I’m running a nonlinear model which is consistently exiting on

solvestat = 2 (solver ran out of iterations). But iterlim is just its

default (10000) and the limit in KNITRO is also set to 10000, and my

model only iterates 29 times.

The KNITRO exit report states “EXIT: Locally optimal solution found.”

According to KNITRO’s manual, this means that the solution satisfies

the tolerance and feasibility stopping criteria. There is a separate

“EXIT: Iteration limit reached” message from KNITRO that would

indicate that an iteration limit was reached, and I’m not getting

that.

These messages from GAMS and KNITRO seem to conflict. Any insight

into this? Is the solvestat value I’m getting just incorrect?

Thanks,

Emily

\

Emily,

Could it be that modelstat is 2 (locally optimal) and solvestat is 1

(normal completion)? That would fit the message you get from knitro.

Could you upload listing and log file so we can dig a little deeper.

Michael Bussieck - GAMSWorld Coordinator

On Sep 9, 6:36 pm, emily wrote:

I’m running a nonlinear model which is consistently exiting on

solvestat = 2 (solver ran out of iterations). But iterlim is just its

default (10000) and the limit in KNITRO is also set to 10000, and my

model only iterates 29 times.

The KNITRO exit report states “EXIT: Locally optimal solution found.”

According to KNITRO’s manual, this means that the solution satisfies

the tolerance and feasibility stopping criteria. There is a separate

“EXIT: Iteration limit reached” message from KNITRO that would

indicate that an iteration limit was reached, and I’m not getting

that.

These messages from GAMS and KNITRO seem to conflict. Any insight

into this? Is the solvestat value I’m getting just incorrect?

Thanks,

Emily

\

How embarrassing! Yes, I’m sorry. I don’t even know how modelstat

got in there, because I’ve never (knowingly) used it before. Only

solvestat. I must have typed it mindlessly while looking at the help

file descriptions.

Sorry again. Mystery solved.

Emily

On Sep 10, 6:12 am, Gamsworld Admin wrote:

Emily,

Could it be that modelstat is 2 (locally optimal) and solvestat is 1

(normal completion)? That would fit the message you get from knitro.

Could you upload listing and log file so we can dig a little deeper.

Michael Bussieck - GAMSWorld Coordinator

On Sep 9, 6:36 pm, emily wrote:

I’m running a nonlinear model which is consistently exiting on

solvestat = 2 (solver ran out of iterations). But iterlim is just its

default (10000) and the limit in KNITRO is also set to 10000, and my

model only iterates 29 times.

The KNITRO exit report states “EXIT: Locally optimal solution found.”

According to KNITRO’s manual, this means that the solution satisfies

the tolerance and feasibility stopping criteria. There is a separate

“EXIT: Iteration limit reached” message from KNITRO that would

indicate that an iteration limit was reached, and I’m not getting

that.

These messages from GAMS and KNITRO seem to conflict. Any insight

into this? Is the solvestat value I’m getting just incorrect?

Thanks,

Emily- Hide quoted text -