[AMPL 15261] Helps BARON to find the feasible solution

classic Classic list List threaded Threaded
4 messages Options
SG
Reply | Threaded
Open this post in threaded view
|

[AMPL 15261] Helps BARON to find the feasible solution

SG
Hi,

I defined the objetive function minimize 

z: X_Am_c + X_Am_g + X_Am_k;

while

# area
param Am_min := 0.5; # min  area [m2]
param Am_max := 10; # max  area [m2]
var X_Am_c >= Am_min, <= Am_max; #  area
var X_Am_g >= Am_min, <= Am_max; #  area
var X_Am_k >= Am_min, <= Am_max; #  area

When I run filter solver I get the optimal value of 

filterSQP (20020316): Optimal solution found, objective = 10.94671379
37 iterations (19 for feasibility)
Evals: obj = 26, constr = 44, grad = 27, Hes = 32
z = 10.9467
X_Am_c = 1.08342
X_Am_g = 5.00282
X_Am_k = 4.86047

Using BARON just gave me the following solution! 

===========================================================================
  Iteration    Open nodes         Time (s)    Lower bound      Upper bound
          1             1             0.12     1.50000         0.100000E+52
        458           291            39.20     1.50000         0.100000E+52
       1561           892            69.24     1.50000         0.100000E+52
       2265          1313           115.63     1.50000         0.100000E+52
       3026          1762           161.79     1.50000         0.100000E+52
       4081          2350           191.82     1.50000         0.100000E+52
       4787          2776           235.34     1.50000         0.100000E+52
       5009          2917           265.37     1.50000         0.100000E+52
       5590          3203           305.81     1.50000         0.100000E+52
       6083          3452           346.52     1.50000         0.100000E+52
       6274          3559           376.57     1.50000         0.100000E+52
       7281          4099           406.63     1.50000         0.100000E+52
       8127          4571           436.66     1.50000         0.100000E+52
       8430          4752           471.06     1.50000         0.100000E+52
       8655          4885           500.39     1.50000          30.0000    

                    *** Max. allowable time exceeded ***      

                   *** No feasible solution was found ***

 Wall clock time:                   500.00
 Total CPU time used:               500.39

 Total no. of BaR iterations:    8655
 Best solution found at node:      -3
 Max. no. of nodes in memory:    4885
 
 All done
===========================================================================
BARON 16.7.29 (2016.07.29): 8655 iterations, CPU time limit reached.
Objective 1e+51
z = 0
X_Am_c = 0
X_Am_g = 0
X_Am_k = 0

I used even the baron options as: # option baron_options 'outlev=1 maxtime=-1 maxiter=-1';

but nothing. 

And all the variables equal to ZERO. That's unbelievable as I limited bounds of variables directly affected the obj function but nothing happened!!

I am wondering if I can improve the rate of convergence? Or somehow trigger the solver to see the global optimum 

Thank you. 

--
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 [hidden email].
To post to this group, send email to [hidden email].
Visit this group at https://groups.google.com/group/ampl.
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

RE: [AMPL 15264] Helps BARON to find the feasible solution

Robert Fourer-2
According to the message from BARON, "no feasible solution was found" -- so no solution was returned to AMPL, and all the variables are still at zero (which is the default value for variables).  This appears to be a hard problem for BARON to find a feasible solution.

To see if a good starting solution would help BARON, you could solve with filterSQP and then set "option solver baron" and solve again, which would cause the local optimum from filterSQP to be used as a start for BARON.  Or you could just use AMPL "let" commands to assign some reasonable starting values to the variables instead of the default values of zero.

Finally, Couenne is another global solver that you could try.

Bob Fourer
[hidden email]

=======

From: [hidden email] [mailto:[hidden email]] On Behalf Of SG
Sent: Thursday, November 30, 2017 10:27 PM
To: AMPL Modeling Language
Subject: [AMPL 15261] Helps BARON to find the feasible solution

I defined the objetive function minimize

z: X_Am_c + X_Am_g + X_Am_k;

while

# area
param Am_min := 0.5; # min  area [m2]
param Am_max := 10; # max  area [m2]
var X_Am_c >= Am_min, <= Am_max; #  area
var X_Am_g >= Am_min, <= Am_max; #  area
var X_Am_k >= Am_min, <= Am_max; #  area

When I run filter solver I get the optimal value of

filterSQP (20020316): Optimal solution found, objective = 10.94671379
37 iterations (19 for feasibility)
Evals: obj = 26, constr = 44, grad = 27, Hes = 32
z = 10.9467
X_Am_c = 1.08342
X_Am_g = 5.00282
X_Am_k = 4.86047

Using BARON just gave me the following solution!

===========================================================================
  Iteration    Open nodes         Time (s)    Lower bound      Upper bound
          1             1             0.12     1.50000         0.100000E+52
        458           291            39.20     1.50000         0.100000E+52
       1561           892            69.24     1.50000         0.100000E+52
       2265          1313           115.63     1.50000         0.100000E+52
       3026          1762           161.79     1.50000         0.100000E+52
       4081          2350           191.82     1.50000         0.100000E+52
       4787          2776           235.34     1.50000         0.100000E+52
       5009          2917           265.37     1.50000         0.100000E+52
       5590          3203           305.81     1.50000         0.100000E+52
       6083          3452           346.52     1.50000         0.100000E+52
       6274          3559           376.57     1.50000         0.100000E+52
       7281          4099           406.63     1.50000         0.100000E+52
       8127          4571           436.66     1.50000         0.100000E+52
       8430          4752           471.06     1.50000         0.100000E+52
       8655          4885           500.39     1.50000          30.0000    

                    *** Max. allowable time exceeded ***      

                   *** No feasible solution was found ***

 Wall clock time:                   500.00
 Total CPU time used:               500.39

 Total no. of BaR iterations:    8655
 Best solution found at node:      -3
 Max. no. of nodes in memory:    4885
 
 All done
===========================================================================
BARON 16.7.29 (2016.07.29): 8655 iterations, CPU time limit reached.
Objective 1e+51
z = 0
X_Am_c = 0
X_Am_g = 0
X_Am_k = 0

I used even the baron options as: # option baron_options 'outlev=1 maxtime=-1 maxiter=-1';

but nothing.

And all the variables equal to ZERO. That's unbelievable as I limited bounds of variables directly affected the obj function but nothing happened!!

I am wondering if I can improve the rate of convergence? Or somehow trigger the solver to see the global optimum


--
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 [hidden email].
To post to this group, send email to [hidden email].
Visit this group at https://groups.google.com/group/ampl.
For more options, visit https://groups.google.com/d/optout.
SG
Reply | Threaded
Open this post in threaded view
|

Re: [AMPL 15271] Helps BARON to find the feasible solution

SG
Dear Robert,

Many thanks. 

I used the neos server to run the model which does not accept baron options NLPSol = 14 for filterSQP. I have to test GAMS to see it works or not.

Regards

On Friday, December 1, 2017 at 1:13:00 PM UTC-5, Robert Fourer wrote:
According to the message from BARON, "no feasible solution was found" -- so no solution was returned to AMPL, and all the variables are still at zero (which is the default value for variables).  This appears to be a hard problem for BARON to find a feasible solution.

To see if a good starting solution would help BARON, you could solve with filterSQP and then set "option solver baron" and solve again, which would cause the local optimum from filterSQP to be used as a start for BARON.  Or you could just use AMPL "let" commands to assign some reasonable starting values to the variables instead of the default values of zero.

Finally, Couenne is another global solver that you could try.

Bob Fourer
<a href="javascript:" target="_blank" gdf-obfuscated-mailto="iGDSbm9eAgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">am...@...

=======

From: <a href="javascript:" target="_blank" gdf-obfuscated-mailto="iGDSbm9eAgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">am...@... [mailto:<a href="javascript:" target="_blank" gdf-obfuscated-mailto="iGDSbm9eAgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">am...@...] On Behalf Of SG
Sent: Thursday, November 30, 2017 10:27 PM
To: AMPL Modeling Language
Subject: [AMPL 15261] Helps BARON to find the feasible solution

I defined the objetive function minimize

z: X_Am_c + X_Am_g + X_Am_k;

while

# area
param Am_min := 0.5; # min  area [m2]
param Am_max := 10; # max  area [m2]
var X_Am_c >= Am_min, <= Am_max; #  area
var X_Am_g >= Am_min, <= Am_max; #  area
var X_Am_k >= Am_min, <= Am_max; #  area

When I run filter solver I get the optimal value of

filterSQP (20020316): Optimal solution found, objective = 10.94671379
37 iterations (19 for feasibility)
Evals: obj = 26, constr = 44, grad = 27, Hes = 32
z = 10.9467
X_Am_c = 1.08342
X_Am_g = 5.00282
X_Am_k = 4.86047

Using BARON just gave me the following solution!

===========================================================================
  Iteration    Open nodes         Time (s)    Lower bound      Upper bound
          1             1             0.12     1.50000         0.100000E+52
        458           291            39.20     1.50000         0.100000E+52
       1561           892            69.24     1.50000         0.100000E+52
       2265          1313           115.63     1.50000         0.100000E+52
       3026          1762           161.79     1.50000         0.100000E+52
       4081          2350           191.82     1.50000         0.100000E+52
       4787          2776           235.34     1.50000         0.100000E+52
       5009          2917           265.37     1.50000         0.100000E+52
       5590          3203           305.81     1.50000         0.100000E+52
       6083          3452           346.52     1.50000         0.100000E+52
       6274          3559           376.57     1.50000         0.100000E+52
       7281          4099           406.63     1.50000         0.100000E+52
       8127          4571           436.66     1.50000         0.100000E+52
       8430          4752           471.06     1.50000         0.100000E+52
       8655          4885           500.39     1.50000          30.0000    

                    *** Max. allowable time exceeded ***      

                   *** No feasible solution was found ***

 Wall clock time:                   500.00
 Total CPU time used:               500.39

 Total no. of BaR iterations:    8655
 Best solution found at node:      -3
 Max. no. of nodes in memory:    4885
 
 All done
===========================================================================
BARON 16.7.29 (2016.07.29): 8655 iterations, CPU time limit reached.
Objective 1e+51
z = 0
X_Am_c = 0
X_Am_g = 0
X_Am_k = 0

I used even the baron options as: # option baron_options 'outlev=1 maxtime=-1 maxiter=-1';

but nothing.

And all the variables equal to ZERO. That's unbelievable as I limited bounds of variables directly affected the obj function but nothing happened!!

I am wondering if I can improve the rate of convergence? Or somehow trigger the solver to see the global optimum


--
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 [hidden email].
To post to this group, send email to [hidden email].
Visit this group at https://groups.google.com/group/ampl.
For more options, visit https://groups.google.com/d/optout.
SG
Reply | Threaded
Open this post in threaded view
|

Re: [AMPL 15272] Helps BARON to find the feasible solution

SG
In reply to this post by Robert Fourer-2
Dear Robert,

Neos-Server is limited, got the message as

 This BARON run may utilize the following subsolver(s)
 For LP/MIP: CLP/CBC                                         
 For NLP: IPOPT, FILTERSD

I cannot change it!

On Friday, December 1, 2017 at 1:13:00 PM UTC-5, Robert Fourer wrote:
According to the message from BARON, "no feasible solution was found" -- so no solution was returned to AMPL, and all the variables are still at zero (which is the default value for variables).  This appears to be a hard problem for BARON to find a feasible solution.

To see if a good starting solution would help BARON, you could solve with filterSQP and then set "option solver baron" and solve again, which would cause the local optimum from filterSQP to be used as a start for BARON.  Or you could just use AMPL "let" commands to assign some reasonable starting values to the variables instead of the default values of zero.

Finally, Couenne is another global solver that you could try.

Bob Fourer
<a href="javascript:" target="_blank" gdf-obfuscated-mailto="iGDSbm9eAgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">am...@...

=======

From: <a href="javascript:" target="_blank" gdf-obfuscated-mailto="iGDSbm9eAgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">am...@... [mailto:<a href="javascript:" target="_blank" gdf-obfuscated-mailto="iGDSbm9eAgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">am...@...] On Behalf Of SG
Sent: Thursday, November 30, 2017 10:27 PM
To: AMPL Modeling Language
Subject: [AMPL 15261] Helps BARON to find the feasible solution

I defined the objetive function minimize

z: X_Am_c + X_Am_g + X_Am_k;

while

# area
param Am_min := 0.5; # min  area [m2]
param Am_max := 10; # max  area [m2]
var X_Am_c >= Am_min, <= Am_max; #  area
var X_Am_g >= Am_min, <= Am_max; #  area
var X_Am_k >= Am_min, <= Am_max; #  area

When I run filter solver I get the optimal value of

filterSQP (20020316): Optimal solution found, objective = 10.94671379
37 iterations (19 for feasibility)
Evals: obj = 26, constr = 44, grad = 27, Hes = 32
z = 10.9467
X_Am_c = 1.08342
X_Am_g = 5.00282
X_Am_k = 4.86047

Using BARON just gave me the following solution!

===========================================================================
  Iteration    Open nodes         Time (s)    Lower bound      Upper bound
          1             1             0.12     1.50000         0.100000E+52
        458           291            39.20     1.50000         0.100000E+52
       1561           892            69.24     1.50000         0.100000E+52
       2265          1313           115.63     1.50000         0.100000E+52
       3026          1762           161.79     1.50000         0.100000E+52
       4081          2350           191.82     1.50000         0.100000E+52
       4787          2776           235.34     1.50000         0.100000E+52
       5009          2917           265.37     1.50000         0.100000E+52
       5590          3203           305.81     1.50000         0.100000E+52
       6083          3452           346.52     1.50000         0.100000E+52
       6274          3559           376.57     1.50000         0.100000E+52
       7281          4099           406.63     1.50000         0.100000E+52
       8127          4571           436.66     1.50000         0.100000E+52
       8430          4752           471.06     1.50000         0.100000E+52
       8655          4885           500.39     1.50000          30.0000    

                    *** Max. allowable time exceeded ***      

                   *** No feasible solution was found ***

 Wall clock time:                   500.00
 Total CPU time used:               500.39

 Total no. of BaR iterations:    8655
 Best solution found at node:      -3
 Max. no. of nodes in memory:    4885
 
 All done
===========================================================================
BARON 16.7.29 (2016.07.29): 8655 iterations, CPU time limit reached.
Objective 1e+51
z = 0
X_Am_c = 0
X_Am_g = 0
X_Am_k = 0

I used even the baron options as: # option baron_options 'outlev=1 maxtime=-1 maxiter=-1';

but nothing.

And all the variables equal to ZERO. That's unbelievable as I limited bounds of variables directly affected the obj function but nothing happened!!

I am wondering if I can improve the rate of convergence? Or somehow trigger the solver to see the global optimum


--
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 [hidden email].
To post to this group, send email to [hidden email].
Visit this group at https://groups.google.com/group/ampl.
For more options, visit https://groups.google.com/d/optout.