- Comp-phys-alps-users - lists.phys.ethz.ch

Lapack linking problem in ALPS 2 compilation on AIX
by jessica.alfonsi＠unipd.it 12 Oct '10

12 Oct '10

Dear Prof. Troyer, did you receive the log file of the make VERBOSE=ON related to the Lapack linking problem of ALPS 2 on AIX ? Do you have any solution for that ? Is there some compiler option which I could try ? Thank you for your help. Best Jessica

2 3

Installing the alps-application version 1.3.5
by Juan Felipe Carrasquilla 07 Oct '10

07 Oct '10

Dear Everyone. I am trying to install the 1.3.5 version of the applications and libraries on a laptop computer with Ubuntu 10.04 LTS. I think I successfully installed the alps libraries and now I am stuck with the applications part. I have downloaded the 5.1 version of lp_solve and compiled it using the a small script I found using sh ccc (ccc is the script). In the folder where I installed it I do ls and obtain: ~/ALPS/lp_solve_5.1/lpsolve51$ ls ccc cccLUSOL.osx cgcc.bat dll.sln liblpsolve51.a lib.sln lpsolve.rc resource.h ccc.hp-ux ccc.osx cvc6.bat dll.vcproj liblpsolve51.so lib.vcproj readme.txt Then I proceed to try to build the applications by first typing configure as follows: ./configure --with-lp_solve=/home/juan/ALPS/lp_solve_5.1/lpsolve51/ Then a big number of checks are performed, among which I see one which I think is related to my problem: checking for make_lp in -llpsolve51... yes checking for make_lp in -llpk5... no checking for make_lp in -llpk... no so llpsolve51 seems to be recognized. Is that OK? Then at the end I see a summary of the applications that are correctly configured: Summary of configuration: Diagonalization Full yes Sparse yes DMRG DMRG yes Particle in box yes Classical Monte Carlo Spin yes Quantum Monte Carlo Checksign yes Looper yes SSE-dirloop no Worms yes Quantum Wang-Landau yes The problem is that I would like to get the SSE-dirloop installed too. Do you people know what is wrong with my installation? Thanks in advance, Juan ---------------------------------------------------------------- SISSA Webmail https://webmail.sissa.it/ Powered by SquirrelMail http://www.squirrelmail.org/

3 3

Re: [ALPS-users] compilation under IBM AIX
by Matthias Troyer 06 Oct '10

06 Oct '10

On 4 Oct 2010, at 09:51, jessica.alfonsi(a)unipd.it wrote: > Dear Prof. Troyer and Alet, > we've made some progress with ALPS 2 beta 3 (Boost 1.43) rev. 4750 under > AIX with xlc 10 (we're not able to go on alps 2 beta 4 and boost 1.44 at > the moment). The patch suggested by Prof. Alet worked fine and we've > reached 61 % of compilation. Now compilation stops with an error on target > pyalea_c, as you can see in the attached files. Could you please provide a > fix for this ? Thank you again for your help Dear Jessica, As we are working towards the final release we do not want to support the old beta3 snapshot anymore if we can avoid it. What is the problem with Boost 1.43 and the beta 4? For the concrete problem in pyaela_c I would propose that you just turn off building of the Python tools in cmake, by setting ALPS_BUILD_PYTHON to OFF. This should not be a problem since all codes can be run without Python, and you can compile the Python evaluation tools on a personal workstation running MacOS, Windows, or Linux. Matthias

2 2

Re: [ALPS-users] Cannot compile alps-1.3.5 on cygwin-1.7.7
by Andriy Zhugayevych 01 Oct '10

01 Oct '10

Thank you, ALPS 2.0 for Windows can be easily installed and works fine except: 1) I set the environment variable ALPS_XML_DIR=C:\Sci\ALPS\lib\xml, but ALPS cannot find ALPS.xsl, lattices.xml, models.xml files, so that I need to copy these files into the working directory. In addition, 2) It would be great to have DLL-versions of ALPS applications for more efficient communication with other programs, e.g. Maple. Best regards, Andriy On 18 Sep 2010, at 08:39, Matthias Troyer wrote: > Hi Andriy, > > You need to use an older version of Boost to install ALPS 1.3.5. But why > don't you try ALPS 2.0b4 of which there is a simple installer available > for Windows? > > Matthias

2 1

compilation under IBM AIX
by jessica.alfonsi＠unipd.it 01 Oct '10

01 Oct '10

Hi all, I'm trying to compile at the moment without success both ALPS 1.3.5 and 2.0.0b4 under IBM SP Power 6. Here are the machine details: https://hpc.cineca.it/docs/guide2010/31IBMSP6UserGuide. The operating system is AIX 6.1, I suspect the reason could be the compiler version, since recent Boost releases have been tested with Visual Age 11: the machine has Visual Age C++ 10.1 (xlC -qversion gives IBM XL C/C++ for AIX, V10.1 Version: 10.01.0000.0008) and GNU g++ 4.2.0 (provided by IBM). Could you please confirm me if this can be an issue ? which compiler version is required for building both ALPS versions (but especially 2.0) on this machine ? which compilation options have to be used ? Please help me !! Jessica Alfonsi

3 6

mpi-implementation of alps-1.3.4, undefined reference to `MPI::Comm::Set_errhandler
by Michael Franz Krenn 28 Sep '10

28 Sep '10

Hey! I'm using alps-1.3.4 and compiled it (X86_64, mpic++.openmpi) successfully with openmpi-1.4.2 and want to use the mpi implementation for my SSE program. Compiling my program (mpi_phonon) I'll get following error: make mpi_phonon mpic++.openmpi -DNDEBUG -I/proj/manybody/pippan/ALPS/sources/boost_1_38_0 -I/proj/manybody/ALPS/lenny/include -I/proj/manybody/libs/openmpi-1.4.2/64bit/lib -o mpi_phonon mpi_phonon.o SSE.Directed.o SSE.Measurements.o SSE.o SSE.Update.o SSE.Initialization.o SSE.Greens.o SSE.PhononInterface.o PhononsInXL2D.o -L/proj/manybody/ALPS/64bit/mpi/lib -pg -lalps -lcomm-mpi -llapack -lf77blas -latlas -lpthread -lm -lmpi_cxx -lmpi -lopen-rte -lopen-pal -lfftw3 -llpk mpi_phonon.o:(.rodata._ZTVN3MPI3WinE[vtable for MPI::Win]+0x28): undefined reference to `MPI::Win::Set_errhandler(MPI::Errhandler const&)' mpi_phonon.o:(.rodata._ZTVN3MPI9IntercommE[vtable for MPI::Intercomm]+0x1a8): undefined reference to `MPI::Comm::Set_errhandler(MPI::Errhandler const&)' mpi_phonon.o:(.rodata._ZTVN3MPI9GraphcommE[vtable for MPI::Graphcomm]+0x1a8): undefined reference to `MPI::Comm::Set_errhandler(MPI::Errhandler const&)' mpi_phonon.o:(.rodata._ZTVN3MPI8CartcommE[vtable for MPI::Cartcomm]+0x1a8): undefined reference to `MPI::Comm::Set_errhandler(MPI::Errhandler const&)' mpi_phonon.o:(.rodata._ZTVN3MPI9IntracommE[vtable for MPI::Intracomm]+0x1a8): undefined reference to `MPI::Comm::Set_errhandler(MPI::Errhandler const&)' mpi_phonon.o:(.rodata._ZTVN3MPI4CommE[vtable for MPI::Comm]+0x1a8): undefined reference to `MPI::Comm::Set_errhandler(MPI::Errhandler const&)' collect2: ld returned 1 exit status make: *** [mpi_phonon] Fehler 1 Is there anyone who knows about this? Am I using the wrong mpi version? (Needs ALPS mpich?) thanks for comments!

3 4

Re: [ALPS-users] ALPS 2.0 vispython problem
by Matthias Troyer 28 Sep '10

28 Sep '10

On 23 Sep 2010, at 01:26, Andreas Nunnenkamp wrote: > Dear Prof Troyer, > > I have recently installed the ALPS package. To check that everything is working I am going through the tutorials. The command line and vistrails (for the most recent version the ALPS patch does not work, so I am using the last-to-recent one) seem to work fine. Vispython runs if I cut and paste the comments directly into the interpreter but if I try to run it from the command line I get the following error message: > > 31-34-185:ed-02-gaps Andreas$ vispython tutorial2a.py > Traceback (most recent call last): > File "tutorial2a.py", line 29, in <module> > import matplotlib.pyplot as plt > File "/Applications/Vistrails/Vistrails.app/Contents/Resources/lib/python2.6/matplotlib/pyplot.py", line 78, in <module> > new_figure_manager, draw_if_interactive, show = pylab_setup() > File "/Applications/Vistrails/Vistrails.app/Contents/Resources/lib/python2.6/matplotlib/backends/__init__.py", line 25, in pylab_setup > globals(),locals(),[backend_name]) > File "/Applications/Vistrails/Vistrails.app/Contents/Resources/lib/python2.6/matplotlib/backends/backend_tkagg.py", line 7, in <module> > import Tkinter as Tk, FileDialog > ImportError: No module named Tkinter > 31-34-185:ed-02-gaps Andreas$ > > I am running Python 2.6.5 (r265:79359, Mar 24 2010, 01:32:55) [GCC 4.0.1 (Apple Inc. build 5493)] on darwin. > Can you help and point me in the right direction? > > Many thanks, > Andreas Dear Andreas, vispython was never meant to be used like that, since Python is an interactive language and we typically use it as such, and not for just running canned Python scripts. The fix to your problem is easy. Just add the following at the beginning of the Python scripts if you run them on MacOS X: import matplotlib matplotlib.use('macosx') import matplotlib.pyplot ALternatively, use the latest nightly snapshot, which does that for you if pyalps is the first package that you include. Best regards Matthias

2 2

Jaynes-Cummings Hubbard model
by Andreas Nunnenkamp 25 Sep '10

25 Sep '10

Hi, I would like to use the ALPS library (possibly mostly ED and DMRG) to study the Jaynes-Cummings-Hubbard model. Has anyone already implemented this model? Below I include the most straightforward way I could think of and which produces the correct basis states but then I do not know how to use the <CONSTRAINT/> tag to pick out the subspace of constant total excitation number. Alternatively I tried using the excitation number (i.e. boson plus spin) and spin quantum numbers as well as excitation number and boson number. In both cases I was not able to get the correct basis states (either I ended up with negative boson number or too large a spin projection). Many thanks, Andreas ********************************************************* <MODELS> <SITEBASIS name="JCHM"> <PARAMETER name="Nmax" default="infinity"/> <QUANTUMNUMBER name="N" min="0" max="Nmax"/> <QUANTUMNUMBER name="Sz" min="-S" max="S"/> <OPERATOR name="Splus" matrixelement="sqrt(S*(S+1)-Sz*(Sz+1))"> <CHANGE quantumnumber="Sz" change="1"/> </OPERATOR> <OPERATOR name="Sminus" matrixelement="sqrt(S*(S+1)-Sz*(Sz-1))"> <CHANGE quantumnumber="Sz" change="-1"/> </OPERATOR> <OPERATOR name="Sz" matrixelement="Sz"/> <OPERATOR name="bdag" matrixelement="sqrt(N+1)"> <CHANGE quantumnumber="N" change="1"/> </OPERATOR> <OPERATOR name="b" matrixelement="sqrt(N)"> <CHANGE quantumnumber="N" change="-1"/> </OPERATOR> <OPERATOR name="n" matrixelement="N"/> <OPERATOR name="x" matrixelement="N+Sz+1/2"/> </SITEBASIS> <BASIS name="JCHM"> <SITEBASIS ref="JCHM"/> </BASIS> <HAMILTONIAN name="JCHM"> <PARAMETER name="Delta" default="0"/> <PARAMETER name="g" default="1"/> <PARAMETER name="J" default="1"/> <BASIS ref="JCHM"/> <SITETERM site="i"> Delta*n(i) - g*(Sminus(i)*bdag(i) + Splus(i)*b(i)) </SITETERM> <BONDTERM source="i" target="j"> -J*(bdag(i)*b(j)+bdag(j)*b(i)) </BONDTERM> </HAMILTONIAN> </MODELS> *****************************************************************

1 0

calculation of local density of states
by Jun Wen 18 Sep '10

18 Sep '10

Dear all, Is it possible to calculate local density of states within the framework of exact diagonalization in ALPS? Thanks a lot! Best -- Jun Wen, Graduate Research Assistant Office: RLM 7.312 Department of Physics 1 University Station C1600 The University of Texas at Austin TX, 78712 Email: jwen(a)physics.utexas.edu junwen.ut(a)gmail.com

3 2

Cannot compile alps-1.3.5 on cygwin-1.7.7
by Andriy Zhugayevych 18 Sep '10

18 Sep '10

I followed the installation instructions for Cygwin but could not compile ALPS-1.3.5 libraries. The error was: make[1]: Entering directory `/sci/alps/tool' g++ -DHAVE_CONFIG_H -I../src -I../src -DNDEBUG -I/usr/src/boost_1_43_0 -g -O2 -MT convert2xml.o -MD -MP -MF .deps/convert2xml.Tpo -c -o convert2xml.o convert2xml.C In file included from /usr/src/boost_1_43_0/boost/integer.hpp:18, ... other 18 files ... from convert2xml.C:34: /usr/src/boost_1_43_0/boost/integer_fwd.hpp:98: error: expected identifier before 'unsigned' /usr/src/boost_1_43_0/boost/integer_fwd.hpp:98: error: multiple types in one declaration In file included from /usr/src/boost_1_43_0/boost/function/function_base.hpp:21, ... other 17 files ... from convert2xml.C:34: /usr/src/boost_1_43_0/boost/integer.hpp:127: error: expected identifier before 'unsigned' /usr/src/boost_1_43_0/boost/integer.hpp:127: error: expected unqualified-id before ':' token make[1]: *** [convert2xml.o] Error 1 make[1]: Leaving directory `/sci/alps/tool' I have Cygwin 1.7.7 with gcc-g++ 3.4.4-999 , boost 1.43.0-1, libxslt 1.1.26-2, LAPACK 3.2.2-1, and other required packages. The results of running configure was as follows: cd /sci/alps ./configure --with-boost=/usr/src/boost_1_43_0 Summary of configuration: ALPS version 1.3.5 prefix /home/admin/ALPS compiler type gnu [g++] optimization yes exceptions yes warnings no Boost version svn MPI support no OpenMP support no HDF5 support no LAPACK support no SQLite support no XML parser native Documentation no Kind regards, Andriy Zhugayevych

2 1