On Friday, January 20, 2017 at 4:56:43 PM UTC, [email protected] wrote: > > Many thanks! > > Usually, I calculate some numbers of graph. Some are linear ones and some > are SDP ones. I'm more familiar with Python, so I want to still to use it > in sage as more as possible. And I have a prejudice that python could be > faster than sage. :P >
Huh, Sage is Python, what are you talking about? :-) Sage implementations are often faster than implementations written in Python, as they extensively use Cython and call C/C++-libraries. The only (very minor) overhead in Sage might come from certain internal machinery for types/classes, but in problems like solving LPs and SDPs this won't hurt performance. E.g. I bet there is no noticeable difference in performance if you call GLPK via MixedIntegerProgram or via cvxopt with solver='glpk' parameter. > > On Friday, January 20, 2017 at 3:45:42 PM UTC+1, Dima Pasechnik wrote: >> >> >> >> On Friday, January 20, 2017 at 2:32:12 PM UTC, Dima Pasechnik wrote: >>> >>> >>> >>> On Friday, January 20, 2017 at 12:31:40 PM UTC, [email protected] >>> wrote: >>>> >>>> Thanks, I'm going to install the newest version of Sage. and `dsdp` is >>>> supported also by default in the cvxopt of sage? >>>> >>> >>> no, dsdp is not supported by default. >>> >> >> In fact, we can include DSDP in sage with only a small effort. IIRC, some >> years ago DSDP did not have any >> license in its distro, which made it quite unclear as to who can use it >> and how. (Or I overlooked it) >> >> Do you need to solve semidefinite optimisation problems (SDP)? >>> What are the optimisation problems you are solving? >>> Sage does have a much better (mixed integer) linear programming interface >>> >>> http://doc.sagemath.org/html/en/thematic_tutorials/linear_programming.html#linear-programming >>> >>> http://doc.sagemath.org/html/en/reference/numerical/sage/numerical/mip.html >>> >>> than the one via cvxopt; it has many backends available, including >>> backends to commercial >>> solvers such as CPLEX etc. >>> >>> It also has a similar interface for SDP, see >>> >>> http://doc.sagemath.org/html/en/reference/numerical/sage/numerical/sdp.html >>> (although the only backend available ATM is cvxopt) >>> >> >> Sage also provides csdp as a package, although one cannot at present use >> it as the backend to the >> SDPs frontend. Someone has to work on this... >> >> >>> >>> >>>> >>>> On Friday, January 20, 2017 at 1:26:16 PM UTC+1, Dima Pasechnik wrote: >>>>> >>>>> There is something broken in your sage installation; for me the >>>>> following works: >>>>> >>>>> c=vector(RDF,[-4,-5]) >>>>> G=matrix(RDF,[[2,1],[1,2],[-1,0],[0,-1]]) >>>>> h=vector(RDF,[3,3,0,0]) >>>>> sol=linear_program(c,G,h,solver='glpk') >>>>> sol['x'] >>>>> >>>>> ouput is >>>>> >>>>> GLPK Simplex Optimizer, v4.60 >>>>> 4 rows, 2 columns, 6 non-zeros >>>>> * 0: obj = 0.000000000e+00 inf = 0.000e+00 (2) >>>>> * 2: obj = -9.000000000e+00 inf = 0.000e+00 (0) >>>>> OPTIMAL LP SOLUTION FOUND >>>>> (1.0, 1.0) >>>>> >>>>> In fact, BUILD_GLPK is set to 1 in the patch applied to cvxopt's >>>>> setup.py at the time sage's cvxopt package >>>>> is built, see build/pkgs/cvxopt/patches/setup.py.patch >>>>> >>>>> We will add a test in the docs, cf. >>>>> https://trac.sagemath.org/ticket/22217 that I just opened. >>>>> >>>>> >>>>> On Friday, January 20, 2017 at 12:11:37 PM UTC, Dima Pasechnik wrote: >>>>>> >>>>>> >>>>>> >>>>>> On Friday, January 20, 2017 at 11:44:46 AM UTC, [email protected] >>>>>> wrote: >>>>>>> >>>>>>> When I run a program it tells me that >>>>>>> `ValueError: invalid option (solver = 'glpk'): cvxopt.glpk is not >>>>>>> installed`. >>>>>>> >>>>>>> Then I try to install cvxopt with glpk from source. That is, change >>>>>>> BUILD_GLPK to 1 in the setup.py and then `python setup.py install` in >>>>>>> the >>>>>>> sage shell. But it gives me error that >>>>>>> `error: command 'gcc' failed with exit status 1`. >>>>>>> >>>>>> >>>>>> This actually should be fixed properly, in the sense that both glpk >>>>>> and cvxopt are standard Sage packages, and so nothing should prevent >>>>>> BUILD_GLPK=1 in cvxopt settings. >>>>>> >>>>>> In fact, I am under impression that it was working in the past, see >>>>>> (very old) >>>>>> https://trac.sagemath.org/ticket/9598 and >>>>>> https://trac.sagemath.org/ticket/6456 >>>>>> >>>>>> >>>>>> -- 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.
