Groups keyboard shortcuts have been updated
Dismiss
See shortcuts

Stochastic programming with continuous probability density function

45 views
Skip to first unread message

mitsuru

unread,
Apr 21, 2025, 11:07:32 AMApr 21
to Pyomo Forum
Dear all

I am going to use pyomo, because it contains stochastic programming packages.
However, I cannot understand whether it can handle my problems.

As my problems, linear-programming forms  (min c^T x) with coefficients (c) of random variables, and their continuous probability density functions (c ~ f(omega)) are known.

Thus, expectation values are integral form (E{c^T x} = integral{c^T x * f(omega)}d_omega).
Is it handled by pyomo with its stochastic programming modules?

David Woodruff

unread,
Apr 21, 2025, 12:01:38 PMApr 21
to pyomo...@googlegroups.com
The short answer is "yes."

If you want to optimize against a worst-case (in a well-defined sense) use Pyros https://pyomo.readthedocs.io/en/6.8.0/contributed_packages/pyros.html
If you want optimize the expected objective function value use mpi-sppy https://github.com/Pyomo/mpi-sppy
(You will need to write code that that samples from your probability distribution)


--
You received this message because you are subscribed to the Google Groups "Pyomo Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pyomo-forum...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pyomo-forum/f929ed1d-7d8b-40ed-8a09-2502b7b604a5n%40googlegroups.com.

西川満

unread,
Apr 21, 2025, 4:40:41 PMApr 21
to pyomo...@googlegroups.com
Dear David Woodruff

Thank you so much for your kind reply.
What I need is expected objective function value,
so that I have to select mpi-sppy.

But it is one problem. Mpi-sppy seems to be developing stage and beta release,
there are little documents and how to solve my own problems.

If you know mpi-sppy already runs robustly and correctly,
I would appreciate if you could tell me.
(Or, should I go back to pysp?)

Best regards

 


2025年4月22日(火) 1:01 David Woodruff <david.l....@gmail.com>:
You received this message because you are subscribed to a topic in the Google Groups "Pyomo Forum" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pyomo-forum/ratfUm8ruAs/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pyomo-forum...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pyomo-forum/CAERj1nWqUWQjReQ1HWQf-uZ76zpadxZd-gWqc81vvPYt1FH%2BZA%40mail.gmail.com.

David Woodruff

unread,
Apr 21, 2025, 5:42:48 PMApr 21
to pyomo...@googlegroups.com
You should use mpi-sppy rather than PySP.

西川満

unread,
Apr 21, 2025, 7:52:21 PMApr 21
to pyomo...@googlegroups.com
Dear Dr. D. Woodruff

Thank you for your advice. There are some examples in mpi-sppy github, I will investigate it for my use.

BTW, do you know if the project of mpi-sppy continues steadily for a long span?
I would appreciate it if you could tell me.

Kind regards



2025年4月22日(火) 6:42 David Woodruff <david.l....@gmail.com>:

David Woodruff

unread,
Apr 21, 2025, 10:06:16 PMApr 21
to pyomo...@googlegroups.com

西川満

unread,
Apr 22, 2025, 3:47:27 AMApr 22
to pyomo...@googlegroups.com
Thank you greatly, I will consider its use.

Kind regards

2025年4月22日(火) 11:06 David Woodruff <david.l....@gmail.com>:

西川満

unread,
Apr 22, 2025, 7:07:34 AMApr 22
to pyomo...@googlegroups.com
Dear Dr. D. Woodruff

I read the first paper about mpi-sppy, and then I understand you are the member of development in mpi-sppy.

Based on that, I would appreciate it if you could give me a lecture.
Stochastic programming seems to have vast applications, particularly for energy planning and financial decisions.
My questions are
(1) The transition from pysp to mpi-sppy is probably because pysp succeeded to get some academic or practical applications. If you know some examples, could you tell me?
(2) The reason why mpi-sppy is now developed should be due to computational complexity of stochastic programming, parallel computing power resolve this. I could not imagine the problem which needs so large scenarios. What is the ongoing main target for mpi-sppy?

Best regards 





2025年4月22日(火) 16:47 西川満 <mit3...@gmail.com>:
Reply all
Reply to author
Forward
0 new messages