0

![The screenshot is from the error result of command:

runph --model-directory=models --instance-directory=scenariodata --solver=ipopt --default-rho=1

]1

Although I could still get outputs, I wonder what does this error warning

"** At least one sub-problem failed to solve! **"

mean? and what should I do next? Thanks for your kind help.

Xin Huang
  • 1
  • 1
  • Question is unclear add some more information – Satendra Nov 17 '17 at 07:53
  • I think it's likely that one of your scenarios might have been infeasible or did not converge within the time limit. – Qi Chen Nov 17 '17 at 17:37
  • @QiChen Thanks man, so it means I have to enlarge my time limit? Due to orginial theory about PHA, it will be convergent if the objective function is convex w.r.t. policy (controls), so is there any check procedure to this issue if my problem is a large and nonlinear case? Thanks again – Xin Huang Nov 18 '17 at 08:18
  • Also thanks to @Satendra I will add more info. soon for you to do me a kind favor. – Xin Huang Nov 18 '17 at 08:19
  • I'm not familiar enough with the progressive hedging algorithm to be able to say anything for sure. But in general the subproblems of stochastic programs tend to be scenario evaluations. – Qi Chen Nov 18 '17 at 18:12

1 Answers1

0

In this case, there is a problem with the constraint, or the value of the objective function goes to infinity.

I recommend to add '--verbose' in your command. This command can help you why this error is occurred. 

runph --model-directory=models --instance-directory=scenariodata --solver=ipopt --default-rho=1 --verbose

This command provides more information about your problem.

bright612
  • 35
  • 7