Question on the result from two problems [2 Attachments]

[Cecile Damiron]

Hello lpsolve community,

I have an lp problem which gives me a set of results I'm confused about.

The first problem "lp_1_0_PB52_2.txt" gives me the set of results I expect, but the second problem "lp_2_0_PB52_2.txt" which changes only by one constraint returns all the decision variables equal to 0 and it is what I don't understand.

Problem 1:
Partial results are:
C122 = 333199.999999471
C125 = 208529.127889622

Problem 2:
Add the constraint: +0.741043 C122 -C125 <= 0;;
Partial results are:
C122 = 0
C125 = 0
I would have expect something like:
C122 = 281399.48 (=208529.127889622/0.741043)
C125 = 208529.127889622

In the problem 1: C122 and C125 are not related and in the problem 2 C122 is limited by C125. Maybe I'm missing something but I cannot figure out why I get a set of 0 in the problem2.

Any help is welcome.

Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15 because it causes some crashes in the application I'm using. But if I check the results with Lp solve IDE 5.5.0.15 I obtain identical sets of results.

Thanks for your help,

Cecile


     

[William H. Patton]

You seem to be right on the cusp of infeasible  +0.741043 C122 -C125 <= 0;

 

I can solve this +0.741041673 C122 -C125 <= 0; but nothing much larger.

+0.741041683 C122 -C125 <= 0;

Found feasibility by dual simplex after            18 iter.

The model is INFEASIBLE

lp_solve unsuccessful after 80 iter and a last best value of -1e+030

So I guess you forgot to check the return code before trying to fetch the
solution to the infeasible problem.

  _____  

From: lp_solve@... [mailto:lp_solve@...] On Behalf
Of Cecile Damiron
Sent: Thursday, November 05, 2009 7:31 PM
To: lp_solve@...
Subject: [lp_solve] Question on the result from two problems [2 Attachments]

 

 

[Attachment(s) from Cecile Damiron included below]

Hello lpsolve community,

I have an lp problem which gives me a set of results I'm confused about.

The first problem "lp_1_0_PB52_2.txt" gives me the set of results I expect,
but the second problem "lp_2_0_PB52_2.txt" which changes only by one
constraint returns all the decision variables equal to 0 and it is what I
don't understand.

Problem 1:
Partial results are:
C122 = 333199.999999471
C125 = 208529.127889622

Problem 2:
Add the constraint: +0.741043 C122 -C125 <= 0;;
Partial results are:
C122 = 0
C125 = 0
I would have expect something like:
C122 = 281399.48 (=208529.127889622/0.741043)
C125 = 208529.127889622

In the problem 1: C122 and C125 are not related and in the problem 2 C122 is
limited by C125. Maybe I'm missing something but I cannot figure out why I
get a set of 0 in the problem2.

Any help is welcome.

Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15
because it causes some crashes in the application I'm using. But if I check
the results with Lp solve IDE 5.5.0.15 I obtain identical sets of results.

Thanks for your help,

Cecile





 

[Cecile Damiron]

Thank you for your very prompt answer.

I checked the function "solve" and it returns 0 (Feasable) for the problem I sent. Maybe it is because I'm using 5.5.0.11?

With Lp solve IDE 5.5.0.15 I obtain the following answer:
log:
Model name:  'LPSolver' - run #1
Objective:   Maximize(R0)

SUBMITTED
Model size:       15 constraints,     234 variables,          260 non-zeros.
Sets:                                   0 GUB,                  0 SOS.

[           +0 < Z < Inf           ]

Using DUAL simplex for phase 1 and PRIMAL simplex for phase 2.
The primal and dual simplex pricing strategy set to 'Devex'.

Found feasibility by dual simplex after            18 iter.

Optimal solution                   0 after         28 iter.
Excellent numeric accuracy ||*|| = 0

MEMO: lp_solve version 5.5.0.15 for 32 bit OS, with 64 bit REAL variables.
In the total iteration count 28, 11 (39.3%) were bound flips.
There were 2 refactorizations, 0 triggered by time and 0 by density.
... on average 8.5 major pivots per refactorization.
The largest [LUSOL v2.2.1.0] fact(B) had 24 NZ entries, 1.0x largest basis.
The constraint matrix inf-norm is 2566.67, with a dynamic range of 6.58779e+006.
Time to load data was 0.020 seconds, presolve used 0.021 seconds,
... 0.008 seconds in simplex solver, in total 0.049 seconds.

msg:
9:51:38 AM: Presolve
9:51:38 AM: LP Feasible

If I "save IDE options...." I get:
TPF0 TLPSolverLPSolverLagTrace
BoundsTighterSimplexTypestDualPrimal
BasisCrashcNothingMessages mPresolve
mIterationmInvertmLpFeasible mLpBetter
mMiLpFeasible
mMiLpEqualmMiLpBettermMiLpOptimal

Sincerely,
Cecile


--- In lp_solve@..., "William H. Patton" <pattonwh@...> wrote:

>
> You seem to be right on the cusp of infeasible  +0.741043 C122 -C125 <= 0;
>
>  
>
> I can solve this +0.741041673 C122 -C125 <= 0; but nothing much larger.
>
> +0.741041683 C122 -C125 <= 0;
>
> Found feasibility by dual simplex after            18 iter.
>
> The model is INFEASIBLE
>
> lp_solve unsuccessful after 80 iter and a last best value of -1e+030
>
> So I guess you forgot to check the return code before trying to fetch the
> solution to the infeasible problem.
>
>   _____  
>
> From: lp_solve@... [mailto:lp_solve@...] On Behalf
> Of Cecile Damiron
> Sent: Thursday, November 05, 2009 7:31 PM
> To: lp_solve@...
> Subject: [lp_solve] Question on the result from two problems [2 Attachments]
>
>  
>
>  
>
> [Attachment(s) from Cecile Damiron included below]
>
> Hello lpsolve community,
>
> I have an lp problem which gives me a set of results I'm confused about.
>
> The first problem "lp_1_0_PB52_2.txt" gives me the set of results I expect,
> but the second problem "lp_2_0_PB52_2.txt" which changes only by one
> constraint returns all the decision variables equal to 0 and it is what I
> don't understand.
>
> Problem 1:
> Partial results are:
> C122 = 333199.999999471
> C125 = 208529.127889622
>
> Problem 2:
> Add the constraint: +0.741043 C122 -C125 <= 0;;
> Partial results are:
> C122 = 0
> C125 = 0
> I would have expect something like:
> C122 = 281399.48 (=208529.127889622/0.741043)
> C125 = 208529.127889622
>
> In the problem 1: C122 and C125 are not related and in the problem 2 C122 is
> limited by C125. Maybe I'm missing something but I cannot figure out why I
> get a set of 0 in the problem2.
>
> Any help is welcome.
>
> Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15
> because it causes some crashes in the application I'm using. But if I check
> the results with Lp solve IDE 5.5.0.15 I obtain identical sets of results.
>
> Thanks for your help,
>
> Cecile
>

[peter_notebaert]

I checked with two other solvers (GLPK and LINDO, a commercial solver) and both agree that the most optimal sulution of your second model is indeed 0 with all zero variables. It is the only possible solution of this model.

Peter


From: im6sous
Sent: Friday, November 06, 2009 19:38
To: lp_solve@...
Subject: [lp_solve] Re: Question on the result from two problems


 
Thank you for your very prompt answer.

I checked the function "solve" and it returns 0 (Feasable) for the problem I sent. Maybe it is because I'm using 5.5.0.11?

With Lp solve IDE 5.5.0.15 I obtain the following answer:
log:
Model name: 'LPSolver' - run #1
Objective: Maximize(R0)

SUBMITTED
Model size: 15 constraints, 234 variables, 260 non-zeros.
Sets: 0 GUB, 0 SOS.

[ +0 < Z < Inf ]

Using DUAL simplex for phase 1 and PRIMAL simplex for phase 2.
The primal and dual simplex pricing strategy set to 'Devex'.

Found feasibility by dual simplex after 18 iter.

Optimal solution 0 after 28 iter.
Excellent numeric accuracy ||*|| = 0

MEMO: lp_solve version 5.5.0.15 for 32 bit OS, with 64 bit REAL variables.
In the total iteration count 28, 11 (39.3%) were bound flips.
There were 2 refactorizations, 0 triggered by time and 0 by density.
... on average 8.5 major pivots per refactorization.
The largest [LUSOL v2.2.1.0] fact(B) had 24 NZ entries, 1.0x largest basis.
The constraint matrix inf-norm is 2566.67, with a dynamic range of 6.58779e+006.
Time to load data was 0.020 seconds, presolve used 0.021 seconds,
... 0.008 seconds in simplex solver, in total 0.049 seconds.

msg:
9:51:38 AM: Presolve
9:51:38 AM: LP Feasible

If I "save IDE options...." I get:
TPF0 TLPSolverLPSolverLagTrace
BoundsTighter SimplexType stDualPrimal
BasisCrashcNothingMessages mPresolve
mIterationmInvert mLpFeasible mLpBetter
mMiLpFeasible
mMiLpEqual mMiLpBetter mMiLpOptimal

Sincerely,
Cecile

--- In lp_solve@..., "William H. Patton" <pattonwh@...> wrote:

>
> You seem to be right on the cusp of infeasible +0.741043 C122 -C125 <= 0;
>
>
>
> I can solve this +0.741041673 C122 -C125 <= 0; but nothing much larger.
>
> +0.741041683 C122 -C125 <= 0;
>
> Found feasibility by dual simplex after 18 iter.
>
> The model is INFEASIBLE
>
> lp_solve unsuccessful after 80 iter and a last best value of -1e+030
>
> So I guess you forgot to check the return code before trying to fetch the
> solution to the infeasible problem.
>
> _____
>
> From: lp_solve@... [mailto:lp_solve@...] On Behalf
> Of Cecile Damiron
> Sent: Thursday, November 05, 2009 7:31 PM
> To: lp_solve@...
> Subject: [lp_solve] Question on the result from two problems [2 Attachments]
>
>
>
>
>
> [Attachment(s) from Cecile Damiron included below]
>
> Hello lpsolve community,
>
> I have an lp problem which gives me a set of results I'm confused about.
>
> The first problem "lp_1_0_PB52_2.txt" gives me the set of results I expect,
> but the second problem "lp_2_0_PB52_2.txt" which changes only by one
> constraint returns all the decision variables equal to 0 and it is what I
> don't understand.
>
> Problem 1:
> Partial results are:
> C122 = 333199.999999471
> C125 = 208529.127889622
>
> Problem 2:
> Add the constraint: +0.741043 C122 -C125 <= 0;;
> Partial results are:
> C122 = 0
> C125 = 0
> I would have expect something like:
> C122 = 281399.48 (=208529.127889622/0.741043)
> C125 = 208529.127889622
>
> In the problem 1: C122 and C125 are not related and in the problem 2 C122 is
> limited by C125. Maybe I'm missing something but I cannot figure out why I
> get a set of 0 in the problem2.
>
> Any help is welcome.
>
> Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15
> because it causes some crashes in the application I'm using. But if I check
> the results with Lp solve IDE 5.5.0.15 I obtain identical sets of results.
>
> Thanks for your help,
>
> Cecile
>

[peter_notebaert]

By the way, not a problem for lp_solve but for others and strange, you have a lot of bounds on variables that are otherwise not used in your model For example variable C286, but there are lots others.

Peter


From: Peter Notebaert
Sent: Friday, November 06, 2009 20:15
To: lp_solve@...
Subject: Re: [lp_solve] Re: Question on the result from two problems


 

I checked with two other solvers (GLPK and LINDO, a commercial solver) and both agree that the most optimal sulution of your second model is indeed 0 with all zero variables. It is the only possible solution of this model.

Peter


From: im6sous
Sent: Friday, November 06, 2009 19:38
To: lp_solve@...
Subject: [lp_solve] Re: Question on the result from two problems


 
Thank you for your very prompt answer.

I checked the function "solve" and it returns 0 (Feasable) for the problem I sent. Maybe it is because I'm using 5.5.0.11?

With Lp solve IDE 5.5.0.15 I obtain the following answer:
log:
Model name: 'LPSolver' - run #1
Objective: Maximize(R0)

SUBMITTED
Model size: 15 constraints, 234 variables, 260 non-zeros.
Sets: 0 GUB, 0 SOS.

[ +0 < Z < Inf ]

Using DUAL simplex for phase 1 and PRIMAL simplex for phase 2.
The primal and dual simplex pricing strategy set to 'Devex'.

Found feasibility by dual simplex after 18 iter.

Optimal solution 0 after 28 iter.
Excellent numeric accuracy ||*|| = 0

MEMO: lp_solve version 5.5.0.15 for 32 bit OS, with 64 bit REAL variables.
In the total iteration count 28, 11 (39.3%) were bound flips.
There were 2 refactorizations, 0 triggered by time and 0 by density.
... on average 8.5 major pivots per refactorization.
The largest [LUSOL v2.2.1.0] fact(B) had 24 NZ entries, 1.0x largest basis.
The constraint matrix inf-norm is 2566.67, with a dynamic range of 6.58779e+006.
Time to load data was 0.020 seconds, presolve used 0.021 seconds,
... 0.008 seconds in simplex solver, in total 0.049 seconds.

msg:
9:51:38 AM: Presolve
9:51:38 AM: LP Feasible

If I "save IDE options...." I get:
TPF0 TLPSolverLPSolverLagTrace
BoundsTighter SimplexType stDualPrimal
BasisCrashcNothingMessages mPresolve
mIterationmInvert mLpFeasible mLpBetter
mMiLpFeasible
mMiLpEqual mMiLpBetter mMiLpOptimal

Sincerely,
Cecile

--- In lp_solve@..., "William H. Patton" <pattonwh@...> wrote:

>
> You seem to be right on the cusp of infeasible +0.741043 C122 -C125 <= 0;
>
>
>
> I can solve this +0.741041673 C122 -C125 <= 0; but nothing much larger.
>
> +0.741041683 C122 -C125 <= 0;
>
> Found feasibility by dual simplex after 18 iter.
>
> The model is INFEASIBLE
>
> lp_solve unsuccessful after 80 iter and a last best value of -1e+030
>
> So I guess you forgot to check the return code before trying to fetch the
> solution to the infeasible problem.
>
> _____
>
> From: lp_solve@... [mailto:lp_solve@...] On Behalf
> Of Cecile Damiron
> Sent: Thursday, November 05, 2009 7:31 PM
> To: lp_solve@...
> Subject: [lp_solve] Question on the result from two problems [2 Attachments]
>
>
>
>
>
> [Attachment(s) from Cecile Damiron included below]
>
> Hello lpsolve community,
>
> I have an lp problem which gives me a set of results I'm confused about.
>
> The first problem "lp_1_0_PB52_2.txt" gives me the set of results I expect,
> but the second problem "lp_2_0_PB52_2.txt" which changes only by one
> constraint returns all the decision variables equal to 0 and it is what I
> don't understand.
>
> Problem 1:
> Partial results are:
> C122 = 333199.999999471
> C125 = 208529.127889622
>
> Problem 2:
> Add the constraint: +0.741043 C122 -C125 <= 0;;
> Partial results are:
> C122 = 0
> C125 = 0
> I would have expect something like:
> C122 = 281399.48 (=208529.127889622/0.741043)
> C125 = 208529.127889622
>
> In the problem 1: C122 and C125 are not related and in the problem 2 C122 is
> limited by C125. Maybe I'm missing something but I cannot figure out why I
> get a set of 0 in the problem2.
>
> Any help is welcome.
>
> Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15
> because it causes some crashes in the application I'm using. But if I check
> the results with Lp solve IDE 5.5.0.15 I obtain identical sets of results.
>
> Thanks for your help,
>
> Cecile
>

[Cecile Damiron]

Thanks again for such a reactivity.

I agree that I have "a lot of bounds on variables that are otherwise not used in your model".
I have an initial problem with all the variables used in the model. I transform the problem in order to get to a simpler problem and all the variables not used anymore can be deduced from the variable still "active" in the model. I'm doing simple algebra to get to it and I keep all the variable in the model just in order to maintain the indexes of all the variables. I assumed that having all the extra bounded variables did not penalize the lp. I can clean my code.
Anyway if I solve the initial problem I get to the same set of results. So I believe that it is not the issue.

For the result being a set of 0 in different solvers I might have done something wrong in the specification of my problem but I have trouble finding what it could be.

The reason I'm using lp solve is to calculate the rates of movement in a system. I build objects allowing users to make a dynamic simulation of their system. These objects allow to simulate the circulation of liquids (tank, valve, diverge blocks etc...). When the objects are put together, I use lpsolve to determine at each instant of the simulation how fast and where the flow moves in the system. lpsolve has been doing a very good job at solving these problems for about 3 years now.

A user sent me a model which didn't give him the expected result and I have narrowed the problem down to the problem I sent. I will keep working on it and I may find something I did wrong.

Thanks for your help,

Cecile

--- In lp_solve@..., "Peter Notebaert" <_peno_@...> wrote:

>
> By the way, not a problem for lp_solve but for others and strange, you have a lot of bounds on variables that are otherwise not used in your model For example variable C286, but there are lots others.
>
> Peter
>
>
> From: Peter Notebaert
> Sent: Friday, November 06, 2009 20:15
> To: lp_solve@...
> Subject: Re: [lp_solve] Re: Question on the result from two problems
>
>
>  
>
> I checked with two other solvers (GLPK and LINDO, a commercial solver) and both agree that the most optimal sulution of your second model is indeed 0 with all zero variables. It is the only possible solution of this model.
>
> Peter
>
>
> From: im6sous
> Sent: Friday, November 06, 2009 19:38
> To: lp_solve@...
> Subject: [lp_solve] Re: Question on the result from two problems
>
>
>  
> Thank you for your very prompt answer.
>
> I checked the function "solve" and it returns 0 (Feasable) for the problem I sent. Maybe it is because I'm using 5.5.0.11?
>
> With Lp solve IDE 5.5.0.15 I obtain the following answer:
> log:
> Model name: 'LPSolver' - run #1
> Objective: Maximize(R0)
>
> SUBMITTED
> Model size: 15 constraints, 234 variables, 260 non-zeros.
> Sets: 0 GUB, 0 SOS.
>
> [ +0 < Z < Inf ]
>
> Using DUAL simplex for phase 1 and PRIMAL simplex for phase 2.
> The primal and dual simplex pricing strategy set to 'Devex'.
>
> Found feasibility by dual simplex after 18 iter.
>
> Optimal solution 0 after 28 iter.
> Excellent numeric accuracy ||*|| = 0
>
> MEMO: lp_solve version 5.5.0.15 for 32 bit OS, with 64 bit REAL variables.
> In the total iteration count 28, 11 (39.3%) were bound flips.
> There were 2 refactorizations, 0 triggered by time and 0 by density.
> ... on average 8.5 major pivots per refactorization.
> The largest [LUSOL v2.2.1.0] fact(B) had 24 NZ entries, 1.0x largest basis.
> The constraint matrix inf-norm is 2566.67, with a dynamic range of 6.58779e+006.
> Time to load data was 0.020 seconds, presolve used 0.021 seconds,
> ... 0.008 seconds in simplex solver, in total 0.049 seconds.
>
> msg:
> 9:51:38 AM: Presolve
> 9:51:38 AM: LP Feasible
>
> If I "save IDE options...." I get:
> TPF0 TLPSolverLPSolverLagTrace
> BoundsTighter SimplexType stDualPrimal
> BasisCrashcNothingMessages mPresolve
> mIterationmInvert mLpFeasible mLpBetter
> mMiLpFeasible
> mMiLpEqual mMiLpBetter mMiLpOptimal
>
> Sincerely,
> Cecile
>
> --- In lp_solve@..., "William H. Patton" <pattonwh@> wrote:
> >
> > You seem to be right on the cusp of infeasible +0.741043 C122 -C125 <= 0;
> >
> >
> >
> > I can solve this +0.741041673 C122 -C125 <= 0; but nothing much larger.
> >
> > +0.741041683 C122 -C125 <= 0;
> >
> > Found feasibility by dual simplex after 18 iter.
> >
> > The model is INFEASIBLE
> >
> > lp_solve unsuccessful after 80 iter and a last best value of -1e+030
> >
> > So I guess you forgot to check the return code before trying to fetch the
> > solution to the infeasible problem.
> >
> > _____
> >
> > From: lp_solve@... [mailto:lp_solve@...] On Behalf
> > Of Cecile Damiron
> > Sent: Thursday, November 05, 2009 7:31 PM
> > To: lp_solve@...
> > Subject: [lp_solve] Question on the result from two problems [2 Attachments]
> >
> >
> >
> >
> >
> > [Attachment(s) from Cecile Damiron included below]
> >
> > Hello lpsolve community,
> >
> > I have an lp problem which gives me a set of results I'm confused about.
> >
> > The first problem "lp_1_0_PB52_2.txt" gives me the set of results I expect,
> > but the second problem "lp_2_0_PB52_2.txt" which changes only by one
> > constraint returns all the decision variables equal to 0 and it is what I
> > don't understand.
> >
> > Problem 1:
> > Partial results are:
> > C122 = 333199.999999471
> > C125 = 208529.127889622
> >
> > Problem 2:
> > Add the constraint: +0.741043 C122 -C125 <= 0;;
> > Partial results are:
> > C122 = 0
> > C125 = 0
> > I would have expect something like:
> > C122 = 281399.48 (=208529.127889622/0.741043)
> > C125 = 208529.127889622
> >
> > In the problem 1: C122 and C125 are not related and in the problem 2 C122 is
> > limited by C125. Maybe I'm missing something but I cannot figure out why I
> > get a set of 0 in the problem2.
> >
> > Any help is welcome.
> >
> > Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15
> > because it causes some crashes in the application I'm using. But if I check
> > the results with Lp solve IDE 5.5.0.15 I obtain identical sets of results.
> >
> > Thanks for your help,
> >
> > Cecile
> >
>

[William H. Patton]

I think your model is probably reacting correctly.

As a real physical model should, you have noise at around 5 or less digits
of precision.

There is a very narrow range near your additional constraint where the model
is infeasible due to this noise.

Otherwse, It seems like C122=C121 and the additional constraint is acting
like a hard capacity limit or a fuse.

It shuts down the fluid flow in the system when a circuit limit blows the
fuse.

 

You could look at elastic constraints to deal with the infeasibility that
pops up.

Or check for the all 0 shutdown return solution and then message about the
"fuse" that just shut down the flow

during your recent iteration of the problem.

http://lpsolve.sourceforge.net/5.5/Infeasible.htm

 

 

  _____  

From: lp_solve@... [mailto:lp_solve@...] On Behalf
Of im6sous
Sent: Friday, November 06, 2009 5:40 PM
To: lp_solve@...
Subject: [lp_solve] Re: Question on the result from two problems

 

 

Thanks again for such a reactivity.

I agree that I have "a lot of bounds on variables that are otherwise not
used in your model".
I have an initial problem with all the variables used in the model. I
transform the problem in order to get to a simpler problem and all the
variables not used anymore can be deduced from the variable still "active"
in the model. I'm doing simple algebra to get to it and I keep all the
variable in the model just in order to maintain the indexes of all the
variables. I assumed that having all the extra bounded variables did not
penalize the lp. I can clean my code.
Anyway if I solve the initial problem I get to the same set of results. So I
believe that it is not the issue.

For the result being a set of 0 in different solvers I might have done
something wrong in the specification of my problem but I have trouble
finding what it could be.

The reason I'm using lp solve is to calculate the rates of movement in a
system. I build objects allowing users to make a dynamic simulation of their
system. These objects allow to simulate the circulation of liquids (tank,
valve, diverge blocks etc...). When the objects are put together, I use
lpsolve to determine at each instant of the simulation how fast and where
the flow moves in the system. lpsolve has been doing a very good job at
solving these problems for about 3 years now.

A user sent me a model which didn't give him the expected result and I have
narrowed the problem down to the problem I sent. I will keep working on it
and I may find something I did wrong.

Thanks for your help,

Cecile

--- In lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com, "Peter
Notebaert" <_peno_@...> wrote:
>
> By the way, not a problem for lp_solve but for others and strange, you
have a lot of bounds on variables that are otherwise not used in your model
For example variable C286, but there are lots others.

>
> Peter
>
>
> From: Peter Notebaert
> Sent: Friday, November 06, 2009 20:15
> To: lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com
> Subject: Re: [lp_solve] Re: Question on the result from two problems
>
>
>
>
> I checked with two other solvers (GLPK and LINDO, a commercial solver) and

both agree that the most optimal sulution of your second model is indeed 0
with all zero variables. It is the only possible solution of this model.

>
> Peter
>
>
> From: im6sous
> Sent: Friday, November 06, 2009 19:38
> To: lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com
> Subject: [lp_solve] Re: Question on the result from two problems
>
>
>
> Thank you for your very prompt answer.
>
> I checked the function "solve" and it returns 0 (Feasable) for the problem

I sent. Maybe it is because I'm using 5.5.0.11?

>
> With Lp solve IDE 5.5.0.15 I obtain the following answer:
> log:
> Model name: 'LPSolver' - run #1
> Objective: Maximize(R0)
>
> SUBMITTED
> Model size: 15 constraints, 234 variables, 260 non-zeros.
> Sets: 0 GUB, 0 SOS.
>
> [ +0 < Z < Inf ]
>
> Using DUAL simplex for phase 1 and PRIMAL simplex for phase 2.
> The primal and dual simplex pricing strategy set to 'Devex'.
>
> Found feasibility by dual simplex after 18 iter.
>
> Optimal solution 0 after 28 iter.
> Excellent numeric accuracy ||*|| = 0
>
> MEMO: lp_solve version 5.5.0.15 for 32 bit OS, with 64 bit REAL variables.
> In the total iteration count 28, 11 (39.3%) were bound flips.
> There were 2 refactorizations, 0 triggered by time and 0 by density.
> ... on average 8.5 major pivots per refactorization.
> The largest [LUSOL v2.2.1.0] fact(B) had 24 NZ entries, 1.0x largest

basis.
> The constraint matrix inf-norm is 2566.67, with a dynamic range of
6.58779e+006.

> Time to load data was 0.020 seconds, presolve used 0.021 seconds,
> ... 0.008 seconds in simplex solver, in total 0.049 seconds.
>
> msg:
> 9:51:38 AM: Presolve
> 9:51:38 AM: LP Feasible
>
> If I "save IDE options...." I get:
> TPF0 TLPSolverLPSolverLagTrace
> BoundsTighter SimplexType stDualPrimal
> BasisCrashcNothingMessages mPresolve
> mIterationmInvert mLpFeasible mLpBetter
> mMiLpFeasible
> mMiLpEqual mMiLpBetter mMiLpOptimal
>
> Sincerely,
> Cecile
>
> --- In lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com,

"William H. Patton" <pattonwh@> wrote:
> >
> > You seem to be right on the cusp of infeasible +0.741043 C122 -C125 <=
0;

> >
> >
> >
> > I can solve this +0.741041673 C122 -C125 <= 0; but nothing much larger.
> >
> > +0.741041683 C122 -C125 <= 0;
> >
> > Found feasibility by dual simplex after 18 iter.
> >
> > The model is INFEASIBLE
> >
> > lp_solve unsuccessful after 80 iter and a last best value of -1e+030
> >
> > So I guess you forgot to check the return code before trying to fetch

the
> > solution to the infeasible problem.
> >
> > _____
> >
> > From: lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com
[mailto:lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com] On
Behalf
> > Of Cecile Damiron
> > Sent: Thursday, November 05, 2009 7:31 PM
> > To: lp_solve@yahoogroup <mailto:lp_solve%40yahoogroups.com> s.com
> > Subject: [lp_solve] Question on the result from two problems [2
Attachments]

> >
> >
> >
> >
> >
> > [Attachment(s) from Cecile Damiron included below]
> >
> > Hello lpsolve community,
> >
> > I have an lp problem which gives me a set of results I'm confused about.
> >
> > The first problem "lp_1_0_PB52_2.txt" gives me the set of results I

expect,
> > but the second problem "lp_2_0_PB52_2.txt" which changes only by one
> > constraint returns all the decision variables equal to 0 and it is what
I

> > don't understand.
> >
> > Problem 1:
> > Partial results are:
> > C122 = 333199.999999471
> > C125 = 208529.127889622
> >
> > Problem 2:
> > Add the constraint: +0.741043 C122 -C125 <= 0;;
> > Partial results are:
> > C122 = 0
> > C125 = 0
> > I would have expect something like:
> > C122 = 281399.48 (=208529.127889622/0.741043)
> > C125 = 208529.127889622
> >
> > In the problem 1: C122 and C125 are not related and in the problem 2

C122 is
> > limited by C125. Maybe I'm missing something but I cannot figure out why
I
> > get a set of 0 in the problem2.
> >
> > Any help is welcome.
> >
> > Remark: I'm using the 5.5.0.11 dll of lpSolve, I do not use the 5.5.0.15
> > because it causes some crashes in the application I'm using. But if I
check
> > the results with Lp solve IDE 5.5.0.15 I obtain identical sets of
results.
> >
> > Thanks for your help,
> >
> > Cecile
> >
>