BARON- Locally Optimal

Archiver · February 11, 2014, 5:54am

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 10:14am

Hi,

in case you have solve status NORMAL COMPLETION you might have to tighten the optimality tolerance optcr: ‘option optcr=0;’ GAMS’ default is 0.1 which lets BARON terminate before the globally optimal solution is found (or proven optimal). If you do not have NORMAL COMPLETION but ran against some limit, you need to increase that limit (e.g. time limit: option relim=2000;’ default is 1000 seconds).

Hope this helps,
Michael Bussieck - GAMSWorld Coordinator

On Monday, February 10, 2014 11:54:29 PM UTC-5, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 10:37am

Dear Mr. Bussieck,

Thank you for your response.

My model is

SCALARS r1 the manufacturer capacity utilization /0.9/
r2 the supplier capacity utilization /0.8/
q deteriorating rate which is a fraction of the on-hand inventory /0.1/
F supplier setup cost per setup /200 /
A manufacturer ordering setup cost per cycle /50/
c supplier production cost per unit /5/
s the unit price charged by the supplier to the manufacturer /10/
b backorder cost /2/
h manufacturer inventory holding cost rate excluding interest charges /0.1/
j supplier inventory holding cost rate excluding interest charges /0.1/
Is supplier opportunity cost per dollar per unit time /0.02/
Ip manufacturer opportunity cost per dollar per unit time /0.06/
Ie manufacturer interest earned per dollar per unit time /0.04/
k constant in the manufacturer demand function representing his market scale /5000000/
e price elasticity of manufacturer demand rate which is alpha in the thesis /1.5/
M the manufacturer trade credit period offered by the supplier /0.166666666666667/
N the customer trade credit period offered by the manufacturer /0.0821917808219178/
g number of shipments from the supplier to the manufacturer per production run /1/ ;

VARIABLES
z1 total net profit ;
POSITIVE VARIABLES
T1 production time with backorder at manufacturer’s side
T2 production time when positive stock builds up and the stock depletes due to the demand and deterioration at manufacturers side
v the unit price charged by the manufacturer to the customers ;
T1.lo = 0.01 ;
T2.lo = 0.01 ;
v.lo = 1 ;

Equations
Profit define objective function
Constraint condition for NP1
PriceConstraint constraint for price ;

Profit … z1 =e= (1/((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))((v-s)(k*(v**(-e)))((1/r1)T1+(T2+((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))))-A-(sh/q)(k(v**(-e)))(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))-(b)((1/2)(T1**2)(k*(v**(-e)))((1/r1)-1)(1/r1))-s*(k*(v**(-e)))((1/r1)T2-(T2+((1/q)(log((1-(1/r1))exp(-qT2)+ 1/r1)))))+(vIe)((k(v**(-e))/(2r1))((M-N)2))-(sIp(k(v*(-e)))/q)(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))) -vIp*(1/2)(k(v**(-e)))((N(2T2+2((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))-N))+N(T1+N-M)(1/r1))+(s-c)((k*(v**(-e))/r1)(T1+T2))-F/(g)- c(j+Is)(((k(v**(-e)))/r1)((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))(T1+T2)(r2(1-((g)/2))+(((g)-1)/2)))-sIs((k*(v**(-e))/r1)*(T1+T2)*M)) ;
Constraint … M =l= T1 ;
PriceConstraint … v =g= s ;

MODEL NetProfit /ALL/ ;
SOLVE NetProfit USING NLP MAXIMIZING z1 ;
option decimals=8;
Display z1.l;
Display T1.l, T2.l, v.l ;

execute_unload
parameter ms, ss;

This would provide you with a number of your model
(you can see here all the messages http://www.gams.com/mccarl/mccarlhtml/modelstat_tmodstat.htm)
The information of this is precious as you can see if your solution is optimal, locally optimal, ifeasible etc.
The second parameter provides you with a status regarding your solver, and how to fix some things
(http://www.gams.com/mccarl/mccarlhtml/solvestat_tsolstat.htm)

ms=NetProfit.modelstat;
ss=NetProfit.solvestat;

and the solver status is Normal Completion. I actually do not know where to put ‘option optcr=0;’. Should I add option optcr as a scalar to the model?

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 10:40am

As suggested in the guidelines of this group you should search GAMS documentation. Entering “optcr” in search on the GAMS web page (http://www.gams.com/search.htm) gives you: http://support.gams.com/doku.php?id=solver:what_is_optca_optcr and many more entries.

Best,
Michael Bussieck - GAMSWorld Coordinator

On Tuesday, February 11, 2014 4:37:15 AM UTC-5, Zohreh wrote:

Dear Mr. Bussieck,

Thank you for your response.

My model is

SCALARS r1 the manufacturer capacity utilization /0.9/
r2 the supplier capacity utilization /0.8/
q deteriorating rate which is a fraction of the on-hand inventory /0.1/
F supplier setup cost per setup /200 /
A manufacturer ordering setup cost per cycle /50/
c supplier production cost per unit /5/
s the unit price charged by the supplier to the manufacturer /10/
b backorder cost /2/
h manufacturer inventory holding cost rate excluding interest charges /0.1/
j supplier inventory holding cost rate excluding interest charges /0.1/
Is supplier opportunity cost per dollar per unit time /0.02/
Ip manufacturer opportunity cost per dollar per unit time /0.06/
Ie manufacturer interest earned per dollar per unit time /0.04/
k constant in the manufacturer demand function representing his market scale /5000000/
e price elasticity of manufacturer demand rate which is alpha in the thesis /1.5/
M the manufacturer trade credit period offered by the supplier /0.166666666666667/
N the customer trade credit period offered by the manufacturer /0.0821917808219178/
g number of shipments from the supplier to the manufacturer per production run /1/ ;

VARIABLES
z1 total net profit ;
POSITIVE VARIABLES
T1 production time with backorder at manufacturer’s side
T2 production time when positive stock builds up and the stock depletes due to the demand and deterioration at manufacturers side
v the unit price charged by the manufacturer to the customers ;
T1.lo = 0.01 ;
T2.lo = 0.01 ;
v.lo = 1 ;

Equations
Profit define objective function
Constraint condition for NP1
PriceConstraint constraint for price ;

Profit … z1 =e= (1/((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))((v-s)(k*(v**(-e)))((1/r1)T1+(T2+((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))))-A-(sh/q)(k(v**(-e)))(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))-(b)((1/2)(T1**2)(k*(v**(-e)))((1/r1)-1)(1/r1))-s*(k*(v**(-e)))((1/r1)T2-(T2+((1/q)(log((1-(1/r1))exp(-qT2)+ 1/r1)))))+(vIe)((k(v**(-e))/(2r1))((M-N)2))-(sIp(k(v*(-e)))/q)(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))) -vIp*(1/2)(k(v**(-e)))((N(2T2+2((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))-N))+N(T1+N-M)(1/r1))+(s-c)((k*(v**(-e))/r1)(T1+T2))-F/(g)- c(j+Is)(((k(v**(-e)))/r1)((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))(T1+T2)(r2(1-((g)/2))+(((g)-1)/2)))-sIs((k*(v**(-e))/r1)*(T1+T2)*M)) ;
Constraint … M =l= T1 ;
PriceConstraint … v =g= s ;

MODEL NetProfit /ALL/ ;
SOLVE NetProfit USING NLP MAXIMIZING z1 ;
option decimals=8;
Display z1.l;
Display T1.l, T2.l, v.l ;

execute_unload
parameter ms, ss;

This would provide you with a number of your model
(you can see here all the messages http://www.gams.com/mccarl/mccarlhtml/modelstat_tmodstat.htm)
The information of this is precious as you can see if your solution is optimal, locally optimal, ifeasible etc.
The second parameter provides you with a status regarding your solver, and how to fix some things
(http://www.gams.com/mccarl/mccarlhtml/solvestat_tsolstat.htm)

ms=NetProfit.modelstat;
ss=NetProfit.solvestat;

and the solver status is Normal Completion. I actually do not know where to put ‘option optcr=0;’. Should I add option optcr as a scalar to the model?

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 10:44am

Does it mean that putting optcr=0 with other solvers such as CONOPT, SNOPT etc. may also provide Global solution if the solver status is normal completion?

On Tue, Feb 11, 2014 at 3:10 PM, Michael Bussieck wrote:

As suggested in the guidelines of this group you should search GAMS documentation. Entering “optcr” in search on the GAMS web page (http://www.gams.com/search.htm) gives you: http://support.gams.com/doku.php?id=solver:what_is_optca_optcr and many more entries.

Best,

Michael Bussieck - GAMSWorld Coordinator

On Tuesday, February 11, 2014 4:37:15 AM UTC-5, Zohreh wrote:

Dear Mr. Bussieck,

Thank you for your response.

My model is

SCALARS r1 the manufacturer capacity utilization /0.9/
r2 the supplier capacity utilization /0.8/
q deteriorating rate which is a fraction of the on-hand inventory /0.1/
F supplier setup cost per setup /200 /
A manufacturer ordering setup cost per cycle /50/
c supplier production cost per unit /5/
s the unit price charged by the supplier to the manufacturer /10/
b backorder cost /2/
h manufacturer inventory holding cost rate excluding interest charges /0.1/
j supplier inventory holding cost rate excluding interest charges /0.1/
Is supplier opportunity cost per dollar per unit time /0.02/
Ip manufacturer opportunity cost per dollar per unit time /0.06/
Ie manufacturer interest earned per dollar per unit time /0.04/
k constant in the manufacturer demand function representing his market scale /5000000/
e price elasticity of manufacturer demand rate which is alpha in the thesis /1.5/
M the manufacturer trade credit period offered by the supplier /0.166666666666667/
N the customer trade credit period offered by the manufacturer /0.0821917808219178/
g number of shipments from the supplier to the manufacturer per production run /1/ ;

VARIABLES
z1 total net profit ;
POSITIVE VARIABLES
T1 production time with backorder at manufacturer’s side
T2 production time when positive stock builds up and the stock depletes due to the demand and deterioration at manufacturers side
v the unit price charged by the manufacturer to the customers ;
T1.lo = 0.01 ;
T2.lo = 0.01 ;
v.lo = 1 ;

Equations
Profit define objective function
Constraint condition for NP1
PriceConstraint constraint for price ;

Profit … z1 =e= (1/((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))((v-s)(k*(v**(-e)))((1/r1)T1+(T2+((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))))-A-(sh/q)(k(v**(-e)))(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))-(b)((1/2)(T1**2)(k*(v**(-e)))((1/r1)-1)(1/r1))-s*(k*(v**(-e)))((1/r1)T2-(T2+((1/q)(log((1-(1/r1))exp(-qT2)+ 1/r1)))))+(vIe)((k(v**(-e))/(2r1))((M-N)2))-(sIp(k(v*(-e)))/q)(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))) -vIp*(1/2)(k(v**(-e)))((N(2T2+2((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))-N))+N(T1+N-M)(1/r1))+(s-c)((k*(v**(-e))/r1)(T1+T2))-F/(g)- c(j+Is)(((k(v**(-e)))/r1)((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))(T1+T2)(r2(1-((g)/2))+(((g)-1)/2)))-sIs((k*(v**(-e))/r1)*(T1+T2)*M)) ;
Constraint … M =l= T1 ;
PriceConstraint … v =g= s ;

MODEL NetProfit /ALL/ ;
SOLVE NetProfit USING NLP MAXIMIZING z1 ;
option decimals=8;
Display z1.l;
Display T1.l, T2.l, v.l ;

execute_unload
parameter ms, ss;

This would provide you with a number of your model
(you can see here all the messages http://www.gams.com/mccarl/mccarlhtml/modelstat_tmodstat.htm)
The information of this is precious as you can see if your solution is optimal, locally optimal, ifeasible etc.
The second parameter provides you with a status regarding your solver, and how to fix some things
(http://www.gams.com/mccarl/mccarlhtml/solvestat_tsolstat.htm)

ms=NetProfit.modelstat;
ss=NetProfit.solvestat;

and the solver status is Normal Completion. I actually do not know where to put ‘option optcr=0;’. Should I add option optcr as a scalar to the model?

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

\

With Regards,

Parul Mathuria
ID- 2010REE505
Research Scholar
MNIT, Jaipur

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 10:48am

Thank you very much.

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 11:30am

Dear Mr. Bussieck,

I set optcr to zero. But I still get the status model as Locally Optimal and the answers are exactly as same as before. I would highly appreciate it if you guide me what the possible problem is.

SCALARS r1 the manufacturer capacity utilization /0.9/
r2 the supplier capacity utilization /0.8/
q deteriorating rate which is a fraction of the on-hand inventory /0.1/
F supplier setup cost per setup /200 /
A manufacturer ordering setup cost per cycle /50/
c supplier production cost per unit /5/
s the unit price charged by the supplier to the manufacturer /10/
b backorder cost /2/
h manufacturer inventory holding cost rate excluding interest charges /0.1/
j supplier inventory holding cost rate excluding interest charges /0.1/
Is supplier opportunity cost per dollar per unit time /0.02/
Ip manufacturer opportunity cost per dollar per unit time /0.06/
Ie manufacturer interest earned per dollar per unit time /0.04/
k constant in the manufacturer demand function representing his market scale /5000000/
e price elasticity of manufacturer demand rate which is alpha in the thesis /1.5/
M the manufacturer trade credit period offered by the supplier /0.166666666666667/
N the customer trade credit period offered by the manufacturer /0.0821917808219178/
g number of shipments from the supplier to the manufacturer per production run /1/ ;

VARIABLES
z1 total net profit ;
POSITIVE VARIABLES
T1 production time with backorder at manufacturer’s side
T2 production time when positive stock builds up and the stock depletes due to the demand and deterioration at manufacturers side
v the unit price charged by the manufacturer to the customers ;
T1.lo = 0.01 ;
T2.lo = 0.01 ;
v.lo = 1 ;

Equations
Profit define objective function
Constraint condition for NP1
PriceConstraint constraint for price ;

Profit … z1 =e= (1/((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))((v-s)(k*(v**(-e)))((1/r1)T1+(T2+((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))))-A-(sh/q)(k(v**(-e)))(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))-(b)((1/2)(T1**2)(k*(v**(-e)))((1/r1)-1)(1/r1))-s*(k*(v**(-e)))((1/r1)T2-(T2+((1/q)(log((1-(1/r1))exp(-qT2)+ 1/r1)))))+(vIe)((k(v**(-e))/(2r1))((M-N)2))-(sIp(k(v*(-e)))/q)(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))) -vIp*(1/2)(k(v**(-e)))((N(2T2+2((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))-N))+N(T1+N-M)(1/r1))+(s-c)((k*(v**(-e))/r1)(T1+T2))-F/(g)- c(j+Is)(((k(v**(-e)))/r1)((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))(T1+T2)(r2(1-((g)/2))+(((g)-1)/2)))-sIs((k*(v**(-e))/r1)*(T1+T2)*M)) ;
Constraint … M =l= T1 ;
PriceConstraint … v =g= s ;

MODEL NetProfit /ALL/ ;

scalars optcr relative gap following the GAMS formula ;
optcr=0;
option decimals=8; display optcr;

SOLVE NetProfit USING NLP MAXIMIZING z1 ;
option decimals=8;
Display z1.l;
Display T1.l, T2.l, v.l;

execute_unload
parameter ms, ss;

This would provide you with a number of your model
(you can see here all the messages http://www.gams.com/mccarl/mccarlhtml/modelstat_tmodstat.htm)
The information of this is precious as you can see if your solution is optimal, locally optimal, ifeasible etc.
The second parameter provides you with a status regarding your solver, and how to fix some things
(http://www.gams.com/mccarl/mccarlhtml/solvestat_tsolstat.htm)

ms=NetProfit.modelstat;
ss=NetProfit.solvestat;

I would be looking forward to hearing your kind response.

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 11, 2014, 8:08pm

Hi,

First of all, you did not set optcr correctly. OptCr is an option not a user scalar. Here is the correct use:

…
MODEL NetProfit /ALL/ ;
option optcr=0, decimals=8;
SOLVE NetProfit USING NLP MAXIMIZING z1 ;
…

When I run the model (with correct optcr), I get clear indicators from GAMS/BARON why BARON can’t prove global optimality:

User did not provide appropriate variable bounds.
Some model expressions are unbounded.
We may not be able to guarantee globality.

and at the end of the run:

*** User did not provide appropriate variable bounds ***
*** Globality is therefore not guaranteed ***

For BARON to prove global optimality everything needs to be nicely bounded. If the variables is not bounded (and BARON can deduce bounds) this is not good. Try to set finite upper bounds (lower you set already).

After setting some upper bounds (T1.up = 1.01; T2.up = 1.01; v.up = 20; not sure if they really make sense), BARON solves to global optimality.

Hope this helps,
Michael Bussieck - GAMSWorld Coordinator

On Tuesday, February 11, 2014 5:30:02 AM UTC-5, Zohreh wrote:

Dear Mr. Bussieck,

I set optcr to zero. But I still get the status model as Locally Optimal and the answers are exactly as same as before. I would highly appreciate it if you guide me what the possible problem is.

SCALARS r1 the manufacturer capacity utilization /0.9/
r2 the supplier capacity utilization /0.8/
q deteriorating rate which is a fraction of the on-hand inventory /0.1/
F supplier setup cost per setup /200 /
A manufacturer ordering setup cost per cycle /50/
c supplier production cost per unit /5/
s the unit price charged by the supplier to the manufacturer /10/
b backorder cost /2/
h manufacturer inventory holding cost rate excluding interest charges /0.1/
j supplier inventory holding cost rate excluding interest charges /0.1/
Is supplier opportunity cost per dollar per unit time /0.02/
Ip manufacturer opportunity cost per dollar per unit time /0.06/
Ie manufacturer interest earned per dollar per unit time /0.04/
k constant in the manufacturer demand function representing his market scale /5000000/
e price elasticity of manufacturer demand rate which is alpha in the thesis /1.5/
M the manufacturer trade credit period offered by the supplier /0.166666666666667/
N the customer trade credit period offered by the manufacturer /0.0821917808219178/
g number of shipments from the supplier to the manufacturer per production run /1/ ;

VARIABLES
z1 total net profit ;
POSITIVE VARIABLES
T1 production time with backorder at manufacturer’s side
T2 production time when positive stock builds up and the stock depletes due to the demand and deterioration at manufacturers side
v the unit price charged by the manufacturer to the customers ;
T1.lo = 0.01 ;
T2.lo = 0.01 ;
v.lo = 1 ;

Equations
Profit define objective function
Constraint condition for NP1
PriceConstraint constraint for price ;

Profit … z1 =e= (1/((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))((v-s)(k*(v**(-e)))((1/r1)T1+(T2+((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))))-A-(sh/q)(k(v**(-e)))(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1)))))-(b)((1/2)(T1**2)(k*(v**(-e)))((1/r1)-1)(1/r1))-s*(k*(v**(-e)))((1/r1)T2-(T2+((1/q)(log((1-(1/r1))exp(-qT2)+ 1/r1)))))+(vIe)((k(v**(-e))/(2r1))((M-N)2))-(sIp(k(v*(-e)))/q)(((1/r1)-1)T2-((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))) -vIp*(1/2)(k(v**(-e)))((N(2T2+2((1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))-N))+N(T1+N-M)(1/r1))+(s-c)((k*(v**(-e))/r1)(T1+T2))-F/(g)- c(j+Is)(((k(v**(-e)))/r1)((1/r1)T1+T2+(1/q)(log((1-(1/r1))exp(-qT2)+(1/r1))))(T1+T2)(r2(1-((g)/2))+(((g)-1)/2)))-sIs((k*(v**(-e))/r1)*(T1+T2)*M)) ;
Constraint … M =l= T1 ;
PriceConstraint … v =g= s ;

MODEL NetProfit /ALL/ ;

scalars optcr relative gap following the GAMS formula ;
optcr=0;
option decimals=8; display optcr;

SOLVE NetProfit USING NLP MAXIMIZING z1 ;
option decimals=8;
Display z1.l;
Display T1.l, T2.l, v.l;

execute_unload
parameter ms, ss;

This would provide you with a number of your model
(you can see here all the messages http://www.gams.com/mccarl/mccarlhtml/modelstat_tmodstat.htm)
The information of this is precious as you can see if your solution is optimal, locally optimal, ifeasible etc.
The second parameter provides you with a status regarding your solver, and how to fix some things
(http://www.gams.com/mccarl/mccarlhtml/solvestat_tsolstat.htm)

ms=NetProfit.modelstat;
ss=NetProfit.solvestat;

I would be looking forward to hearing your kind response.

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.

Archiver · February 12, 2014, 4:11am

Dear Mr. Bussieck,

Thank you very much for your clear answer. It was really helpful.

Best regards.

On Tuesday, February 11, 2014 12:54:29 PM UTC+8, Zohreh wrote:

Dear Friends,

As far as I have found, BARON is a global optimizer. But, after solving my model, its status is locally optimal. Does it mean that the model in not concave (for maximum) and the global optimum does not exist?

I would highly appreciate it if anybody can help me understanding this issue.

I would be looking forward to hearing your kind response.

Best regards.

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.