On Wednesday, May 11, 2016 at 6:10:52 PM UTC+1, Sihuang Hu wrote: > > I think I downloaded the binaries. Thanks for your suggestions. I will try > it. >
well, you might try Sage 7.1 binary, perhaps it would work better... (or build from source, on such a machine it won't take much time, perhaps 3-4 hours) It would be good to get to the root cause of this, just to make sure that our way of building Sage binaries which should run on any CPU is correct. > > 在 2016年5月11日星期三 UTC+3下午6:14:37，Dima Pasechnik写道： >> >> Have you built Sage from source? If not, it could happen, in principle, >> that lapack/atlas/blas shipped >> do not quite work on your system... >> >> If you built it from source, did you do anything special to build Atlas >> in particular? >> >> You can also try building csdp completely separately from Sage, using >> Ubuntu system-supplied >> libraries, and see whether theta works. >> Namely, you can clone it from https://github.com/dimpase/csdp >> and build as described on that page. >> >> Sorry for not being able to help more, but as long we cannot reproduce >> the error, it's hard to tell what exactly is wrong. >> >> And of course you can use gdb to debug theta - for this you need to make >> sure it's compiled with right switches, of course... >> >> >> >> On Wednesday, May 11, 2016 at 2:50:30 PM UTC+1, Sihuang Hu wrote: >>> >>> Yes, it still works on many other graphs. >>> >>> --------------------------------------------------------------------------------------------------------------------------- >>> (sage-sh) shhu@shhu-ee-tau:~$ ldd `which theta` >>> linux-vdso.so.1 => (0x00007ffea69d6000) >>> libsdp.so.0 => /home/shhu/Documents/SageMath/local/lib/libsdp.so.0 >>> (0x00007f9e721e9000) >>> liblapack.so.3 => >>> /home/shhu/Documents/SageMath/local/lib/liblapack.so.3 (0x00007f9e71907000) >>> libcblas.so.3 => >>> /home/shhu/Documents/SageMath/local/lib/libcblas.so.3 (0x00007f9e716e3000) >>> libf77blas.so.3 => >>> /home/shhu/Documents/SageMath/local/lib/libf77blas.so.3 (0x00007f9e714c2000) >>> libatlas.so.3 => >>> /home/shhu/Documents/SageMath/local/lib/libatlas.so.3 (0x00007f9e70e91000) >>> libgfortran.so.3 => >>> /home/shhu/Documents/SageMath/local/lib/libgfortran.so.3 >>> (0x00007f9e70b73000) >>> libquadmath.so.0 => >>> /home/shhu/Documents/SageMath/local/lib/libquadmath.so.0 >>> (0x00007f9e70936000) >>> libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f9e70615000) >>> libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f9e7024f000) >>> libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 >>> (0x00007f9e70031000) >>> libgcc_s.so.1 => >>> /home/shhu/Documents/SageMath/local/lib/libgcc_s.so.1 (0x00007f9e6fe1b000) >>> /lib64/ld-linux-x86-64.so.2 (0x00005609d1eac000) >>> >>> ------------------------------------------------------------------------------------------------------------------------------ >>> The CSDP log is attached. >>> >>> Thanks for your kindly help! >>> >>> 在 2016年5月11日星期三 UTC+3下午4:43:04，Dima Pasechnik写道： >>>> >>>> It does work on some graphs, still, right? >>>> It's a high-end relatively new CPU, perhaps Atlas and/or gcc has a >>>> problem with it? >>>> >>>> Can you post the output of 'ldd theta', i.e. >>>> >>>> (sage-sh) dimpase@clpc171:sage$ ldd `which theta` >>>> linux-vdso.so.1 (0x00007ffd2c34f000) >>>> libsdp.so.0 => /home/scratch/dimpase/sage/sage/local/lib/libsdp.so.0 >>>> (0x00007f74f6ec3000) >>>> liblapack.so.3 => >>>> /home/scratch/dimpase/sage/sage/local/lib/liblapack.so.3 >>>> (0x00007f74f65fc000) >>>> libcblas.so.3 => >>>> /home/scratch/dimpase/sage/sage/local/lib/libcblas.so.3 >>>> (0x00007f74f63da000) >>>> libf77blas.so.3 => >>>> /home/scratch/dimpase/sage/sage/local/lib/libf77blas.so.3 >>>> (0x00007f74f61b8000) >>>> libatlas.so.3 => >>>> /home/scratch/dimpase/sage/sage/local/lib/libatlas.so.3 >>>> (0x00007f74f5b2a000) >>>> libgfortran.so.3 => /lib64/libgfortran.so.3 (0x00007f74f57ca000) >>>> libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007f74f558b000) >>>> libm.so.6 => /lib64/libm.so.6 (0x00007f74f5288000) >>>> libc.so.6 => /lib64/libc.so.6 (0x00007f74f4ec7000) >>>> libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f74f4caa000) >>>> libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f74f4a92000) >>>> /lib64/ld-linux-x86-64.so.2 (0x000055f7fadea000) >>>> >>>> as well as the CSDP installation log: >>>> >>>> SAGE_ROOT/logs/pkgs/csdp-6.2.log >>>> >>>> On Wednesday, May 11, 2016 at 12:58:43 PM UTC+1, Sihuang Hu wrote: >>>>> >>>>> I just did what you suggested, and it didn't work. I got those >>>>> messages: >>>>> >>>>> (sage-sh) shhu@shhu-ee-tau:~$ theta 2k2 >>>>> Segmentation fault (core dumped) >>>>> >>>>> My Cpu information is attached. >>>>> Ubuntu 14.04LTS >>>>> SageMath Version 7.0, Release Date: 2016-01-19 >>>>> >>>>> >>>>> 在 2016年5月10日星期二 UTC+3下午11:55:55，Dima Pasechnik写道： >>>>>> >>>>>> here is the input for theta executable: >>>>>> >>>>>> 4 >>>>>> 2 >>>>>> 1 2 >>>>>> 3 4 >>>>>> >>>>>> (also attached) >>>>>> save it to a file named, say, 2k2, fire up "sage -sh" >>>>>> and run >>>>>> >>>>>> theta 2k2 >>>>>> >>>>>> I get >>>>>> >>>>>> $ theta 2k2 >>>>>> Graph is of size 4 2 >>>>>> C block 1, blocksize, 4 >>>>>> Checking constraint 1 >>>>>> Checking constraint 2 >>>>>> Checking constraint 3 >>>>>> Iter: 0 Ap: 0.00e+00 Pobj: 1.0666667e+02 Ad: 0.00e+00 Dobj: >>>>>> 0.0000000e+00 >>>>>> [... edited] >>>>>> Iter: 15 Ap: 1.00e+00 Pobj: 2.0000000e+00 Ad: 9.00e-01 Dobj: >>>>>> 2.0000000e+00 >>>>>> Success: SDP solved >>>>>> Primal objective value: 2.0000000e+00 >>>>>> Dual objective value: 2.0000000e+00 >>>>>> Relative primal infeasibility: 1.11e-16 >>>>>> Relative dual infeasibility: 2.12e-10 >>>>>> Real Relative Gap: 8.56e-10 >>>>>> XZ Relative Gap: 9.63e-10 >>>>>> DIMACS error measures: 1.11e-16 0.00e+00 5.30e-10 0.00e+00 8.56e-10 >>>>>> 9.63e-10 >>>>>> The Lovasz Theta Number is 2.0000000e+00 >>>>>> >>>>>> Does this work for you? >>>>>> >>>>>> -- You received this message because you are subscribed to the Google Groups "sage-support" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/sage-support. For more options, visit https://groups.google.com/d/optout.
