Unfortunately, I’m not familiar enough with LP to know what you mean by that 
statement.






Sent from Windows Mail





From: Ted Dunning
Sent: ‎Thursday‎, ‎May‎ ‎22‎, ‎2014 ‎9‎:‎13‎ ‎PM
To: Reginald Johnson
Cc: [email protected]







Is it possible that you encode the symmetry constraint into the function itself 
by only allowing adaptation of the upper triangle?










On Thu, May 22, 2014 at 8:36 PM, Reginald Johnson <[email protected]> 
wrote:



I agree, and in fact my original formulation used that same format (A_ij=A_ji) 
for the constraint.  However, I didn't (and still don't) see a way to create a 
constraint in the optimization class that will let me use anything other than a 
number for the right hand side.  

  Do you know if this library has this ability?



On May 22, 2014 8:33 AM, "Ted Dunning" <[email protected]> wrote:

In electrical and fluid flow problems, it is customary to encode flow to a node 
as positive and flow from as negative.  This reduces your constraints to a much 
simpler form

Sum_j A_ij = 0
A_ij = A_ji

Sent from my iPhone

Reply via email to