Page 1 of 1

Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 Passes

Posted: 23 Dec 2015, 15:29
by xiongyan21
Dear all
The sequential compilation of the just released LSDALTON1.0 on OS X EI Capitan10.11.2 has been successful, with Xcode 7.2, Cmake 3.3.2, mpich 3.1.4_1 and GCC 5.2.0 installed, and all the tests, using ctest -j4 and original parameters, can pass except the failure of the following
210 - dectests/decmp2_energy (Failed)
218 - dectests/fullcc2 (Failed)
228 - dectests/snoopmp2 (Failed)
229 - dectests/snoopmp2_ghost (Failed)
230 - dectests/snooprimp2f12 (Failed)
272 - dectests/decccsd_restart (Failed)
273 - dectests/decmp2_restart (Failed)
278 - dectests/rpa_rsh (Failed)
279 - dectests/decrpa_rsh (Failed)
280 - dectests/ccsd_exci_right (Failed)
281 - dectests/ccsd_exci_left (Failed)
287 - dectests/fullccsdpt_print (Failed)

There are many new, exciting and valuable tests added, from which I believe we all will benefit.

Congratulations!
US Oak Ridge Leadership Computing Facility (OLCF) has selected LSDALTON as the electronic structure application into its Center for Accelerated Application Readiness (CAAR) program.
Best Regards!

Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 Passe

Posted: 24 Dec 2015, 05:45
by xiongyan21
The error, e.g., in the case of test ccsd_exci_left for LiH, is perhaps caused by the related memory not clearly defined.
After I changed the keyword into .memory, the test has passed .

ccsd_exci_left
...

JAC CC excitation energies
JAC ******************************************************************************
JAC
JAC Exci. Hartree eV cm-1 ||T1|| (%)
JAC 1 0.1297317 3.530181 28472.81 93.63


...


Full CCSD calculation is done ! *
*****************************************************************************





**********************************************************
* CC ENERGY SUMMARY *
**********************************************************

E: Hartree-Fock energy : -7.6903387011
E: Correlation energy : -0.0178025193
E: Total CCSD energy : -7.7081412204
------------------------------------------------------
Total CPU time used in CC : 0.577611 s
Total Wall time used in CC : 0.599000 s
------------------------------------------------------


Hostname :
Job finished : Date: 24/12/2015 Time: 12:51:00



=============================================================================
-- end of CC program --
=============================================================================



>>> CPU Time used in *SCF is 0.89 seconds
>>> wall Time used in *SCF is 0.91 seconds

Total no. of matmuls used: 422
Total no. of Fock/KS matrix evaluations: 14
*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*
Memory statistics
*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*
Allocated memory (TOTAL): 0 byte - Should be zero - otherwise a leakage is present
...



Best Regards!

Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 Passe

Posted: 24 Dec 2015, 06:05
by xiongyan21
Dear all

I ran test decmp2_energy separately, which could pass, and got the following result
...
MP2 occupied correlation energy : -0.3373917322
MP2 virtual correlation energy : -0.3374894323
MP2 Lagrangian correlation energy : -0.3373881537



=============================================================================




**********************************************************
* DEC ENERGY SUMMARY *
**********************************************************

E: Hartree-Fock energy : -188.6610339040
E: Correlation energy : -0.3373917322
E: Total MP2 energy : -188.9984256362
E: DEC energy + skipped pairs : -188.9984256362

E: Estimated DEC err 1 : 0.0001012786
E: Estimated DEC err 2 : 0.0010000000
E: Estimated DEC err 3 : 0.7907125526


>>> CPU Time used in DEC FINAL is 0.03 seconds
>>> wall Time used in DEC FINAL is 0.03 seconds

CC Memory summary
-----------------
Allocated memory for array4 : 0.3853E-02 GB
Memory in use for array4 : 0.000 GB
Max memory in use for array4 : 0.4993E-03 GB

The estimated DEC err 3 energy is significantly different from that in the original output attached, and there are no tables for the timings of different levels matrix operations and integral routines.

Through this and the method used above all the tests in LSDALTON seem to pass.

Best Regards!

Re: Compilation of LSDALTON1.0 on OS EI Capitan 10.11.2 Pas

Posted: 24 Dec 2015, 23:03
by pett
Hi xiongyan21,

To your first question:
I completely agree that the keyword .USE_SYS_MEM_INFO is maybe not the best input for a test calculation because it is known to work only on linux systems where a /proc/meminfo exists, or mac os x systems if the top command returns an output of the expected format. The .MEMORY is a far more robust choice but it requires more from the user.

To the second question. Please ignore the estimated DEC err 3, we may remove it with an upcoming patch. It sneaked into the release but it is essentially an untested feature even though it has been around for some time.

Best regards

Re: Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 P

Posted: 10 Jan 2016, 09:33
by xiongyan21
xiongyan21 wrote:The error, e.g., in the case of test ccsd_exci_left for LiH using STO-3G, is perhaps caused by the related memory not clearly defined.
After I changed the keyword into .memory, the test has passed.

ccsd_exci_left
...

JAC ******************************************************************************
JAC CC excitation energies
JAC ******************************************************************************
JAC
JAC Exci. Hartree eV cm-1 ||T1|| (%)
JAC 1 0.1297317 3.530181 28472.81 93.63
...

Re: Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 Pa

Posted: 19 Mar 2016, 11:33
by xiongyan21
Dear all
The sequential compilation of LSDalton1.2 released on 15 Mar. has been successful on MAC OS EI Capitan 10.11.3 having Xcode 7.2.1, Cmake 3.3.2, mpich 3.1.4_1 and GCC 5.2.0 installed so that all the tests can pass if the previously stated methods are used.
Best Regards!

Re: Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 Pa

Posted: 09 Apr 2016, 07:38
by xiongyan21
It is the case for LSDalton1.2 released on 15 Mar. on MAC OS EI Capitan 10.11.4 having Xcode 7.3, Cmake 3.3.2, mpich 3.1.4_1 and GCC 5.2.0 installed

Re: Compilation of LSDALTON1.0 on OS X EI Capitan 10.11.2 Pa

Posted: 15 Jul 2016, 13:47
by xiongyan21
The sequential compilation of LSDalton1.2 released on 15 Mar. has been successful on MAC OS EI Capitan10.11.5 having Xcode7.3.1, Cmake3.3.2, mpich3.1.4_1 and patched GCC5.2 installed, but the following tests failed with the original parameters
210 - dectests/decmp2_energy (Failed)
218 - dectests/fullcc2 (Failed)
272 - dectests/decccsd_restart (Failed)
273 - dectests/decmp2_restart (Failed)
287 - dectests/fullccsdpt_print (Failed)