PEST: "Cannot open error file "LF90.EER", error # 122"

[Giovanni Firmani]

has anybody experiences the error:

"Cannot open error file "LF90.EER", error # 122"

when running PEST with MODFLOW?

Thanks a lot for your help

Gio

Sent with MailTrack

[Richard Winston]

Lf90.err is the name of a file for Lahey Fortran 90. It contains error messages. If you had it, you would have gotten a more meaningful error message. 

--
You received this message because you are subscribed to the Google Groups "MODFLOW Users Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to modflow+u...@googlegroups.com.
To post to this group, send email to mod...@googlegroups.com.
Visit this group at https://groups.google.com/group/modflow.
For more options, visit https://groups.google.com/d/optout.

[Giovanni Firmani]

Thanks for your email.

I googled the error message and found this website:

http://support.esri.com/en/knowledgebase/techarticles/detail/12543

apparently the error means:  "122 | Unable to position file "

It is like the file Lf90.err cannot be created (?). No idea what it means.

I get the error after the 13th PEST optimisation. I tried to start pest with the best parameters I got so far and then it stopped again at the 5th optimisation.

I have never seen a problem like this before :(

Sent with MailTrack

--

Giovanni Firmani
Borgo Angelico, 22 - 00193 Roma (Italy)

Phone +39 06 98353792
Mobile +39 340 5341198

Skype gio.vanni.f

email g.fi...@gmail.com
webpage http://giovannifirmani.blogspot.com

[Giovanni Firmani]

Just to add more context to the problem, I get the error right after the following message on the screen:

"Broyden's update of Jacobian matrix undertaken."

Thanks again

Sent with MailTrack

[Hazar Bouzourra]

MODFLOW has been compiled using the” Lahey Fortran 90 extended memory compiler “.

 

The file "LF90.EER" must be located in the bin directory: “ modflow \bin “

 

If    “LF90.ERR” cannot be found, the error will only be identified by number

 

Cordially

Hazar BOUZOURRA

ENIT-LMHE

[Richard Winston]

Lf90.err is distributed to purchasers of Lahey Fortran 90. They can distribute it with programs compiled with Lahey Fortran 90.

If you are right about what that error number means, that suggests that Pest was trying to move to a location in a file before the beginning or after the end of a file. It might try to do that if it was trying to extract an observation from a file produced by a model run in which the model stopped prematurely.

[Giovanni Firmani]

Hi Richard,

thanks a lot for your answer. Your explanation seems reasonable. However, I doubt that it is the problem. According to the PEST manual:

".... when the files that PEST expects to read are not present, it will cease execution with an appropriate error message...."

I refuse to think that such a case was not considered by the PEST developer and the "appropriate error message" they refer to should be a Lahey Fortran message. It does not make any sense. How can an user understand the problem if he/she does not have the software installed?

In addition I checked the last model simulation run by PEST before the crash, and there was not anything wrong: it converged without any problem. I tried to run PEST in parallel mode too and got the same problem. As I used 24 CPUs I was able to check the last 24 simulations and they were OK. As I stated in a previous email the program stopped when calculating the "Broyden's update of Jacobian matrix undertaken." Find below the last lines of the PEST output:

       Lambda =  2.78256E-04 ----->

       running model .....

          Phi =   18.316      (  0.984 of starting phi)

    Meas. fn. =   10.556    

   Regul. fn. =   22295.    

       param "kpkp029" frozen: gradient and update vectors out of bounds

       param "kpkp067" frozen: gradient and update vectors out of bounds

       param "kpkp161" frozen: - update vector out of bounds

       Lambda =  3.59381E-05 ----->

       running model .....

          Phi =   18.318      (  0.985 of starting phi)

    Meas. fn. =   10.553    

   Regul. fn. =   22310.    

       param "kpkp608" frozen: gradient and update vectors out of bounds

       param "kpkp151" frozen: - update vector out of bounds

       Lambda =  4.64159E-06 ----->

       running model .....

          Phi =   18.319      (  0.985 of starting phi)

    Meas. fn. =   10.551    

   Regul. fn. =   22319.    

       Lambda =  5.99484E-07 ----->

       running model .....

          Phi =   18.321      (  0.985 of starting phi)

    Meas. fn. =   10.551    

   Regul. fn. =   22325.    

Cannot open error file "LF90.EER", error # 122

Whilst the last lines of the .rec file are:

       Lambda =  2.78256E-04 ----->

          Phi =   18.316      (  0.984 of starting phi)

    Meas. fn. =   10.556    

   Regul. fn. =   22295.    

       Broyden's update of Jacobian matrix undertaken.

       param "kpkp029" frozen: gradient and update vectors out of bounds

       param "kpkp067" frozen: gradient and update vectors out of bounds

       param "kpkp161" frozen: - update vector out of bounds

       Lambda =  3.59381E-05 ----->

          Phi =   18.318      (  0.985 of starting phi)

    Meas. fn. =   10.553    

   Regul. fn. =   22310.    

       Broyden's update of Jacobian matrix undertaken.

       param "kpkp608" frozen: gradient and update vectors out of bounds

       param "kpkp151" frozen: - update vector out of bounds

       Lambda =  4.64159E-06 ----->

          Phi =   18.319      (  0.985 of starting phi)

    Meas. fn. =   10.551    

   Regul. fn. =   22319.    

       Broyden's update of Jacobian matrix undertaken.

       Lambda =  5.99484E-07 ----->

          Phi =   18.321      (  0.985 of starting phi)

    Meas. fn. =   10.551    

   Regul. fn. =   22325.    

       Broyden's update of Jacobian matrix undertaken.

Thanks again

Gio

Sent with MailTrack

[Simon Woodward]

Which version of MODFLOW are you using?

I have run PEST quite a bit with Groundwater Vistas, and I know they use modified versions of MODFLOW (and other codes like MODPATH etc) that work better with PEST.

I don’t know the details however.

Are you running under Windows? I know that PEST expects certain kinds of EOL characters so if you have switched operating systems this could be a problem.

 

Simon

 

 

 

On 6 May 2016 at 12:11, Richard Winston <rbwi...@mindspring.com> wrote:

Lf90.err is the name of a file for Lahey Fortran 90. It contains error messages. If you had it, you would have gotten a more meaningful error message. 

 

-------- Original message --------

From: Giovanni Firmani

Date:05/05/2016 12:45 PM (GMT-05:00)

To: mod...@googlegroups.com

Subject: [MODFLOW] PEST: "Cannot open error file "LF90.EER", error # 122"

 

has anybody experiences the error:

 

"Cannot open error file "LF90.EER", error # 122"

 

when running PEST with MODFLOW?

 

Thanks a lot for your help

 

Gio

 

 

 

--

[Giovanni Firmani]

I managed to solve the problem. I found the lf90.err file in the original PEST package that I downloaded from the developer website. Then I re-run the simulation and got a more meaningful error message. The file that the program was not able to position was the SVD file. I was actually running the model with the SVD assist option enabled even though I did not use the power of the method. I discovered that when the SVD file becomes bigger than 2 Gb PEST crashes. I do not understand why. In the end I removed the SVD assist option and re-run the model and everything went OK.

thanks everyone for the support.

Gio

Sent with MailTrack

[Richard Winston]

SVD Assist must be compiled as a 32-bit program and so uses 32-bit integers to specify file positions.  The largest 32-bit integer doesn't allow for file sizes larger than 2 GB. You need a 64-bit version of the program. 

--