[Mesa-users] using FreeEOS_builder

[Ebraheem Farag]

Thanks for giving it a try Rob,

I corrected my Cmake library path.
Similar to you, I also had to manually set the fortran flag to legacy and
compiler path to mesasdk, though I failed to mention this initially.

-EbF

On Tue, May 9, 2023 at 7:39 AM Rob Farmer <robert.j.farmer37 at gmail.com>
wrote:

> Hi Eb,
> I don't have a answer for your problem (as i find the same result) but a
> few things I noticed while trying:
>
> The sdk has moved the lapack libraries since those instructions where
> written so you should do:
>
> export
>  CMAKE_LIBRARY_PATH=$MESASDK_ROOT/lib:$MESASDK_ROOT/math-slots/crmath/lib
>
> (you can see this in your cmake.out file that it found xcode version
> instead of the sdk's)
>
> I also needed to do
> export FC=$MESASDK_ROOT/bin/gfortran
>
> to get the SDK compiler instead of the system one, though yours seems to
> have picked up the sdk compiler. I also needed
>
> export FFLAGS="-std=legacy"
>
> set before calling cmake.
>
> Rob
>
>
> On Fri, 5 May 2023 at 22:26, Ebraheem Farag via Mesa-users <
> mesa-users at lists.mesastar.org> wrote:
>
>> Hello all,
>>
>> I'm wondering if anyone has used the FreeEOS builder package in MESA
>> recently to build new tables or run the test case.
>>
>> From $MESA_DIR/eos/eosFreeEOS_builder. I have version 2.2.1 downloaded
>> and installed following https://cococubed.com/mesa_market/free_eos.html
>> <https://urldefense.com/v3/__https://cococubed.com/mesa_market/free_eos.html__;!!IKRxdwAv5BmarQ!fuqGlgaFexlG-uCROjbVTjI3Mbda1lsVAue6hbZlMxbYeny3RU2Ht1KSQAg_PEyxvIxNIv7ZW4BG_oMSGKm9DwPO1A$>
>> .
>>
>> I'm in MESA r15140 for this, but it's essentially identical to r22.11.1,
>> which I have also tested with identical results. I checked the mesa github
>> to see what changed between versions, and it only seems to be related to
>> the location of the FreeEOS package.
>> ------------------
>> The primary issue I am running into is with the test case, where I get an
>> error "STOP bfgs_linesearch: internal bracketing logic error" before the
>>  './free_eos_table   0.02   0.70   mesa-FreeEOS_02z70x.data' is able to
>> write that last few table lines.
>>
>> I've attached the build and install log, and a text file of test case
>> output ./free_eos_table 0.02 0.70 mesa-FreeEOS_02z70x.data
>> <https://urldefense.com/v3/__https://drive.google.com/file/d/1dmaVTlvS9JtzkcaT3eezpLzxJwAsgxT6/view?usp=gmail__;!!IKRxdwAv5BmarQ!fuqGlgaFexlG-uCROjbVTjI3Mbda1lsVAue6hbZlMxbYeny3RU2Ht1KSQAg_PEyxvIxNIv7ZW4BG_oMSGKmyggJyNQ$>
>> .
>>
>> I know this error is coming from line 715 in 'free_eos-2.2.1/src/bfgs.f',
>> but I have little idea what is happening:
>>
>>      *do* *while*(need345.or.*.not.*acceptable)
>>
>>         *if*(not_need345) *then*
>>
>>           *if*(debug) *then*
>>
>>             *write*(0,*) 'ai, fai, fprimeai = ', ai, fai, fprimeai
>>
>>             *write*(0,*) 'bi, fbi = ', bi, fbi
>>
>>             *write*(0,*) 'fai.eq.fmin', fai*.eq.*fmin
>>
>>             *write*(0,*) 'fai.le.f0+ai*rho*fprime0',
>>
>>      &        fai*.le.*f0+ai*rho*fprime0
>>
>>             *write*(0,*) '(bi-ai)*fprimeai.lt.0.d0.and.'//
>>
>>      &        'abs(fprimeai).gt.-sigma*fprime0 = ',
>>
>>      &        (bi-ai)*fprimeai*.lt.*0.d0.and.
>>
>>      &        abs(fprimeai)*.gt.*-sigma*fprime0
>>
>>             *write*(0,*) 'fbi.gt.f0+bi*rho*fprime0.or.fbi.ge.fai = ',
>>
>>      &        fbi*.gt.*f0+bi*rho*fprime0.or.fbi*.ge.*fai
>>
>>           *endif*
>>
>> C          test assertion given by PMOO, 2.6.3
>>
>>           *if*(*.not.*(
>>
>>      &      (fai*.eq.*fmin)*.and.*
>>
>>      &      (fai*.le.*f0+ai*rho*fprime0)*.and.*
>>
>>      &      ((bi-ai)*fprimeai*.lt.*0.d0)*.and.*
>>
>>      &      (abs(fprimeai)*.gt.*-sigma*fprime0)*.and.*
>>
>>      &      (fbi*.gt.*f0+bi*rho*fprime0.or.fbi*.ge.*fai)))
>>
>>      &      *stop* 'bfgs_linesearch: internal bracketing logic error'
>>
>> p.s. I'm on Mac OS 12.6, but I get the same result on Ubuntu 20LTS. I
>> rebuilt and ran in multiple MESASDKs on Mac os and Ubuntu from r15140 to
>> date, to ensure I get the same result regardless. (The build files look
>> identical between each and the test case fails at the same location.)
>>
>> Any help/advice would be much appreciated, thank you!
>>
>> -EbF
>> _______________________________________________
>> mesa-users at lists.mesastar.org
>> https://lists.mesastar.org/mailman/listinfo/mesa-users
>> <https://urldefense.com/v3/__https://lists.mesastar.org/mailman/listinfo/mesa-users__;!!IKRxdwAv5BmarQ!fuqGlgaFexlG-uCROjbVTjI3Mbda1lsVAue6hbZlMxbYeny3RU2Ht1KSQAg_PEyxvIxNIv7ZW4BG_oMSGKmwNZaouQ$>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20230509/074428d9/attachment.htm>