[Symphony] Issue: get No solution in Symphony

[Menal Guzelsoy]

Vladyslav,

I could not replicate the error. Can you try reading the problem from
these files and see if you are still getting the error? My guess is
that the constraint matrix that you create is not well defined (check
the fractional constants in MPS file) and the LP solver declares the
problem infeasible due to some tolerance issues.

Menal.

Menal Guzelsoy
1610 4620455



On Tue, Mar 29, 2011 at 6:19 AM, Vladyslav Kolbasin <marcell at insart.com> wrote:
>
> Here is problem in both formats. It is rather large, but on smaller
> problems i couldn't reproduce it.
>
> Also simplest solution i know in it:
> variables:
>  x0 = 250
>  x396 = 500
>  x813 = 1
>  x886 = 1
>  all others = 0
>
> Thanks a lot. Vladyslav
>
> On Mon, 28 Mar 2011 13:42:57 -0400, Menal Guzelsoy <megb at lehigh.edu>
> wrote:
>> Can you send us the problem in LP or MPS format?
>>
>> Menal Guzelsoy
>> 1610 4620455
>>
>>
>>
>> On Mon, Mar 28, 2011 at 1:11 PM, Vladyslav Kolbasin
>> <marcell at insart.com> wrote:
>>>
>>> No i didn't set common bounds (upper_bound & lower_bound or
>>> ..._bound_estimate)
>>>
>>> I tryed to set them (upper_bound & lower_bound) in wide range - nothing
>>> changed and i still have NO_SOLUTION
>>>
>>>
>>> Thanks a lot.
>>>
>>> On Mon, 28 Mar 2011 11:32:14 -0400, Ted Ralphs <tkralphs at Lehigh.EDU>
>>> wrote:
>>>> If you would be so kind as the post your inquiry to the mailing list,
>>>> we will try to get you an answer. Fomr the output, it looks as thought
>>>> you must be setting an a priori upper bound and the solver is not able
>>>> to imprve on that.
>>>>
>>>> Cheers,
>>>>
>>>> Ted
>>>>
>>>> On Mon, Mar 28, 2011 at 11:29 AM, Vladyslav Kolbasin
>>>> <marcell at insart.com> wrote:
>>>>>
>>>>>
>>>>> Good Day!
>>>>>
>>>>>   I'm using RSymphony (package that uses Symphony for R).
>>>>>   I'm running rather large LP problems : about 1000-3000 variables and
>>>>> 800-2000 constraints.
>>>>>
>>>>>   For some my test-cases i get return code 226 (NO_SOLUTION). But i know
>>>>> that there are simple solutions and may be rather complex too.
>>>>>
>>>>>   I set param verbosity to 140 and get such return:
>>>>>
>>>>>
>>>>>
>>>>> Skipping Preprocessor
>>>>> Problem has
>>>>>          834 constraints
>>>>>          1065 variables
>>>>>          4366 nonzero coefficients
>>>>>
>>>>> Solving...
>>>>>
>>>>> granularity set at 0.000001
>>>>>
>>>>> TM: tree size: 0 , 0
>>>>>
>>>>> ****************************************************
>>>>> * Now processing NODE 0 LEVEL 0 (from TM)
>>>>> ****************************************************
>>>>>
>>>>> Diving set to 2
>>>>>
>>>>>
>>>>>
>>>>> **** Starting iteration 1 ****
>>>>>
>>>>> solving root lp relaxation
>>>>> Terminating due to high cost -- fathoming node (no more cols to check)
>>>>>
>>>>>
>>>>> ****************************************************
>>>>> * Branch and Cut Finished                          *
>>>>> * Now displaying stats and best solution found...  *
>>>>> ****************************************************
>>>>> .....
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>   Also if i do some mixing constraints & variables then get solution - it is
>>>>> most strange issue.
>>>>>
>>>>>
>>>>>
>>>>>   Please Can you help me:
>>>>> - explain why i can't get solution
>>>>> - advise any params (or may be combination) that can help me to get any
>>>>> solution (already tryed some from "LP parameters" and some others)
>>>>>
>>>>> Thanks a lot.
>>>
>>> _______________________________________________
>>> Symphony mailing list
>>> Symphony at list.coin-or.org
>>> http://list.coin-or.org/mailman/listinfo/symphony
>>>
>