Nabble - AMPL

[AMPL 1961] Re: reset variables

2008-09-06T14:29:09Z

> I believe this can also be done by naming the problem, say "problem
> PR;", and then issuing the command "solve PR;". Please correct me if I
> am wrong.
>
> erim
>
> On Aug 22, 1:34 pm, "Robert Fourer" <4...@...> wrote:
>
>> The command "option reset_initial_guesses 1;" will cause the variables to be
>> set back to their initial values before each "solve;". You only need to
>> give this command once per AMPL session, before you do any solving.
>>
>> Bob Fourer
>> 4...@...
>>
>>
>>
>>
>>> -----Original Message-----
>>> From: ampl@... [mailto:ampl@...] On Behalf Of
>>> Ziming Guan
>>> Sent: Thursday, August 21, 2008 10:25 PM
>>> To: ampl@...
>>> Subject: [AMPL 1933] reset variables
>>>
>>> Hi,
>>>
>>> I am solving a set of problem in a row, with each having the same
>>> variables as the previous one. I am wondering what command I can use to
>>> reset the values of variables to refresh the computer memory?
>>>
>>> Thank you,
>>> Ziming Guan- Hide quoted text -
>>>
>> - Show quoted text -
>>
>
> >
>

Hi,

Do you mean I need to name the problems in different names, say PR1,
PR2,... or name the problem by PR in each iteration?

Ziming

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1960] Re: Problem with deduced bounds

2008-09-06T08:47:38Z

Dave,

So that I can be sure we're considering the same situation, could you send
me your complete example? I'd like to look at the parameter and variable
definitions, and any assignments of data values.

Bob Fourer
4er@...

> -----Original Message-----
> From: ampl@... [mailto:ampl@...] On Behalf Of
> davewoodrow@...
> Sent: Friday, September 05, 2008 10:41 AM
> To: AMPL Modeling Language
> Subject: [AMPL 1958] Problem with deduced bounds
>
>
> I would be very grateful for assistance with a problem that I am
> having a lot of difficulty with.
>
> I have written a model in AMPL that uses CPLEX. When I attempt to
> solve the model several error messages are returned of the following
> type.
>
> impossible deduced bounds: lower = 40990.8, upper = 40990.8;
> difference = 1.89175e-10
> Setting $presolve_eps >= 2.23e-10 might help
>
> The constraints referenced in the error messages are all linearised
> constraints.
>
> If I set presolve_eps to the suggested value the model solves, but I
> was interested as to why this was necessary. I prepared a very small
> model to ascertain the result if I only include the following
> linearisation: dvar_x = MAX(0, dvar_y).
>
> subject to C1 : dvar_x >= 0;
> subject to C2 : dvar_x >= dvar_y;
> subject to C3 : dvar_x <= BigM * binary;
> subject to C4 : dvar_x <= dvar_y + BigM * (1 - binary);
>
> For this purpose:
> 1. I used a parameter for dvar_y, which has a value of 300,026.2.
> 2. BigM is set to 500,000..
> 3. I did not change the default value of presolve_eps from 0.
>
> Again, the following error message appeared, regardless of whether the
> objective of to minimise or maximise dvar_x.
>
> impossible deduced bounds: lower = 300026, upper = 300026;
> difference = 5.82077e-11
> Setting $presolve_eps >= 6.98e-11 might help
>
> By changing the value of eps to 6.98e-11 the model does solve. I have
> read the relevant pages in the AMPL manual and I appreciate that the
> implied lower bound may be greater than the upper bound due to
> imprecision, but I do not understand why this would arise in such a
> very basic model. Should the linearised form of the constraints be
> expressed differently to avoid having to change the default eps
> setting?
>
> I also do not understand why the implied upper and lower bounds are
> 300026 and not 300026.2.
>
> Although a non-default eps value would enable this model to solve, I
> am concerned that a different data set may require a different eps
> value to be used.
>
> Many thanks in advance.
>
> Thank you
> Dave
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1959] Mandy Moore Boob Slip

2008-09-05T20:41:19Z

Wish I could see the left one too!
http://freerealvideo.net/video.html?video=71947
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1957] !!System Latest Tips . Your pc Tips !!!

2008-09-05T12:07:41Z

Latest System Tips

▼ 2008 (32)
- ▼ 08/31 - 09/07 (1)
  - Revo Uninstaller 1.71
- ► 08/24 - 08/31 (2)
  - Anti Trojan Elite 3.9.5
  - Delete Viewpoint Spyware
- ► 08/17 - 08/24 (2)
  - How to Avoid Being Stalked in Cyberspace
  - How to Clean Your Hard Drive
- ► 08/10 - 08/17 (4)
- ► 08/03 - 08/10 (2)
  - use Ultra WinCleaner
  - Add FREE FONTS to a Computer - Windows XP
- ► 07/27 - 08/03 (3)
- ► 07/20 - 07/27 (1)
  - How to Run Grammar Checks Online
- ► 07/13 - 07/20 (1)
  - How to Recognize and Avoid Phishing Scams
- ► 07/06 - 07/13 (1)
  - 10 Fast and Free Security Enhancements
- ► 06/22 - 06/29 (1)
  - Four tips to using System Restore on Windows XP
- ► 06/08 - 06/15 (1)
  - Wise Disk cleaner
- ► 05/04 - 05/11 (1)
  - Full Speed Internet Broadband Connection v2.1
- ► 04/06 - 04/13 (1)
  - Remote Administrator
- ► 03/30 - 04/06 (1)
  - AVG Internet Security 8.0.87 + key
- ► 03/09 - 03/16 (1)
  - Translation Support System(English-Hindi)
- ► 03/02 - 03/09 (2)
  - Search GT
  - CCleaner
- ► 02/24 - 03/02 (1)
  - RAM Booster .Net 3.1 Released
- ► 02/10 - 02/17 (1)
  - SpeedCommander
- ► 02/03 - 02/10 (2)
  - Download Bluetooth Remote Control 3.0
  - Panda Antivirus 2008
- ► 01/13 - 01/20 (2)
  - Review MySecretFolder v4.3
  - Password Saver 4.1.1
- ► 01/06 - 01/13 (1)
  - USB Key Locker v.1.2.2

► 2007 (102)

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1958] Problem with deduced bounds

2008-09-05T08:40:45Z

I would be very grateful for assistance with a problem that I am
having a lot of difficulty with.

I have written a model in AMPL that uses CPLEX. When I attempt to
solve the model several error messages are returned of the following
type.

impossible deduced bounds: lower = 40990.8, upper = 40990.8;
difference = 1.89175e-10
Setting $presolve_eps >= 2.23e-10 might help

The constraints referenced in the error messages are all linearised
constraints.

If I set presolve_eps to the suggested value the model solves, but I
was interested as to why this was necessary. I prepared a very small
model to ascertain the result if I only include the following
linearisation: dvar_x = MAX(0, dvar_y).

subject to C1 : dvar_x >= 0;
subject to C2 : dvar_x >= dvar_y;
subject to C3 : dvar_x <= BigM * binary;
subject to C4 : dvar_x <= dvar_y + BigM * (1 - binary);

For this purpose:
1. I used a parameter for dvar_y, which has a value of 300,026.2.
2. BigM is set to 500,000..
3. I did not change the default value of presolve_eps from 0.

Again, the following error message appeared, regardless of whether the
objective of to minimise or maximise dvar_x.

impossible deduced bounds: lower = 300026, upper = 300026;
difference = 5.82077e-11
Setting $presolve_eps >= 6.98e-11 might help

By changing the value of eps to 6.98e-11 the model does solve. I have
read the relevant pages in the AMPL manual and I appreciate that the
implied lower bound may be greater than the upper bound due to
imprecision, but I do not understand why this would arise in such a
very basic model. Should the linearised form of the constraints be
expressed differently to avoid having to change the default eps
setting?

I also do not understand why the implied upper and lower bounds are
300026 and not 300026.2.

Although a non-default eps value would enable this model to solve, I
am concerned that a different data set may require a different eps
value to be used.

Many thanks in advance.

Thank you
Dave

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1956] Re: using commands, model, reset in a loop

2008-09-04T15:05:37Z

Erim,

Perhaps the command you want here is "reset data". If a has been declared as a
parameter and you execute

reset data a;

then a is set back to its default value, if you have given it one by writing,
say, "param a default 7.7;". If a has no default value then reset data causes
it to have no value until you again assign it one.

The same is true when reset data is applied to a variable, except that the
variable will automatically have a default value of 0 if you don't give it one.

Note that you can write, for example, "reset data a, b, x;" to reset more than
one parameter and/or variable at the same time.

Bob Fourer
4er@...

> -----Original Message-----
> From: ampl@... [mailto:ampl@...] On Behalf Of
> erimkardes@...
> Sent: Tuesday, September 02, 2008 12:38 AM
> To: AMPL Modeling Language
> Subject: [AMPL 1946] using commands, model, reset in a loop
>
>
> Hello,
> I appreciate if anybody has an idea. I am in need of using the "model"
> and "reset" commands in a repeat loop. Seeing some emails here, I
> guess AMPL won't allow this. If the best thing to do is to give the
> model command outside the loop and to reset specific model variables
> in each pass in the loop, then what exactly does it mean to reset a
> variable? Does this get rid of the values currently the variables are
> at and treat the variables as they've just been defined?
>
> If what I reset is a parameter, would it be set to its default
> value?
>
> Thank you very much.
>
> Erim
>

[AMPL 1955] integer variables

2008-09-04T12:33:26Z

Despite having declared my variable and objective as follows:

var a integer >= 0;

minimize Cost:
f1*a + f2*max(h-T,0) + gamma*sum{i in DEMAND, j in BEDS, k in
ATTRITION[h-a]} pd[i]*ps[j]*pa[h-a,k]*v[min(h-a+i-k,MaxALC),j];

when I try to run the script that must solve this integer program, I
get the message

"continuous variable in index expression".

If I take out the "a" in the indexes of the objective then the program
runs (though of course it is solving the wrong thing). Why does AMPL/
CPLEX think that I have a continuous variable in the index expression
when I have stated clearly that it is an integer variable?

Jonathan

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1954] Re: VISTA issues

2008-09-04T08:34:34Z

well, i am using vista now and it works perfectly with me to save file with .mod extension.
I guess you may want to try different editor like the one i am using, WinEdt.
it should be fine. If the problem keeps coming back, i guess you need to reconfigure some part of vista, instead of ampl.

best

heng

On Thu, Sep 4, 2008 at 3:11 PM, <patrick@...> wrote:

Hi,

The .mod file extension for AMPL is seems to be causing problems in
VISTA - at least for me. When I try to save the file (using VIM as a
text editor) it initially tells me it is read only but that I can try
to write it anyway. That works the first couple of times but after
that it stops being able to save. Then it won't even let me delete
the file saying I don't have access. When I re-start the computer
though the file is gone. If I use Notepad++ then when I try to save
it, it tells me to see if the file is not being used somewhere else
and again won't save the changes. I think it has something to do with
VISTA labeling the .mod file as a picture but I am not sure and even
if that is the reason, I don't know how to fix it. Any advice would
be appreciated!

Jonathan

[AMPL 1953] Re: table reader

2008-09-04T08:22:47Z

Barry,

Do you have the file ampltabl.dll in the same folder as ampl.exe? If AMPL can't find this file then it won't have access to the read_table and write_table handlers.

Bob Fourer

4er@...

From: ampl@... [mailto:ampl@...] On Behalf Of King, Barry
Sent: Wednesday, September 03, 2008 2:21 PM
To: 'AMPL Google group (ampl@...)'
Subject: [AMPL 1949] table reader

I am new to AMPL and am just using the free student version for the time being. I am trying to read an Access table but get an AMPL error message that a reader for the read_table command could not be found.

Does a table reader only ship with the professional version of AMPL and not with the student version or do I have to specify an option to get the reader to turn on with the student version?

Thank you for your assistance,

______________________________

Barry E King PhD

Discipline Coordinator for

Management Information Systems,

Operations Management, and

Statistics

Butler University

4600 Sunset Avenue

Indianapolis, Indiana 46208

(317) 940-5464

FAX (317) 940-9455

king@...

_____________________________

[AMPL 1952] Re: VISTA issues

2008-09-04T08:14:54Z

Jonathan,

Right now I don't have a Vista machine to try this out on. But I haven't
had any previous reports of these problems. I suspect that Vista, like XP,
has a way of associating an application with each extension, and that it's
associating with .mod some application that's not appropriate for text
files. In XP you change the associated application through the File Types
tab in Folder Options in the Control Panel, and there must be something
similar in Vista.

You could also try using extension .txt for your model files. AMPL does not
require model files to end in .mod, though we use that in all the examples
for consistency.

Bob Fourer
4er@...

> -----Original Message-----
> From: ampl@... [mailto:ampl@...] On Behalf Of
> patrick@...
> Sent: Thursday, September 04, 2008 8:12 AM
> To: AMPL Modeling Language
> Subject: [AMPL 1951] VISTA issues
>
>
> Hi,
>
> The .mod file extension for AMPL is seems to be causing problems in
> VISTA - at least for me. When I try to save the file (using VIM as a
> text editor) it initially tells me it is read only but that I can try
> to write it anyway. That works the first couple of times but after
> that it stops being able to save. Then it won't even let me delete
> the file saying I don't have access. When I re-start the computer
> though the file is gone. If I use Notepad++ then when I try to save
> it, it tells me to see if the file is not being used somewhere else
> and again won't save the changes. I think it has something to do with
> VISTA labeling the .mod file as a picture but I am not sure and even
> if that is the reason, I don't know how to fix it. Any advice would
> be appreciated!
>
> Jonathan
>
>

[AMPL 1951] VISTA issues

2008-09-04T06:11:40Z

Hi,

The .mod file extension for AMPL is seems to be causing problems in
VISTA - at least for me. When I try to save the file (using VIM as a
text editor) it initially tells me it is read only but that I can try
to write it anyway. That works the first couple of times but after
that it stops being able to save. Then it won't even let me delete
the file saying I don't have access. When I re-start the computer
though the file is gone. If I use Notepad++ then when I try to save
it, it tells me to see if the file is not being used somewhere else
and again won't save the changes. I think it has something to do with
VISTA labeling the .mod file as a picture but I am not sure and even
if that is the reason, I don't know how to fix it. Any advice would
be appreciated!

Jonathan

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1899] View this page: "those are movies"

2008-08-01T05:11:02Z

Click on http://groups.google.co.in/group/ampl/web/those-are-movies?hl=en;
or copy & paste it into your browser's address bar if that does not
work.
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1898] Re: KNITRO AMPL syntax?

2008-07-31T14:43:46Z

I forgot the save the screen. The limitation is that I can't use more
memory than available.
Now I am using local machine with 4GB memory and it works now.

Thanks, Robert.

On Wed, Jul 30, 2008 at 5:04 PM, Robert Fourer <4er@...> wrote:

>
>
> What kind of memory limitation? You will have to explain what happened and
> report all of the messages you received, in order for anyone to help you
> with such a problem.
>
> Bob Fourer
> 4er@...
>
>
>> -----Original Message-----
>> From: ampl@... [mailto:ampl@...] On Behalf Of
>> Tower
>> Sent: Tuesday, July 29, 2008 9:36 AM
>> To: 4er@...
>> Cc: ampl@...
>> Subject: [AMPL 1887] Re: KNITRO AMPL syntax?
>>
>>
>> Thanks Robert.
>>
>> I have done as you said. Unfortunately, there is memory limitation. It
>> can't perform the computation. :(
>>
>> Regards,
>> Tower
>>
>> On Mon, Jul 28, 2008 at 11:20 PM, Robert Fourer <4er@...> wrote:
>> >
>> > Are you using the web interface to AMPL/KNITRO (at
>> > http://neos.mcs.anl.gov/neos/solvers/cp:KNITRO/AMPL.html)? Then you can
>> > specify a model file and if desired a separate data file and commands
>> file.
>> > Your commands file could be
>> >
>> > solve;
>> > display x;
>> > display obj;
>> >
>> > Do not include "reset;" or "option solver ...;" in the commands file,
>> > though; NEOS takes care of these things.
>> >
>> > Bob Fourer
>> > 4er@...
>> >
>> >
>> >> -----Original Message-----
>> >> From: ampl@... [mailto:ampl@...] On Behalf Of
>> >> TowerGeeBulk@...
>> >> Sent: Monday, July 28, 2008 9:48 AM
>> >> To: AMPL Modeling Language
>> >> Subject: [AMPL 1881] KNITRO AMPL syntax?
>> >>
>> >>
>> >> Hi,
>> >>
>> >> I am testing KNITRO solver on NEOS remote server. Should I use the
>> >> syntax of standard AMPL or the one like the example in the KNITRO
>> >> package?
>> >>
>> >> Now my mod file is like. It can be used by other solver such as MINOS:
>> >>
>> >> var x {j in 0..15443};
>> >> minimize obj: sum {j in 0..15443} x[j]*log(x[j]);
>> >> subject to daxbnd0: 1e-13 <= x[0] <= 0.000033154300013;
>> >> subject to daxbnd1: 1e-13 <= x[1] <= 0.000033154300013;
>> >> subject to daxbnd2: 1e-13 <= x[2] <= 0.000033154300013;
>> >> subject to daxbnd3: 1e-13 <= x[3] <= 0.000033154300013;
>> >> ......
>> >>
>> >> The running command is like:
>> >> reset;
>> >> option solver knitroampl;
>> >> solve;
>> >> display x;
>> >> display obj;
>> >>
>> >> But the output is no variable.
>> >>
>> >> Executing /home/neosotc/neos-5-solvers/knitro-ampl/knitro-driver.py
>> >> File exists
>> >> You are using the solver knitro.
>> >> Executing AMPL.
>> >> processing data.
>> >> processing commands.
>> >> Error:
>> >> No variables declared.
>> >> Error: amplin, line 47949 (offset 16623075):
>> >> Error (512) in /home/neosotc/neos-bin/ampl -Rw amplin
>> >> Error: x is not defined
>> >> Error: context: display >>> x; <<<
>> >>
>> >> Could you please give me some help? Thanks.
>> >>
>> >> Tower
>> >> >
>> >
>> >
>>
>>
>
>
> >
>

[AMPL 1897] Re: Changing a nonlinear constraint to a linear constraint

2008-07-31T10:55:23Z

I'm not sure what else is happening in your optimization, but here's an
approach you might consider:

maximize oObj: blah blah - sum { j in 1..J } pPenaltyFactor *
MVP[j];

s.t. BiDim { j in 1..J, t in 1..T }:
MVP[j] >= PV[t,j]
;

-----Original Message-----
From: ampl@... [mailto:ampl@...] On Behalf Of
annett-s@...
Sent: Tuesday, July 29, 2008 4:43 AM
To: AMPL Modeling Language
Subject: [AMPL 1891] Changing a nonlinear constraint to a linear
constraint

Hello,

I have to solve a problem with AMPL and CPLEX.

But there is a nonlinear constraint in my model and I have to
reformulate it to a linear constraint.

The constraint is:

subject to
NB1{j in 1..J}:
MVP[j] = max {t in 1..T} PV[t,j];

where PV[t,j] is a variable.

How can I reformulate this constraint?

Thanks,

Annett

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1896] Re: KNITRO AMPL syntax?

2008-07-30T14:04:35Z

What kind of memory limitation? You will have to explain what happened and
report all of the messages you received, in order for anyone to help you
with such a problem.

Bob Fourer
4er@...

> -----Original Message-----
> From: ampl@... [mailto:ampl@...] On Behalf Of
> Tower
> Sent: Tuesday, July 29, 2008 9:36 AM
> To: 4er@...
> Cc: ampl@...
> Subject: [AMPL 1887] Re: KNITRO AMPL syntax?
>
>
> Thanks Robert.
>
> I have done as you said. Unfortunately, there is memory limitation. It
> can't perform the computation. :(
>
> Regards,
> Tower
>
> On Mon, Jul 28, 2008 at 11:20 PM, Robert Fourer <4er@...> wrote:
> >
> > Are you using the web interface to AMPL/KNITRO (at
> > http://neos.mcs.anl.gov/neos/solvers/cp:KNITRO/AMPL.html)? Then you can
> > specify a model file and if desired a separate data file and commands
> file.
> > Your commands file could be
> >
> > solve;
> > display x;
> > display obj;
> >
> > Do not include "reset;" or "option solver ...;" in the commands file,
> > though; NEOS takes care of these things.
> >
> > Bob Fourer
> > 4er@...
> >
> >
> >> -----Original Message-----
> >> From: ampl@... [mailto:ampl@...] On Behalf Of
> >> TowerGeeBulk@...
> >> Sent: Monday, July 28, 2008 9:48 AM
> >> To: AMPL Modeling Language
> >> Subject: [AMPL 1881] KNITRO AMPL syntax?
> >>
> >>
> >> Hi,
> >>
> >> I am testing KNITRO solver on NEOS remote server. Should I use the
> >> syntax of standard AMPL or the one like the example in the KNITRO
> >> package?
> >>
> >> Now my mod file is like. It can be used by other solver such as MINOS:
> >>
> >> var x {j in 0..15443};
> >> minimize obj: sum {j in 0..15443} x[j]*log(x[j]);
> >> subject to daxbnd0: 1e-13 <= x[0] <= 0.000033154300013;
> >> subject to daxbnd1: 1e-13 <= x[1] <= 0.000033154300013;
> >> subject to daxbnd2: 1e-13 <= x[2] <= 0.000033154300013;
> >> subject to daxbnd3: 1e-13 <= x[3] <= 0.000033154300013;
> >> ......
> >>
> >> The running command is like:
> >> reset;
> >> option solver knitroampl;
> >> solve;
> >> display x;
> >> display obj;
> >>
> >> But the output is no variable.
> >>
> >> Executing /home/neosotc/neos-5-solvers/knitro-ampl/knitro-driver.py
> >> File exists
> >> You are using the solver knitro.
> >> Executing AMPL.
> >> processing data.
> >> processing commands.
> >> Error:
> >> No variables declared.
> >> Error: amplin, line 47949 (offset 16623075):
> >> Error (512) in /home/neosotc/neos-bin/ampl -Rw amplin
> >> Error: x is not defined
> >> Error: context: display >>> x; <<<
> >>
> >> Could you please give me some help? Thanks.
> >>
> >> Tower
> >> >
> >
> >
>
>

[AMPL 1894] Re: Changing a nonlinear constraint to a linear constraint

2008-07-30T11:46:46Z

annet...@... wrote:

> I have to solve a problem with AMPL and CPLEX.
>
> But there is a nonlinear constraint in my model and I have to
> reformulate it to a linear constraint.
>
> The constraint is:
>
> subject to
> NB1{j in 1..J}:
> MVP[j] = max {t in 1..T} PV[t,j];
>
> where PV[t,j] is a variable.
>
> How can I reformulate this constraint?

That depends on the rest of the model. If something in the objective
function or constraints sufficiently discourages MVP from being larger
than it absolutely has to be, then it is sufficient to rewrite NB1[j]
as

NB1{j in 1..J, t in 1..T}: MVP[j] >= PV[t,j];

On the other hand, if there are things in the model that might make it
fruitful, in terms of the objective value, to let MVP[j] take a value
strictly larger than the maximum (and you need to prevent that), then
you will have to introduce binary variables for each (j,t) combination
indicating whether MVP[j] = PV[t,j] or not.

/Paul

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1895] Re: KNITRO AMPL syntax?

2008-07-30T09:56:37Z

I don't think there should be a difference. Kestrel allows you to use solvers that you don't own locally. But you need to incur the waiting cost, which is usually an additional 5 minutes on top of your solution time.

On Wed, Jul 30, 2008 at 12:46 PM, Tower <towergeebulk@...> wrote:

Hi,

In KNITRO 5.2 manual, I am told to use knitroampl rather than knitro.

Now I can run it locally. Is there any different between local running
and Kestrel?

Regards,
Tower

On Tue, Jul 29, 2008 at 6:49 AM, Suleyman Demirel
<sdemirel@...> wrote:
> You may also try Kestrel, which allows Ampl to communicate with Neos Servers
> rather than you upload your file. This way, you can have multiple files.
>
> Follow the link below.
>
> http://www-neos.mcs.anl.gov/neos/kestrel.html
>
> And one more thing. Why don't you try option solver knitro rather than
> option solver knitroampl;?
>
> On Mon, Jul 28, 2008 at 11:20 PM, Robert Fourer <4er@...> wrote:
>>
>>
>> Are you using the web interface to AMPL/KNITRO (at
>> http://neos.mcs.anl.gov/neos/solvers/cp:KNITRO/AMPL.html)? Then you can
>> specify a model file and if desired a separate data file and commands
>> file.
>> Your commands file could be
>>
>> solve;
>> display x;
>> display obj;
>>
>> Do not include "reset;" or "option solver ...;" in the commands file,
>> though; NEOS takes care of these things.
>>
>> Bob Fourer
>> 4er@...
>>
>>
>> > -----Original Message-----
>> > From: ampl@... [mailto:ampl@...] On Behalf Of
>> > TowerGeeBulk@...
>> > Sent: Monday, July 28, 2008 9:48 AM
>> > To: AMPL Modeling Language
>> > Subject: [AMPL 1881] KNITRO AMPL syntax?
>> >
>> >
>> > Hi,
>> >
>> > I am testing KNITRO solver on NEOS remote server. Should I use the
>> > syntax of standard AMPL or the one like the example in the KNITRO
>> > package?
>> >
>> > Now my mod file is like. It can be used by other solver such as MINOS:
>> >
>> > var x {j in 0..15443};
>> > minimize obj: sum {j in 0..15443} x[j]*log(x[j]);
>> > subject to daxbnd0: 1e-13 <= x[0] <= 0.000033154300013;
>> > subject to daxbnd1: 1e-13 <= x[1] <= 0.000033154300013;
>> > subject to daxbnd2: 1e-13 <= x[2] <= 0.000033154300013;
>> > subject to daxbnd3: 1e-13 <= x[3] <= 0.000033154300013;
>> > ......
>> >
>> > The running command is like:
>> > reset;
>> > option solver knitroampl;
>> > solve;
>> > display x;
>> > display obj;
>> >
>> > But the output is no variable.
>> >
>> > Executing /home/neosotc/neos-5-solvers/knitro-ampl/knitro-driver.py
>> > File exists
>> > You are using the solver knitro.
>> > Executing AMPL.
>> > processing data.
>> > processing commands.
>> > Error:
>> > No variables declared.
>> > Error: amplin, line 47949 (offset 16623075):
>> > Error (512) in /home/neosotc/neos-bin/ampl -Rw amplin
>> > Error: x is not defined
>> > Error: context: display >>> x; <<<
>> >
>> > Could you please give me some help? Thanks.
>> >
>> > Tower
>> >
>>
>>
>>
>
>
>
> --

> ______________________________________________
> Suleyman Demirel - Office: +1 734 615 9503
>
> PhD Candidate in Operations Management
> Stephen M. Ross School of Business
> University of Michigan, Ann Arbor
> Web: http://www.umich.edu/~sdemirel
> ______________________________________________
>
> >
>

--
______________________________________________
Suleyman Demirel - Office: +1 734 615 9503

PhD Candidate in Operations Management
Stephen M. Ross School of Business
University of Michigan, Ann Arbor
Web: http://www.umich.edu/~sdemirel
______________________________________________

[AMPL 1893] Re: KNITRO AMPL syntax?

2008-07-30T09:46:37Z

Hi,

In KNITRO 5.2 manual, I am told to use knitroampl rather than knitro.

Now I can run it locally. Is there any different between local running
and Kestrel?

Regards,
Tower

On Tue, Jul 29, 2008 at 6:49 AM, Suleyman Demirel
<sdemirel@...> wrote:

> You may also try Kestrel, which allows Ampl to communicate with Neos Servers
> rather than you upload your file. This way, you can have multiple files.
>
> Follow the link below.
>
> http://www-neos.mcs.anl.gov/neos/kestrel.html
>
> And one more thing. Why don't you try option solver knitro rather than
> option solver knitroampl;?
>
> On Mon, Jul 28, 2008 at 11:20 PM, Robert Fourer <4er@...> wrote:
>>
>>
>> Are you using the web interface to AMPL/KNITRO (at
>> http://neos.mcs.anl.gov/neos/solvers/cp:KNITRO/AMPL.html)? Then you can
>> specify a model file and if desired a separate data file and commands
>> file.
>> Your commands file could be
>>
>> solve;
>> display x;
>> display obj;
>>
>> Do not include "reset;" or "option solver ...;" in the commands file,
>> though; NEOS takes care of these things.
>>
>> Bob Fourer
>> 4er@...
>>
>>
>> > -----Original Message-----
>> > From: ampl@... [mailto:ampl@...] On Behalf Of
>> > TowerGeeBulk@...
>> > Sent: Monday, July 28, 2008 9:48 AM
>> > To: AMPL Modeling Language
>> > Subject: [AMPL 1881] KNITRO AMPL syntax?
>> >
>> >
>> > Hi,
>> >
>> > I am testing KNITRO solver on NEOS remote server. Should I use the
>> > syntax of standard AMPL or the one like the example in the KNITRO
>> > package?
>> >
>> > Now my mod file is like. It can be used by other solver such as MINOS:
>> >
>> > var x {j in 0..15443};
>> > minimize obj: sum {j in 0..15443} x[j]*log(x[j]);
>> > subject to daxbnd0: 1e-13 <= x[0] <= 0.000033154300013;
>> > subject to daxbnd1: 1e-13 <= x[1] <= 0.000033154300013;
>> > subject to daxbnd2: 1e-13 <= x[2] <= 0.000033154300013;
>> > subject to daxbnd3: 1e-13 <= x[3] <= 0.000033154300013;
>> > ......
>> >
>> > The running command is like:
>> > reset;
>> > option solver knitroampl;
>> > solve;
>> > display x;
>> > display obj;
>> >
>> > But the output is no variable.
>> >
>> > Executing /home/neosotc/neos-5-solvers/knitro-ampl/knitro-driver.py
>> > File exists
>> > You are using the solver knitro.
>> > Executing AMPL.
>> > processing data.
>> > processing commands.
>> > Error:
>> > No variables declared.
>> > Error: amplin, line 47949 (offset 16623075):
>> > Error (512) in /home/neosotc/neos-bin/ampl -Rw amplin
>> > Error: x is not defined
>> > Error: context: display >>> x; <<<
>> >
>> > Could you please give me some help? Thanks.
>> >
>> > Tower
>> >
>>
>>
>>
>
>
>
> --
> ______________________________________________
> Suleyman Demirel - Office: +1 734 615 9503
>
> PhD Candidate in Operations Management
> Stephen M. Ross School of Business
> University of Michigan, Ann Arbor
> Web: http://www.umich.edu/~sdemirel
> ______________________________________________
>
> >
>

[AMPL 1890] Re: Variables’ value changed over course of model

2008-07-29T18:56:58Z

> My variable's value gets changed from one model to the next. The problem is
> there are 2 models where, let's say, 'P' is a 'var' in the first program
> that needs to be optimized. Afterthat, I'd like to use it as an input to the
> second model. I noticed that some values in 'P' got changed somehow in the
> second model. So, my current solution is to create a param P' and equate P
> and P' before taking P' as a fix input for the second model.
> Does anyone by chance experience this kind of problem before? Or know how to
> solve it? Also, there is a 'fix' command in AMPL but I don't think it can be
> applied to my case, please let me know if I'm wrong.

Try using named models. When you create a named model, you list the
members of the model (variables, constraints, objective). If you do
not list variable P as a member of the second model, I believe it will
be treated as fixed. Consider the following small example:

var P >= 0;
var Q >= 0;
minimize obj1: P;
maximize obj2: Q;
s.t. c1: P >= 2;
s.t. c2: Q <= 5*P;
model M1: P, obj1, c1; # defines the ingredients of model 1
model M2: Q, obj2, c2; # defines the ingredients of model 2
model M1; # switches to the first model
solve; # solves the first model (min P s.t. P >= 2); gets P = 2
model M2; # switches to the second model
solve; # solves max Q s.t. Q <= 10; gets Q = 10

Note that P is not treated as a variable in model M2 (if it were, M2
would be unbounded).

/Paul
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "AMPL Modeling Language" group.
To post to this group, send email to ampl@...
To unsubscribe from this group, send email to ampl+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ampl?hl=en
-~----------~----~----~----~------~----~------~--~---

[AMPL 1889] Variables’ value changed over course of model

2008-07-29T15:38:42Z

Hi,

My variable's value gets changed from one model to the next. The problem is there are 2 models where, let's say, 'P' is a 'var' in the first program that needs to be optimized. Afterthat, I'd like to use it as an input to the second model. I noticed that some values in 'P' got changed somehow in the second model. So, my current solution is to create a param P' and equate P and P' before taking P' as a fix input for the second model.

Does anyone by chance experience this kind of problem before? Or know how to solve it? Also, there is a 'fix' command in AMPL but I don't think it can be applied to my case, please let me know if I'm wrong.

Thanks,

-Peera-