Originally posted by: BabakArh

Hello

I catch this error "Access violation" i attached the error. this is a bit weird becaz

1- for one instance I get it sometimes and it doesn't appear sometime.

2-it appear when I add the 4th column in column generation method

3- before adding OPLALL i added more than 10000 columns and there was no sign of this error

4- if I break the error I could continue to add another columns

Is it possible that you are close to the memory limits of your machine? Do you have a Java heap dump or a dump from the JVM that includes a backtrace for the native code?

Originally posted by: BabakArh

hello thank you for replying

there is something weird about this error.

1- my free memory is about 3 gig

2-i could add another columns in this problem after i click on break and continue, i can add more than 100 column if i continue

3-it happens after OPLALL

What exactly do you mean by "it happens after OPLALL"? What is OPLALL?

Whether you can do anything useful with your application after a memory access violation is more or less random.

Originally posted by: BabakArh

thank u for replying

before OPLALL i used CPLEX reference, and this problem never occurred. after I start to use constraint programing in column generation I forced to use oplall.dll instead of cplex.dll

It is not random, I break the error and I continue it all the time. it is like this:

1-I start to solve, and i get error in e.g 4th column

2-i break the error and continue solving

3-i add more than e.g 100 column, again I catch that error.

4-break the error and continue to solve

If I had the access violation error in 4th column shouldn't I have that error in all column I add after that error?

this is the weirdest error Ive faced it.

plz if u had any idea it would help me a lot. i appreciate if u share it.

Can you configure your JVM to dump a backtrace of the native code in case of the crash and post that here?

Also, do you have a minimal example that you can post here and that reproduces the error so that we can check here?
 

Originally posted by: Babak ARH

I develop a new column generation model (a brief simple model )

I used OPLALL.dll i get this error in 16763th column. you can run it then wait for catching error.

I also used the same model with Cplex.dll, and I didnt catch the error even after 36000th column

So far I have not managed to reproduce your problem here. It seems like you are using a very old version of CPLEX (12.4). Can you upgrade to a more recent version?

Originally posted by: BabakArh

thank you for your consideration

I have some reason which I cant accept your answer basde on

1- with OPLALL.dll i get this error but with ilog.CPLEX.dll i won't,

2- there is something wierd about this error it randomly happen. first run in 200 the second run in 1300.

it might be useful for you:

I get this exception in column 405 (column name=Y_n_s405), when I export the model;

in one constraint (constraint name=day_cross57_4  ,,,, line=4337 ) this column appeared 2 times and at the end of the model, in variable condition part it happened 2 time (line 4795 and 4796).

[[    i should mention constraint name=day_cross57_4 means i=57 j=4  and this column has an object i=57  ]]

but in the other constraint (constraint name=pat_fnish57  ,,,, line=1245) this column should be appeared with coefficient 12 but it didn't. furthermore in the OBJECTIVE this column should be appeared by cost 300 but it didnt either.

I will attach the model, thank you for your time. I look forward to hearing your answer.

Originally posted by: BoJensen

Daniel could not reproduce this issue, is it possible to upgrade to our newest version ? If there is a problem it might already be fixed.

I ran your problem with the interactive commandline :

<log snippet>

Parallel mode: deterministic, using up to 8 threads for concurrent optimization.
Tried aggregator 1 time.
LP Presolve eliminated 2661 rows and 3204 columns.
All rows and columns eliminated.

Notice your problem is solved in presolve i.e CPLEX can with simple logic solve your problem, without starting one of the optimizers. This may be completely OK, but you should check your model makes sense, often this is an indication the model is not what you expect.

Originally posted by: BabakArh

hello and thank you for your consideration.

I think because of my English I couldn't explain the problem well.

consider one problem which you coded in c#. you have 2 codes for that problem. the first code is based on OPLALL.dll

the scond one is based on and Cplex.dll and concert.dll.

(as you know OPLALL.dll aggregate some library like CP.dll,Cplex.dll and Concert ...)

if I run the first code I face the error, but in second, I don't.

the last thing which throw the exception is this function in concert tech :

concert_wrapPINVOKE.new_IloNumVar__SWIG_8(IntPtr jarg1, Double jarg2, Double jarg3, Int32 jarg4, String jarg5)

I dont know what is it but I know the concert in OPLALL throw this exception.

You may think concert tech in OPLALL is exactly as same as the separate concert tech. I assure you they r different. because some of their syntax are different. for example in the oplall you have Column which is IColumn in separate concert.

Originally posted by: BoJensen

The OPL concert version and the CPLEX concert version may act differently. The libraries are created with different internal methods.

Could you please upgrade to the newest version or confirm it's not possible for you to upgrade ?

Originally posted by: BabakArh

hi egain

I have the same problem with 12.6

Sorry, I still cannot reproduce your problem here. I have no idea what else to try.

My best guess still is that you are running out of memory and that is not handled well, but you said that you have 3GB free memory when the crash happens (is all that memory actually available to your process?).

Originally posted by: BabakArh

hello.

I dont know how much memory is available for the process but i have enough memory.

if there is a way to assign memory for the process or for the cplex, i can try.

thank u for your concern, 

Yes, the operating system usually has a way to limit the amount of memory available to a process. On Linux this is for example done with the ulimit command. For Windows I'm sorry, I really don't know. I google a little and it seems a good way to emulate this might be the Application Verifier. See for example this answer on stackoverflow: http://stackoverflow.com/a/1289945.

Running your application through application verifier may be a good idea in any case since it may reveal errors/problems that are otherwise hidden.

Originally posted by: BabakArh

thank you for considering my problem. I didnt get my answer at all and i tried everything, and thank you for your comments and ideas.

it was a good experience to share my problem here.

I am sorry that we could not figure out what caused the crash in your application. if the problem re-appears with a different model instance or in a different setup then don't hesitate to come back here. In a different setup it may be easier to figure out what is going wrong without actually sitting in front of your machine.