Re: [Help-glpk] SOLVED : problem with glp_ios_heur_sol

[Giorgio Sartor]

Just to clarify:

glp_simplex : LP solution

glp_iocp.presolve = ON : LP solution + preprocessing


- Gioker -

Il giorno 06/ago/2013, alle ore 14:55, Le Dinh Danh <address@hidden> ha scritto:

> OK, I find it clearer now, Thank you so much for your help man. Good work!
> 
> On Tue, Aug 6, 2013 at 2:35 PM, Giorgio Sartor <address@hidden> wrote:
>> The difference is the preprocessing.
>> If the problem has, for example, x variables (say 192), it is possible that 
>> after preprocessing (activated with presolve) the problem has less than x 
>> variables (say 186). glp_ios_heur_sol would need 186 but you give him 192 
>> variables. That's why you had those problems with initial solutions.
>> 
>> 
>> - G. S. -
>> 
>> Il giorno 06/ago/2013, alle ore 13:13, Le Dinh Danh <address@hidden> ha 
>> scritto:
>> 
>>> Oh great! It works now. Thank you so much G.S. I try your way and it
>>> is correct now.
>>> However, I really don't know why. I thought that the default presolve
>>> in GLPK using simplex method and it should work the same as we call
>>> simplex solver (with parm.presolve = OFF).
>>> It takes me a lot of time. Thank you once again and good work!
>>> 
>>> On Tue, Aug 6, 2013 at 12:42 PM, Giorgio Sartor <address@hidden> wrote:
>>>> Ok. I really think that the problem is the presolve because I had the same 
>>>> problem months ago.
>>>> At least just try adding glp_simplex before glp_intopt and disable 
>>>> presolve.
>>>> Presolve is not default in GLPK.
>>>> 
>>>> 
>>>> - G. S. -
>>>> 
>>>> Il giorno 06/ago/2013, alle ore 12:31, Le Dinh Danh <address@hidden> ha 
>>>> scritto:
>>>> 
>>>>> Oh I forgot to tell you that I suppose to use the default presolve in
>>>>> GLPK and do not explicit use any other for it. I know that if I don't
>>>>> use the presolve default I have to use some presolve like simplex or
>>>>> so. So in my case  parm.presolve = ON is neccesary.
>>>>> For my heuristic solutions I am sure that  they are correct, because I
>>>>> can verify it manually myself.
>>>>> 
>>>>> On Tue, Aug 6, 2013 at 12:23 PM, Giorgio Sartor <address@hidden> wrote:
>>>>>> Well, I don't know what you mean exactly but setting parm.presolve = ON 
>>>>>> is
>>>>>> never NECESSARY! (It is necessary only if you don't want to use 
>>>>>> glp_simplex
>>>>>> before glp_intopt). If you don't use it, my suggestion is to use 
>>>>>> glp_simplex
>>>>>> before glp_intopt and disable parm.presolve. In this case, if the integer
>>>>>> solution found is incorrect means that the solution you pass is 
>>>>>> incorrect.
>>>>>> 
>>>>>> 
>>>>>> 2013/8/6 Le Dinh Danh <address@hidden>
>>>>>>> 
>>>>>>> Hi G.S,
>>>>>>> It is neccesary to set parm.presolve=ON before solving MIP. In fact, 
>>>>>>> I've
>>>>>>> tried with parm.presolve=OFF but the results are always 0. Anyway, thank
>>>>>>> you.
>>>>>>> 
>>>>>>> 
>>>>>>> On Mon, Aug 5, 2013 at 5:34 PM, Giorgio Sartor <address@hidden> wrote:
>>>>>>>> 
>>>>>>>> Ok. Just try parm.presolve=OFF and let me know if then works.
>>>>>>>> 
>>>>>>>> 
>>>>>>>> - G. S. -
>>>>>>>> 
>>>>>>>> Il giorno 05/ago/2013, alle ore 16:51, Le Dinh Danh <address@hidden> ha
>>>>>>>> scritto:
>>>>>>>> 
>>>>>>>> Hi Giorgio Sartor,
>>>>>>>> 
>>>>>>>> On Mon, Aug 5, 2013 at 4:14 PM, Giorgio Sartor <address@hidden>
>>>>>>>> wrote:
>>>>>>>>> 
>>>>>>>>> Is glp_iocp.presolve activated?
>>>>>>>> 
>>>>>>>> Yes, I have already set parm.presolve = ON.
>>>>>>>>> 
>>>>>>>>> A common cause of incorrect solutions is that your solution is correct
>>>>>>>>> for the original problem but not correct for the current problem 
>>>>>>>>> (which
>>>>>>>>> could have been modified by GLPK during the solving process).
>>>>>>>>> Stupid question: does your solution start from position 1 of the 
>>>>>>>>> array?
>>>>>>>> 
>>>>>>>> Yes, of course.
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> - G. S. -
>>>>>>>>> 
>>>>>>>>> Il giorno 05/ago/2013, alle ore 14:58, Le Dinh Danh <address@hidden>
>>>>>>>>> ha scritto:
>>>>>>>>> 
>>>>>>>>> With "the result is still incorrect" I mean after calling the callback
>>>>>>>>> function, the solver find the solution but it is not correct one. I 
>>>>>>>>> don't
>>>>>>>>> know why. If I don't call the cacllback, the solve provide a good 
>>>>>>>>> solution.
>>>>>>>>> I can check it manually.
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> On Mon, Aug 5, 2013 at 1:19 PM, Giorgio Sartor <address@hidden>
>>>>>>>>> wrote:
>>>>>>>>>> 
>>>>>>>>>> What do you mean with "the result is still incorrect"?
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> - G. S. -
>>>>>>>>>> 
>>>>>>>>>> Il giorno 05/ago/2013, alle ore 12:57, Le Dinh Danh <address@hidden>
>>>>>>>>>> ha scritto:
>>>>>>>>>> 
>>>>>>>>>>> the result is still incorrect.
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> --
>>>>>>>>> Le Dinh Danh
>>>>>>>>> LIRMM, 161 rue Ada, 34095 Montpellier Cedex 5
>>>>>>>>> Email: address@hidden; address@hidden
>>>>>>>>> Tel: (+84) 912.449.666; (+33) 6.52.98.59.66
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> --
>>>>>>>> Le Dinh Danh
>>>>>>>> LIRMM, 161 rue Ada, 34095 Montpellier Cedex 5
>>>>>>>> Email: address@hidden; address@hidden
>>>>>>>> Tel: (+84) 912.449.666; (+33) 6.52.98.59.66
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> --
>>>>>>> Le Dinh Danh
>>>>>>> LIRMM, 161 rue Ada, 34095 Montpellier Cedex 5
>>>>>>> Email: address@hidden; address@hidden
>>>>>>> Tel: (+84) 912.449.666; (+33) 6.52.98.59.66
>>>>> 
>>>>> 
>>>>> 
>>>>> --
>>>>> Le Dinh Danh
>>>>> LIRMM, 161 rue Ada, 34095 Montpellier Cedex 5
>>>>> Email: address@hidden; address@hidden
>>>>> Tel: (+84) 912.449.666; (+33) 6.52.98.59.66
>>> 
>>> 
>>> 
>>> --
>>> Le Dinh Danh
>>> LIRMM, 161 rue Ada, 34095 Montpellier Cedex 5
>>> Email: address@hidden; address@hidden
>>> Tel: (+84) 912.449.666; (+33) 6.52.98.59.66
> 
> 
> 
> -- 
> Le Dinh Danh
> LIRMM, 161 rue Ada, 34095 Montpellier Cedex 5
> Email: address@hidden; address@hidden
> Tel: (+84) 912.449.666; (+33) 6.52.98.59.66