great.  But when I run on the problem I'm having the constraints don't have a .pi attribute after the solve (unlike these).  The problem has an optimal solution but my constraints don't have duals at the end.  Is there a flag that I need to set?

To unsubscribe from this group and stop receiving emails from it, send an email to pulp-or-discu...@googlegroups.com.

To post to this group, send email to pulp-or...@googlegroups.com.

Visit this group at http://groups.google.com/group/pulp-or-discuss?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Sean Whipple]

No they it's all continuous.  And if I run from cplex command line I can see the duals in the .sol file

[Sean Whipple]

Interesting follow up...when I ran the test above I did it with CPLEX 12.4...when I ran my own code it was on a machine with CPLEX 12.5.  Is it possible that it's no longer supported?  (both using the same version of PuLP)

[Stuart Mitchell]

Could you please run the test on the computer with CPLEX 12.5.

Also can you post the results of

>>> import pulp

>>> pulp.pulpTestAll()

on the machine that is having problems.

Thanks

--

[fatzm...@gmail.com]

Hello,

I am using pulp with cplex to solve a column generation problem and I need to use the dual variables. I try the constraint.pi but it always returns None even with the small example that has been posted on this discussion. Any idea why?

Thanks in advance

[Stuart Mitchell]

Are you using the cplex command line or cplex through the python interface?

Stu

--

You received this message because you are subscribed to the Google Groups "pulp-or-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pulp-or-discu...@googlegroups.com.
To post to this group, send email to pulp-or...@googlegroups.com.

Visit this group at http://groups.google.com/group/pulp-or-discuss.
For more options, visit https://groups.google.com/d/optout.

[Nathan Petty]

Hey Stu,  what about the variable bound constraints automatically generated when you declare a variable?  How do I get those shadow price?  For example I have a continuous variable with bounds [0,1] and I need the shadow price for those constraints.   

Am I going to have to explicitly declare those bounds as constraints?

The other option: For variable bounds in an LP, the shadow price equals the reduced cost on the variable.  So there's no need to add constraints?; just retrieve the reduced cost on the variable?

[Nathan Petty]

To answer my own question, I did just use explicit constraints for bounds to get shadows

[Rebecca Jeffers]

Hi :)

I am having this same problem using cplex with pulp on a simple problem where all the varlables are continuous. I am using the latest version of pulp with cplex 12.8

[Rebecca Jeffers]

It seems that it couldn't find CPLEX_CMD because I had not put the lib, bin and include paths separately in my PATH variable. CPLEX() was therefore using CPLEX_PY and this gave "None"s for the dual variables. Once I had corrected the path I used CPLEX_CMD directly and there were values for the duals

[GiaLLuh]

Hi Stu,

I'm having the same problem using the GUROBI_CMD solver (none output).

I know that Gurobi has its own syntax to retrieve shadow prices for constraints (.Pi attribute), but i have already written my model with the pulp syntax since it gets along well with pandas dataframes which I'm importing my input data from.

Do I have to change the variable path into the solver option like Rebecca suggested? Where? How?

Thanks

Il giorno lunedì 19 gennaio 2015 21:31:21 UTC+1, Stuart Mitchell ha scritto:

Are you using the cplex command line or cplex through the python interface?

Stu

On Wed, Jan 14, 2015 at 6:11 AM, <fatzm...@gmail.com> wrote:

Hello,

I am using pulp with cplex to solve a column generation problem and I need to use the dual variables. I try the constraint.pi but it always returns None even with the small example that has been posted on this discussion. Any idea why?

Thanks in advance

--
You received this message because you are subscribed to the Google Groups "pulp-or-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to pulp-or...@googlegroups.com.

To post to this group, send email to pulp-or...@googlegroups.com.

Visit this group at http://groups.google.com/group/pulp-or-discuss.
For more options, visit https://groups.google.com/d/optout.

[GiaLLuh]

Problem solved using the pulp.GUROBI() solver instead of pulp.GUROBI_CMD() solver. Calling the name of the constraint plus .solverConstraint plus .pi or .Pi gives the dual value.