[Coin-ipopt] Segmentation Fault (Quasi Newton option)

damien at khubla.com damien at khubla.com
Wed May 24 17:33:26 EDT 2006


I've had this on Linux and on Windows.  The pointer is pointing to junked
memory.  Can you repeat it consistently?  I can get it sometimes by
returning false from one of the evaluation calls but it's not consistent
enough for me to give instructions to duplicate it.

Damien

> Hi,
>
> I am working with the 3.1.0 version of IPOpt and am using the Quasi
> Newton option to approximate the Hessian.
> After several (100) iterations IPOpt gives a segmentation fault. The
> call stack is -
>
> =======================================================
> Program received signal SIGSEGV, Segmentation fault.
> 0x0815df07 in Ipopt::SmartPtr<Ipopt::LowRankUpdateSymMatrixSpace
> const>::operator-> ()
> (gdb) where
> #0  0x0815df07 in Ipopt::SmartPtr<Ipopt::LowRankUpdateSymMatrixSpace
> const>::operator-> ()
> #1  0x0815f72a in Ipopt::LowRankAugSystemSolver::Solve ()
> #2  0x081064bf in Ipopt::AugRestoSystemSolver::Solve ()
> #3  0x08152e40 in Ipopt::LeastSquareMultipliers::CalculateMultipliers ()
> #4  0x081494d0 in Ipopt::IpoptAlgorithm::AcceptTrialPoint ()
> #5  0x081455cb in Ipopt::IpoptAlgorithm::Optimize ()
> #6  0x081a6964 in Ipopt::MinC_1NrmRestorationPhase::PerformRestoration ()
> #7  0x0810f8a5 in Ipopt::BacktrackingLineSearch::FindAcceptableTrialPoint
> ()
> #8  0x08147515 in Ipopt::IpoptAlgorithm::ComputeAcceptableTrialPoint ()
> #9  0x08145533 in Ipopt::IpoptAlgorithm::Optimize ()
> #10 0x080890bf in Ipopt::IpoptApplication::call_optimize ()
> #11 0x08087d84 in Ipopt::IpoptApplication::OptimizeNLP ()
> #12 0x08087434 in Ipopt::IpoptApplication::OptimizeTNLP ()
> =========================================================
>
> Someone else faced the same problem about a month back. Was wondering
> if someone knows the reason or has a solution for this problem.
>
> Thanks,
> ~ Sonia
>
> _______________________________________________
> Coin-ipopt mailing list
> Coin-ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-ipopt
>





More information about the Coin-ipopt mailing list