[Amir_line]

Hello, everyone!

I try to do a simulation with CFX and using a user Fortran subroutine in Linux machine. When I compile and lance CFX solver, I obtain this error .

my subroutine integral a function from 0 => +infinity. When I use integral from 0.2 => +infinity it works normally but when I integrate from 0 or 0.1 => +infinity it gives me this error. (I tested my integral in online compiler without CFX and it works normally)

Parallel run: Received message from slave

-----------------------------------------

Slave partition : 14

Slave routine : ErrAction

Master location : Message Handler

Message label : 001100279

Message follows below - :



+--------------------------------------------------------------------+

| ERROR #001100279 has occurred in subroutine ErrAction. |

| Message: |

| Signal caught: Segmentation violation |

| |

| |

| |

| |

| |

+--------------------------------------------------------------------+



Parallel run: Received message from slave

-----------------------------------------

Slave partition : 14

Slave routine : ErrAction

Master location : Message Handler

Message label : 001100279

Message follows below - :



+--------------------------------------------------------------------+

| ERROR #001100279 has occurred in subroutine ErrAction. |

| Message: |

| Stopped in routine FPX: SIG_HANDLER |

| |

| |

| |

| |

| |

+--------------------------------------------------------------------+



+--------------------------------------------------------------------+

| ERROR #001100279 has occurred in subroutine MESG_RETRIEVE. |

| Message: |

| Stopping the run due to error(s) reported above |

| |

| |

| |

| |

| |

+--------------------------------------------------------------------+



+--------------------------------------------------------------------+

| An error has occurred in cfx5solve: |

| |

| The ANSYS CFX solver exited with return code 1. No results file |

| has been created. |

+--------------------------------------------------------------------+

can anyone help me, please?

[ghorrocks]

It looks like your function is not working as you expect. You need to do some basic debugging on your routine, get it to print out values from the routine and check they are what you expect.

[Opaque]

From the output, it seems you are running in parallel, correct?

I will follow Glenn's suggestion of debugging your routine first in serial mode, and once it is working to your expectations try it in parallel mode, and repeat the debugging if needed.

[Gert-Jan]

Your routine seems to work fine if you integrate starting from 0.2, but not from 0.

Then everything looks fine, except for the function that you integrate. So the question is: how does your function look like?

[Amir_line]

Thanks Glenn for your reply!
I have checked that the values are good if I integrate from 0.2 => infinity (I compared them with a python code and with an online compiler Fortran).

[Amir_line]

Originally Posted by Opaque

From the output, it seems you are running in parallel, correct?

I will follow Glenn's suggestion of debugging your routine first in serial mode, and once it is working to your expectations try it in parallel mode, and repeat the debugging if needed.

Thanks Opaque!
I tested in serial mode but the same problem appears.

[Amir_line]

Originally Posted by Gert-Jan

Your routine seems to work fine if you integrate starting from 0.2, but not from 0.

Then everything looks fine, except for the function that you integrate. So the question is: how does your function look like?

Thanks Gert,

My function is Gaussian. See the picture below.

[Gert-Jan]

Nice picture, but it doesn't explain a lot.
What functions in CFX are present on your X and Y-axis?

[Amir_line]

Hello!
Thanks all, we find the problem, it was in the cluster (Linux machine). When we try the integral in windows machine, it works normally.

[Opaque]

Glad to hear you are getting results.

However, it is a very bad sign that it works on a Windows cluster and not on the Linux cluster.

From experience, it is a sign that something is wrong with the code and the different initialization paradigms between Windows and Linux can produce a false positive that the code is working. Just beware that something else is lurking in those simulations.