For the following file SDPA does not find a solution and exits with phase.value = pdINF cvxpy finds the solution OK the latest sdpa code is run as: sdpa sdpaTest.dat out.dat >diag.txt see s.zip with the results when called differently it also prints mu thetaP thetaD objP objD alphaP alphaD beta 0 1.0e+04 1.0e+00 1.0e+00 -0.00e+00 +6.17e+05 9.1e-01 2.7e-01 2.74e+00 pdINF criteria :: line 1194 in sdpa_parts.cpp 1 2.7e+04 8.9e-02 7.3e-01 +4.71e+03 +1.24e+06 9.1e-01 2.7e-01 2.74e+00
For the following file SDPA does not find a solution and exits with phase.value = pdINF cvxpy finds the solution OK the latest sdpa code is run as: sdpa sdpaTest.dat out.dat >diag.txt see s.zip with the results
Have you tried increasing the value of lambdaStar in the parameter file? If you choose this parameter too small, then SDPA will start with a very small initial step size and runs into the problem that you describe.
When I try to compile SDPA-QD with GCC 11 I get the error "use of an operand of type ‘bool’ in ‘operator++’ is forbidden in C++17". This is very easy to fix, one just needs to replace the line LP_CNonZeroCount[blockNumber[l2]+j]++; in the file sdpa_io.cpp with LP_CNonZeroCount[blockNumber[l2]+j] = true; The same problem affects SDPA-GMP, and I assume all other versions of SDPA.
I have the same problem. Have you solved it now?
Hello! I think I am having some problems with installation of sdpa. I downloaded sdpa-7.3.8 and compiled it via ./configure and make, which seemed to succeed. But then I tried to run sdpa A.dat A.result where A.dat is the example downloaded from http://plato.asu.edu/ftp/sdpa_format.txt . In this example the following 4x4 matrix should be SDP: x1-1 0 0 0 ; 0 x1+x2-2 0 0 ; 0 0 5*x2-3 2*x2 ; 0 0 2*x2 6*x2-4 The output (copied below) does not make any sense to me. In particular, the solution xVec = {-4.414e+00,+6.121e+00}...
Dear Professor Yamashita, Dear all, Recently I tried to install SDPARA following the usual instructions. I have Open MPI 2.1.1 isntalled, with ScaLAPACK and BLACS, and I compiled MUMPS in the distribution version. When I'm trying to do "make", it gives me error messages of sorts: mumps_part9.F:(.text+0x18cb7): undefined reference to `mpi_allreduce_' ./mumps/build/lib/libmumps_common.a(mumps_part9.o): In function `mumps_645_': Could you please provide any advice on how to complete the linking process?...
Dear developers, I generated a .dat-s file from Ncpol2sdpa in python, and tried to solve it in SDPA, but no matter I used SDPA in Python or in Command line, the result from SDPA is a total blank. I checked the Task Manager in Windows while running sdpa.exe in command line. It normally takes lower than 50% of my CPU and Memories. But after few hours, the sdpa.exe didn't work anymore which was shown at the Task Manager. Thanks!
Dear Rui Li, When I compiled SDPA to build an Windows binary, I used mingw64 on a Linux machine. On Windows, I guess cygwin or msys2 are available to compile SDPA. Thanks, Makoto Yamashita
Yeah. You're right. I also spent a half year to learn the documentation, technical points, and installation. I think updated documentation is necessary.
Hi all, Does anyone have experience on running the C++ SDPA code on a Window operating system? I transform from the Ubuntu system to a Window system. However, I encountered more than 100 errors when I was running the original code. Thank you. Regards, Ricky
Thanks a lot for the fix. We appreciate your support.
Dear Makoto, A fellow student has managed to fix the problem - by changing the timings.h file in spooles/build so that //static struct timezone TZ ; is commented out and gettimeofday(&TV, TZ) ; \ is replaced by gettimeofday(&TV, NULL) ; \ the MakeFile for spooles also needs to be editited, so that the files are not extracted from archive each time. - commenting out the line rm -rf build ; mkdir build ; cd build ; tar xvfz ../archive/spooles.2.2.tgz does this. this is because in the current version...
Thank you very much for your interst on SDPA. Now, timezone is defined in, for example, /usr/include/x86_64-linux-gnu/sys/time.h. Could you try #include </usr> ? Best Regards, Makoto Yamashita
Hello there, I am looking to use the QD variant of SDPA, and have been following the relevant installation instructions. I can install the QD library no problem, but when I try to make SDPA-qd I get the following error many times: ../../Utilities/MM.h:47:12: warning: format '%d' expects argument of type 'int', but argument 3 has type 'long unsigned int' [-Wformat=] "\n ALLOCATE error : bytes %d, line %d, file %s", \ ^ basics.c:27:1: note: in expansion of macro 'ALLOCATE' ALLOCATE(etree, struct _ETree,...
I found the SDPA-C solver usually fails to get the optimal solution due to some numerical issues while other solves like SeDumi can solve it efficiently. As the attached two data file. How could I fix it? SDPA-C performs Cholesky factorization to solve the Schur complement system, and this step is not easy to be parallelized. So, I am wondering is it possible to lever some iterative methods to replace Cholesky factorization to develop a parallel one. SDPA-C do not perform the Mehrotra predictor-corrector...
Dear all, I am writing regarding some trouble in prapreing the data file for the SDPA-C solver. Since SDPA-C embedded the sparse clique factorization into the PDIP method, my trobule is whether I need to obtain the Chordal graph and maximal cliques of the problem before putting the data file problem into the solver. If I do this, how the SDPA-C figure out this since the data file has on infromation on the clique facotirzation. If not, does the solver do this inside? Another question is the sparse-clique...
I have received an answer by mail which did not get picked up by SourceForge. It turned out that my Python 2 build was failing because I was missing the SPOOLES library. Python 3 support for SDPA-P has not been implemented so far.
Hello, when I run python2 setupy.py install, compilation fails with the following error: building 'sdpap.spcolo.spcoloext' extension […] gcc -pthread -DNDEBUG -march=x86-64 -mtune=generic -O2 -pipe -fstack-protector-strong -fno-plt -march=x86-64 -mtune=generic -O2 -pipe -fstack-protector --param=ssp-buffer-size=4 -D_FORTIFY_SOURCE=2 -fPIC -Isdpap/spcolo/cmodule -I/usr/include/spooles/ -I/usr/include/python2.7 -c sdpap/spcolo/cmodule/spcolo_ordering.cpp -o build/temp.linux-x86_64-2.7/sdpap/spcolo/cmodule/spcolo_ordering.o...
Hi all, I just install the SDPA-C solver on ubuntu. However, I encountered a mistake when I execute the 'make all' command. Here is the screen output. Could you tell me how can I figure out this issue? Thank yo so much. (cd /home/ricky/sdpa-c; \ rm -rf OpenBLAS; \ git clone git://github.com/xianyi/OpenBLAS; \ mv -f OpenBLAS /home/ricky/sdpa-c/OpenBLAS; \ cd /home/ricky/sdpa-c/OpenBLAS; \ make NO_WARMUP=1 USE_THREAD=1 USE_OPENMP=1 FC=gfortran CC=gcc libs netlib) Cloning into 'OpenBLAS'... remote:...
Hi all, I use the SDPA and Sedumi to solver a large sparse SDP. I found the Sedumi can solve the problem while SDPA told me that 'the step length is too small'. I am wondering is it possible for me to make SDPA to solve the problem by changing this condition? Best, Rui
Hi all, I am using the SDPA-C to solve a SDP problem on Liunx, however, I failed to install the SDPA-C, due to the targeted address: http://www.cise.ufl.edu/research/sparse/SuiteSparse/SuiteSparse-4.0.2.tar.gz is wrong. I am wondering how can I change this address. Thank you so much. Rui
Hi all, I am using the SDPA-C to solve a SDP problem on Liunx, however, I failed to install the SDPA-C, due to the targeted address: http://www.cise.ufl.edu/research/sparse/SuiteSparse/SuiteSparse-4.0.2.tar.gz is wrong. I am wondering how can I change this address. Thank you so much. Rui
Hi, Yikui, Have you solved the problem? I also want to use SDPA on windows, but I failed it. Best, Rui
Is it possible for me to install SDPA-C and SDPARA-C on Windows operating system?
Did SDPAM7.39 include the SDPA-C 7.38?
hi, i have a standard Semi-Definite Program and I dont Know how can I find F1,f2,... https://ibb.co/dUfH2v thanks
hi, i have a standard Semi-Definite Program and I dont Know how can I find F1,f2,... thanks
Dear SDPA developers, I am trying to write a Julia SDPA wrapper (see here: https://github.com/blegat/SDPA.jl)....
I have found a workaround, I replace the line with: cd build; echo "OPTF = " '-I$$(topdir)/libseq'...
I have found a workaround, I replace the line with: cd build; echo "OPTF = " '-I$$(topdir)/libseq'...
I have found a workaround, I replace the line with: cd build; echo "OPTF = " -I$(topdir)/libseq...
I have spotted the issue. The problem comes from line 35 of mumps/Makefile. You do...
I see that in mumps/src/Makefile there is only -I../include and not -I../libseq....
I just had an idea. Here (http://beige.ucs.indiana.edu/I590/node100.html), it seems...
I just had an idea. Here (http://beige.ucs.indiana.edu/I590/node100.html), it seems...
When I execute one of the examples, I get the error "ERROR in MPI_ALLREDUCE, DATATYPE=...
Hello SDPA team, I am working on solving 4 real quadratic equations in 4 real unknowns,...
Dear SDPA developers, I am trying to write a Julia SDPA wrapper (see here: https://github.com/blegat/SDPA.jl)....
Dear SDPA developpers: First of all, thank you for this nice software.I am trying...
Hi, I downloaded the sdpa/sdpa-7.3.9-windows.zip. There is an application in it and...
Hi, I am trying to precisely follow the instruction given for installing SDPA-M on...
Dear SDPA developers, First of all, thank you for this nice software, and I really...
I have a similar problem both in linux/unix and windows environment: everything crashes...
Hello, Everytime I write a problem using SDPAP and call the SDPAP.solve(A,b,c,K,J)...
Dear All; I am a little bit confused regarding setting the input parameters of the...
Dear All; I am a little bit confused regarding setting the input parameters of the...
Dear, I have just read the config.log file generated after the ./configure command....
Dear, I am trying to use the MATLAB interface for SDPA. To do so, I have performed...
When I applied SDPA-GMP to hinf10 and hinf11, problems from SDPLIB, the procedures...
I am attempting to install SDPA-C and then SDPARA-C. Everything goes fine installing...
Thank you. I found my error in the blockstructure.
The LP parts are handled as diagonal matrices. So, if p-th line of the input file...
Hi, I am just trying to solve my first Problem with sdpa. I get the error message:...
Thank you.
Thanks a lot.
Dear Giraffem, When SDPA and SDPA-M are installed by apt-get, the example files are...
Dear Giraffem, As you wrote, could you expand the matrix by the larger matrix? SDPA...
Can SDPA handle complex numbers or do I have to rewrite everything as A <=> (real(A)...
I installed sdpa and sdpam via $ sudo apt-get install sdpa $ sudo apt-get install...
I was not sure, whether to create a new thread or reply to this one. Matlab is crashing...
Dear Prof. Yamashita, I really appreciate your advice. I am sorry that you faced...
Dear Jehandad, Thank you for being interested in this topic. Does it crash on the...
Dear Jehandad, Thank you very much for using SDPA. To change the output accuracy...
SDPA employs the primal-dual interior-point method. So, it outputs both the primal...
When I repeat SDPA more than 500 times, I also encounter the same situation. From...
Dear Prof Yamashita, I am using Yalmip to solve a SDP. While sdpa for matlab is able...
A naive question perhaps, but can come tell me how to solve the dual form of a problem...
Does it crash on the first iteration ?
Dear all, I have a problem on executing SDPA-M on Windows. When I execute the following...
Thanks !
MEX completed successfully. Thank you for your help.
I think that the interface of the ScaLAPACK from netlib assumes 32 bit integers....
Thanks a lot, indeed i linked it against SCALAPACK ( from netlib ) which linked and...
Thank you very much for using SDPA-M. For the first error message, as you write,...
I am sorry that the available version of SDPARA can not link with 64bit scalapack...
Dear all, I have problems compiling SDPA-M. I am working under CentOS 6.5 First,...
Hi All, I have compiled SDPARA using intel mkl's flavor of BLAS + LAPACK + BLACS...
Dear Prof Yamashita, Indeed, that does solve the problem. Thanks a lot for the helpful...
Dear Prof Yamashita, Indeed such high accuracies are not required in my research....
To pass new compilers, could you modify the error line from===== printf("invalid...
Dear Jehandad, I have tried your data, and SDPA have returned a result like this....
I am trying to compile SDPA-GMP using gcc version 4.7.2 on Fedora 17. GMP 4.3.2 compiles...
Impressive !
Dear Prof Yamashita, I tried to change the parameter as you had directed, but the...
Dear Prof Yamashita, Thanks a lot for the kind reply. I will try and tweak the parameter...
Dear Jehandad, I am sorry that SDPA does not switch the Cholesky factorization to...
Dear Prof Yamashita, I am using SDPA to solve for a Lyapunov stable matrix which...
Hi, I've downloaded sdpa-gmp and I manage to get it to work on small examples but...
Hi All, First off thanks for the great software, it has worked really well for me...
SDPA-M for 64bit Windows MATLAB is now available