Decision Optimization

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization

Yes, this function can be used to create at most two nodes.  But you can work around this limitation by creating, at each step, a 'real' node, and a 'fake' one.  In the fake node, you will create a second 'real' node, and yet another fake node, etc.  This will let you create as many 'real' nodes as you want, as long as you actually organise them in a binary tree.

------------------------------
Xavier
------------------------------

Original Message:
Sent: Wed September 23, 2020 08:45 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization

Thanks for the reply!

I am wondering how I can recognize the fake node later on in my code? Possibly I should create a dummy variable for this purpose?

------------------------------
Amin
------------------------------

Original Message:
Sent: Thu September 24, 2020 06:33 AM
From: Xavier Nodet
Subject: branching on constraint and limitation on offspring nodes

Yes, this function can be used to create at most two nodes.  But you can work around this limitation by creating, at each step, a 'real' node, and a 'fake' one.  In the fake node, you will create a second 'real' node, and yet another fake node, etc.  This will let you create as many 'real' nodes as you want, as long as you actually organise them in a binary tree.

------------------------------
Xavier

Original Message:
Sent: Wed September 23, 2020 08:45 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization

In support of Xavier's answer, I actually did a series of three blog posts (the first one ten years ago), including sample code (using the Java API):

• When the OctoMom Solves MILPs
• Birthing a Litter in CPLEX
• Multiple Children - Again!

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University
------------------------------

Original Message:
Sent: Wed September 23, 2020 08:45 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization

Thanks Paul for the reply!
Unfortunately, I am not familiar with Java, so I could not figure out your code. Please look at my above question in reply to Xavier as well! Another challenge that I am probably facing is that this is like one-time branching that I wish it to occur only in the root node. The root node has an index of zero, so it is easy to recognize, but I guess the indexing of the other nodes in the tree happens more in a random fashion, so I cannot recognize at what level of the B&B tree I am now.

------------------------------
Amin
------------------------------

Original Message:
Sent: Thu September 24, 2020 05:17 PM
From: Paul Rubin
Subject: branching on constraint and limitation on offspring nodes

In support of Xavier's answer, I actually did a series of three blog posts (the first one ten years ago), including sample code (using the Java API):

• When the OctoMom Solves MILPs
• Birthing a Litter in CPLEX
• Multiple Children - Again!

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University

Original Message:
Sent: Wed September 23, 2020 08:45 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization

I'm not familiar with the Python API, but I assume it has the following in common with the Java API. CPLEX assigns a unique ID to every node, with the ID of the root node being zero. The various methods for creating branches in a callback all return the ID number of the child node being created, and there are methods to get the ID of the current node. So you just create a memory structure in your callback that associates ID numbers with information about the node (what constraints you used to create the node, or what your next step at that node should be, or whatever). You only need to insert entries there for nodes that need follow-up work (such as the dummy children). If your callback encounters a node whose ID is not in the table, it assumes this node is further down the tree and just lets CPLEX branch as it wishes there.

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University
------------------------------

Original Message:
Sent: Sat September 26, 2020 08:15 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Thanks Paul for the reply!
Unfortunately, I am not familiar with Java, so I could not figure out your code. Please look at my above question in reply to Xavier as well! Another challenge that I am probably facing is that this is like one-time branching that I wish it to occur only in the root node. The root node has an index of zero, so it is easy to recognize, but I guess the indexing of the other nodes in the tree happens more in a random fashion, so I cannot recognize at what level of the B&B tree I am now.

------------------------------
Amin

Original Message:
Sent: Thu September 24, 2020 05:17 PM
From: Paul Rubin
Subject: branching on constraint and limitation on offspring nodes

In support of Xavier's answer, I actually did a series of three blog posts (the first one ten years ago), including sample code (using the Java API):

• When the OctoMom Solves MILPs
• Birthing a Litter in CPLEX
• Multiple Children - Again!

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University

Original Message:
Sent: Wed September 23, 2020 08:45 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization

Thanks Paul!
I did not know that we could get the ID of the created child node.

------------------------------
Amin
------------------------------

Original Message:
Sent: Sun September 27, 2020 02:33 PM
From: Paul Rubin
Subject: branching on constraint and limitation on offspring nodes

I'm not familiar with the Python API, but I assume it has the following in common with the Java API. CPLEX assigns a unique ID to every node, with the ID of the root node being zero. The various methods for creating branches in a callback all return the ID number of the child node being created, and there are methods to get the ID of the current node. So you just create a memory structure in your callback that associates ID numbers with information about the node (what constraints you used to create the node, or what your next step at that node should be, or whatever). You only need to insert entries there for nodes that need follow-up work (such as the dummy children). If your callback encounters a node whose ID is not in the table, it assumes this node is further down the tree and just lets CPLEX branch as it wishes there.

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University

Original Message:
Sent: Sat September 26, 2020 08:15 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Thanks Paul for the reply!
Unfortunately, I am not familiar with Java, so I could not figure out your code. Please look at my above question in reply to Xavier as well! Another challenge that I am probably facing is that this is like one-time branching that I wish it to occur only in the root node. The root node has an index of zero, so it is easy to recognize, but I guess the indexing of the other nodes in the tree happens more in a random fashion, so I cannot recognize at what level of the B&B tree I am now.

------------------------------
Amin

Original Message:
Sent: Thu September 24, 2020 05:17 PM
From: Paul Rubin
Subject: branching on constraint and limitation on offspring nodes

In support of Xavier's answer, I actually did a series of three blog posts (the first one ten years ago), including sample code (using the Java API):

• When the OctoMom Solves MILPs
• Birthing a Litter in CPLEX
• Multiple Children - Again!

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University

Original Message:
Sent: Wed September 23, 2020 08:45 PM
From: Amin
Subject: branching on constraint and limitation on offspring nodes

Hello all,

I have used python API for coding an MIP, and I am trying to branch on certain constraints using .make_branch.
However,  if I create more than two subproblems in each node, I will receive an error. Is there any limitation on the number of offspring nodes that can be created in each node?

Thanks,

------------------------------
Amin
------------------------------

#DecisionOptimization