Discussion group for the ARIA software

[Brian Smith <Brian.Smith AT glasgow.ac.uk>]

This is a better fix:

To address this edit source/machvar.f and replace the subroutine with the following

C ======================================================
C SUBROUTINE SETFPEPS
C IMPLICIT NONE
C I/O
      INCLUDE 'numbers.inc'
      INCLUDE 'machvar.inc'
C
C Determine the machine epsilon i.e. the smallest FPEPS such that both
C 1+FPEPS and 1-FPEPS are different from 1.
C
C local
      DOUBLE PRECISION ONEDP
      DOUBLE COMPLEX DBCOMP
C
C external
C     NONE
C
C begin
C
C ONEDP doesn't really need to be defined - EPSILON() only needs to know
C the type
      _ONEDP_=(1.00)
      FPEPS = EPSILON(ONEDP)
      DBCOMP = DCMPLX(ZERO,ZERO)
      CALL DECLAR('FP_EPSILON', 'DP', ' ', DBCOMP, FPEPS)
      RETURN
      END

C========================================================

Dr. Brian O. Smith --------------------------- Brian Smith at glasgow ac uk
Institute of Molecular, Cell and Systems Biology & School of Life Sciences,
          College of Medical, Veterinary & Life Sciences,
  Joseph Black Building, University of Glasgow, Glasgow G12 8QQ, UK.
Tel: 0141 330 5167/6459/3089

https://www.gla.ac.uk/researchinstitutes/biology/research/structuralbiologybiophysicalcharacterisationfacility/

----------------------------------------------------------------------
The University of Glasgow, charity number SC004401

---

From: aria-discuss-request AT services.cnrs.fr <aria-discuss-request AT services.cnrs.fr> on behalf of Julien CAPPELE <julien.cappele AT univ-lorraine.fr>
Sent: 02 October 2020 09:45
To: aria-discuss AT services.cnrs.fr <aria-discuss AT services.cnrs.fr>
Subject: Re: [aria-discuss] Error during PSF-file generation

 

Hi everyone,

I have good news, I finally found the solution thanks to David L, in those two related forums on Haddock which also use CNS.

It happened that the problem was with gfortran, the compiler used in this exact case (so it seems)

Firstly, because the log said "Increase MXFPEPS value and compile again", I changed the value inside machvar.f (1024 to 2048). The first "machine epsilon" error was solved, but it raised another one which is "epsilon value is too low". SO, in this case I tried the other solution which is to hunt down the flag "-ffast-math" in the compiler "makefiles". I had to change it in multiple files of the cns_solve_1.21 directory.

After a few minutes of a quite intense hunting session (I used "grep -r ffast-math" to find all the files with it), I compiled again and now I can execute the cns executable without error.

I will finally be able to hear my poor laptop struggle during NMR structure calculations !

Thanks again everyone for your help !

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

Julien Cappèle

Doctorant - 2ème année - ED C2MP

Université de Lorraine

CRM² - UMR CNRS 7036

julien.cappele@univ-lorraine.fr

Tel: (+33)6 99 18 59 03

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

Le jeu. 1 oct. 2020 à 14:58, David Langelaan <dlangela AT dal.ca> a écrit :

Hello Julien,

 

I feel your pain and have struggled with these issues in the past. FYI I generally just launch aria from the terminal using ‘aria2 -g’ in the past launching aria from within ccpnmr failed, so I just because used to importing ccpnmr projects into the aria2 gui.

 

As for the epsilon issue, I have faced this myself and it is a bit odd. In short this is from the cns program crashing, and is not an issue with aria or ccpnmr. A quick way to test this is to launch the cns executable in the terminal, it will probably crash with the same error message. This error has been observed in the HADDOCK community as well.

 

This issue has something to with the machine.f file in the cns source code, and there are several options to fix it. One involves modifying the makefile/install files to remove a -fastmath compiler flag, the other involves modifying the machvar.f file to prevent the crash. I have used both in the past with success.

 

https://www.researchgate.net/post/Why_has_this_error_in_HADDOCK_occurredProtein-Protein_Docking_software_runningError_in_the_topology_generation_prot_psf_A_could_not_be_created

https://ask.bioexcel.eu/t/cns-errors-before-after-recompilation/54

 

I hope that this helps put you on the right track, good luck.

 

David

 

 

From: <aria-discuss-request AT services.cnrs.fr> on behalf of Julien CAPPELE <julien.cappele AT univ-lorraine.fr>
Reply-To: "aria-discuss AT services.cnrs.fr" <aria-discuss AT services.cnrs.fr>
Date: Thursday, October 1, 2020 at 6:47 AM
To: Benjamin Bardiaux <bardiaux AT pasteur.fr>
Cc: "aria-discuss AT services.cnrs.fr" <aria-discuss AT services.cnrs.fr>
Subject: Re: [aria-discuss] Error during PSF-file generation

 

CAUTION: The Sender of this email is not from within Dalhousie.

Hi Benjamin, thanks for your answer.

 

I forgot to mention that the error about ARIA Gui is specific to CcpNmr. I can launch it from the terminal (aria -g) but when I click on the "Launch Aria gui" in CcpNmr, it says Tix cannot be imported. I assume this error is more on the CcpNmr side. The check.py says it's all OK, especially :

"""

Tkinter and Tix modules...
Tkinter imported (Tk version 8.600)

Tix imported.

"""

 

On the actual generation problem:

I tried to generate the PSF-file myself just to see if it was related to rights or something, and it produced the exact same output which is :

 

"""

 %SETFPEPS increase value of MXFPEPS2 and recompile
 %SETFPEPS error encountered: Could not determine machine epsilon
   (CNS is in mode: SET ABORT=NORMal END)
 *****************************************************
 ABORT mode will terminate program execution.
 *****************************************************
 Program will stop immediately.
          ============================================================
           Maximum dynamic memory allocation:           0 bytes
           Maximum dynamic memory overhead:             0 bytes
           Program started at: \00\00\00\00\00\00\00\00 on \00\00\00\00\00\00\00\00\00\00\00
           Program stopped at: 13:25:55 on 01-Oct-2020
           CPU time used:       0.0017 seconds
          ============================================================

"""

 

I really do not know how to fix that epsilon thingy. I tried to google it, but I have no idea how to force CNS to use one.

The CPU in the laptop i'm using is a quad-core Intel i7-4710MQ

 

Thanks again,

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

Julien Cappèle

Doctorant - 2ème année - ED C2MP

Université de Lorraine

CRM² - UMR CNRS 7036

julien.cappele@univ-lorraine.fr

Tel: (+33)6 99 18 59 03

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

 

 

Le jeu. 1 oct. 2020 à 11:35, Benjamin Bardiaux <bardiaux AT pasteur.fr> a écrit :

Hi Julien,

On 01/10/2020 11:11, julien.cappele AT univ-lorraine.fr wrote:
> Good morning/afternoon everyone,
>
> I am a PhD student and I am currently trying to locally calculate some NMR
> structures with Aria, on a Ubuntu 20.04 LTS plateform.
> To be honest, I've been trying for a week now and I finally installed
> "correctly" all the softwares (Aria2.3, CNS1.2) and I am using CcpNmr 2.5.0.

What does the check.py script tells you about Tix being installed ?

>
> I'm doing the tutorial dedicated to structure calculation of CcpNmr.
> However, now that the project is finally set up, I get an error when i'm
> starting the calculation with Aria.
> Using this command: $ aria2 /path/to/my/project/project.xml, I get a strange
> error that I cannot understand.
>
> "IOError: USER ERROR <aria.cns.CNS> File "/some/path/to/aria2.3.2_05.08.2019/
> aria2.3/src/py/aria/cns.py", line 2038 in prepare
> An error occurered during creation of the PSF-file: "/some/path/to/Tutorials/
> CcpnCourse3a/run1/cns/begin/CcpnCourse3a.psf" could not be found."

check the run1/cns/begin/generate.out for ERROR, this should tell you more about the issue.

Best,

Benjamin

>
> Also, I cannot use the GUI interface with Aria because I get a strange error
> about Tix package being not installed and I didn't find any solution to it.
>
> Does someone have a solution to this problem ?
> Thanks a lot for your time.
>
> JC
>


--
---------------------------------------------
Dr Benjamin Bardiaux      bardiaux AT pasteur.fr
Unité de Bioinformatique Structurale
CNRS UMR3528 - Institut Pasteur
25,28 rue du Docteur Roux 75015 Paris, France
---------------------------------------------

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

Julien Cappèle

Doctorant - 2ème année - ED C2MP

Université de Lorraine

CRM² - UMR CNRS 7036

julien.cappele@univ-lorraine.fr

Tel: (+33)6 99 18 59 03

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

 

 

Le jeu. 1 oct. 2020 à 11:35, Benjamin Bardiaux <bardiaux AT pasteur.fr> a écrit :

Hi Julien,

On 01/10/2020 11:11, julien.cappele AT univ-lorraine.fr wrote:
> Good morning/afternoon everyone,
>
> I am a PhD student and I am currently trying to locally calculate some NMR
> structures with Aria, on a Ubuntu 20.04 LTS plateform.
> To be honest, I've been trying for a week now and I finally installed
> "correctly" all the softwares (Aria2.3, CNS1.2) and I am using CcpNmr 2.5.0.

What does the check.py script tells you about Tix being installed ?

>
> I'm doing the tutorial dedicated to structure calculation of CcpNmr.
> However, now that the project is finally set up, I get an error when i'm
> starting the calculation with Aria.
> Using this command: $ aria2 /path/to/my/project/project.xml, I get a strange
> error that I cannot understand.
>
> "IOError: USER ERROR <aria.cns.CNS> File "/some/path/to/aria2.3.2_05.08.2019/
> aria2.3/src/py/aria/cns.py", line 2038 in prepare
> An error occurered during creation of the PSF-file: "/some/path/to/Tutorials/
> CcpnCourse3a/run1/cns/begin/CcpnCourse3a.psf" could not be found."

check the run1/cns/begin/generate.out for ERROR, this should tell you more about the issue.

Best,

Benjamin

>
> Also, I cannot use the GUI interface with Aria because I get a strange error
> about Tix package being not installed and I didn't find any solution to it.
>
> Does someone have a solution to this problem ?
> Thanks a lot for your time.
>
> JC
>


--
---------------------------------------------
Dr Benjamin Bardiaux      bardiaux AT pasteur.fr
Unité de Bioinformatique Structurale
CNRS UMR3528 - Institut Pasteur
25,28 rue du Docteur Roux 75015 Paris, France
---------------------------------------------