MINOS 5.51: unbounded (or badly scaled) problem

[Bekir Kırkıcı]

Hello everyone, I try to solve a model given below but i get the error message "MINOS 5.51: unbounded (or badly scaled) problem". I looked closely to my model but couldnt find the unboundedness problem. 

MODEL IS:

set DEPT;

set SCEN;

set PROD;

var Produce{PROD} >= 0 ;

var OT_Produced{SCEN, PROD} >= 0 ;

var Promise{PROD} >= 0 ;

var OT_hours_bought{DEPT, SCEN} >= 0 ;

param price{PROD} ;

param OT_sell_price{PROD} ; 

param prob{SCEN} ;

param cost_OT{DEPT} ; 

param hours_spent{DEPT, PROD} ;

param initial_time {DEPT} ;

param time_decrease {DEPT, SCEN} ;

param Quota {DEPT} ;

maximize Profit: ( ( sum {k in PROD} Produce[k] * price[k] ) + 

( sum {k in PROD, j in SCEN} OT_Produced[j, k] * OT_sell_price[k] * prob[j] ) ) -

( sum {j in SCEN, i in DEPT} prob[j] * cost_OT[i] * OT_hours_bought[i, j])            ;

subject to mise {j in SCEN, k in PROD}: Produce[k] * OT_Produced[j, k] <= Promise[k] ;

subject to Availability {i in DEPT, j in SCEN, k in PROD}: Promise[k] * hours_spent[i, k] <= initial_time[i] - time_decrease[i, j] + OT_hours_bought[i, j] ;

subject to Quota_on_time {i in DEPT, j in SCEN}: OT_hours_bought[i, j] <= Quota[i];

subject to Non_neg {k in PROD}: Promise[k] >= 0 ;

WITH DATA:

data ;

set DEPT := cut sew fin ins;

set SCEN := A B none;

set PROD := std lux;

param: price OT_sell_price :=

std 10 8

lux 9 0;

param prob := 

A 0.5

B 0.4

none 0.1 ;

param: initial_time Quota cost_OT :=

cut 630 999999 5

sew 600 999999 6

fin 708 100 8

ins 135 999999 4 ;

param hours_spent :

std lux :=

cut 0.7 1

sew 0.5 0.833

fin 1 0.667

ins 0.1 0.25 ;

param time_decrease :

A B none :=

cut 50 30 0

sew 40 50 0

fin 80 70 0

ins 10 15 0 ;

[Robert Fourer]

After solving with MINOS, display the variables with

display Produce, Promise;
display OT_Produced;
display OT_hours_bought;

Look for variables that have very large values. Those are the variables that are going to infinity in the unbounded solution.

Bob Fourer
4...@ampl.com

=======

[Michael Saunders]

Bekir, if your model has nonlinear constraints and the

starting point is not very good, Minos may have trouble

solving the problem.  The approximate solution may diverge

(the opposite of convergence) and the final message may be

misleading.

Normally this won't happen if you have linear constraints.

Let's know which case you have.

If they're nonlinear, take care with the starting point.

Always take care with scaling.

Michael

--
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ampl+unsubscribe@googlegroups.com.
To post to this group, send email to am...@googlegroups.com.
Visit this group at https://groups.google.com/group/ampl.
For more options, visit https://groups.google.com/d/optout.