[Coin-ipopt] LIWMA/LA problem?

Andreas Waechter andreasw at us.ibm.com
Mon Dec 6 20:48:50 EST 2004


Hi Kranthi,

If the program segfaults at that point, you probably gave invalid 
information about the structure of the Jacobian (AVAR or NNZA) to Ipopt. 
Please check that you populate AVAR and ACON and NNZA correctly.

As for LIWMA and LA, those are related to the work space that MA27 
requires for performing and storing the factorization of the iteration 
matrix.  Since the fill-in of the factorization is not known before it is 
attempted, we take a guess of how much it could be, reserve that memory, 
try the factorization, and if that turns out not to be enough, MA27 
complaints and we refactorize the matrix after increasing the size of the 
work space - this is when that message is printed.  If you really want to 
know more, please consult the documentation of MA27 (which is available at 
Harwell, where you also downloaded MA27).  Usually, seeing this message 
means that the linear system is ill-conditioned, but in your case that 
might be because some information about the Jacobian structure is 
incorrect (see the seg fault).

I hope this helps,

Andreas





Kranthi K Gade <kranthi at cs.nyu.edu>
Sent by: coin-ipopt-bounces at list.coin-or.org
12/04/2004 11:34 PM
 
        To:     coin-ipopt at list.coin-or.org
        cc: 
        Subject:        [Coin-ipopt] LIWMA/LA problem?


Hello all,

I have been trying to use IPOPT to optimize a problem with around 2000 
variables and 2 constraints.  The program gives seg fault on iteration 1.

Here's the output:

******************************************************************************
This program contains IPOPT, a program for large-scale nonlinear 
optimization.
    IPOPT is released as open source under the Common Public License 
(CPL).
    For more information visit www.coin-or.org/Ipopt
******************************************************************************

Number of variables           :     2105
    of which are fixed         :        0
Number of constraints         :        2
Number of lower bounds        :        2
Number of upper bounds        :        0
Number of nonzeros in Jacobian:     4210
Number of nonzeros in Hessian :     2105

ITER     ERR       MU      ||C||   ||YPY||  ||PZ||     ||D||   ALFA(V) 
ALFA(X)     NU   #LS       F      #cor   Regu    CPU(s)
0 .987E+05p .100E+00 .140E+06 .000E+00 .000E+00  .000E+00 .000E+00 
.000E+00  .000E+00  0 -.57231356E+03  0 .000E+00 .313E-01

ma27_call:  LIWMA increased from 73675 to 1307435
ma27_call:  LA increased from 63145 to 1307435

1 .134E+06p .100E+00 .138E+06 .000E+00 .000E+00  .621E+04 .976E+00d 
.100E+01h .000E+00  1 -.82994516E+03  2 .100E-03 .938E-01

Can somebody tell me what are these LIWMA and LA?  And is there a 
problem if they are increased?  Also is there a good source (FAQ list or 
something) to which I can refer to when faced with such problems?

I am attaching some debugging information just in case you might find it 
useful.

Thanks a lot.
kranthi
Program received signal SIGSEGV, Segmentation fault.
0x0043b113 in constr_order_a__ (nzorig=0x4fe8e0, norig=0x3d8340, 
m=0x3d8344, nind=0x4fe0f0, ivar=0x75edd4, nfix=0x4fe120, nz=0x4fe9b0, 
a=0x650020, acon=0x740048, avar=0x744210,
    nzc=0x4fe9a0, liw=0x22f240, iw=0x760eb8, ierr=0x22fee4) at 
constr.F:2310
2310          k = IW(p_ivar1+AVAR(ln2))
Current language:  auto; currently fortran
(gdb) bt
#0  0x0043b113 in constr_order_a__ (nzorig=0x4fe8e0, norig=0x3d8340, 
m=0x3d8344, nind=0x4fe0f0, ivar=0x75edd4, nfix=0x4fe120, nz=0x4fe9b0, 
a=0x650020, acon=0x740048, avar=0x744210,
    nzc=0x4fe9a0, liw=0x22f240, iw=0x760eb8, ierr=0x22fee4) at 
constr.F:2310
#1  0x0043a9b1 in constr_new_a__ (task=0x4f4a70, nzorig=0x4fe8e0, 
norig=0x3d8340, xorig=0x3d85f8, m=0x3d8344, nind=0x4fe0f0, ivar=0x75edd4, 
nfix=0x4fe120, ifix=0x740048, cscale=0x700f00,
    nz=0x4fe9b0, a=0x650020, acon=0x740048, avar=0x744210, nzc=0x4fe9a0, 
liw=0x22f428, iw=0x760eb8, ierr=0x22fee4, ev_f__=0x402d86 <EV_F>, 
ev_c__=0x402e30 <EV_C>, ev_g__=0x402ddb <EV_G>,
    ev_a__=0x402e8e <EV_A>, ev_h__=0x402f01 <EV_H>, ev_hlv__=0x402fa5 
<EV_HLV>, ev_hov__=0x402ff0 <EV_HOV>, ev_hcv__=0x403034 <EV_HCV>, dat=0x0, 
idat=0x3d8340) at constr.F:1818
#2  0x0043a443 in constr_ (task=0x4f06f8, iter=0x4f06f4, n=0x4fe0e0, 
nind=0x4fe0f0, m=0x3d8344, ivar=0x75edd4, nfix=0x4fe120, ifix=0x740048, 
norig=0x3d8340, xorig=0x3d85f8,
    cscale=0x700f00, vin=0x22f840, vout=0x700f10, ivec1=0x22f8a4, 
ivec2=0x22f8a4, lrs=0x4fe020, rs=0x650020, lis=0x4fe028, is=0x740048, 
lrw=0x22f6f8, rw=0x700f20, liw=0x22f6fc,
    iw=0x760eb8, ierr=0x22fee4, ev_f__=0x402d86 <EV_F>, ev_c__=0x402e30 
<EV_C>, ev_g__=0x402ddb <EV_G>, ev_a__=0x402e8e <EV_A>, ev_h__=0x402f01 
<EV_H>, ev_hlv__=0x402fa5 <EV_HLV>,
    ev_hov__=0x402ff0 <EV_HOV>, ev_hcv__=0x403034 <EV_HCV>, dat=0x0, 
idat=0x3d8340) at constr.F:1502
#3  0x00407cba in get_scale__ (iter=0x22ff64, norig=0x3d8340, 
xorig=0x3d85f8, n=0x4fe0e0, x=0x6fcd28, nind=0x4fe0f0, m=0x3d8344, 
ivar=0x75edd4, nfix=0x4fe120, ifix=0x740048,
    nlbo=0x3d8348, ilbo=0x3d83a0, bnds_lo__=0x3d83c0, nubo=0x3d834c, 
iubo=0xbaadf00d, bnds_uo__=0xbaadf00d, cscale=0x700f00, kconstr=0x4fe020, 
lrs=0x4fe040, rs=0x650020, lis=0x4fe050,
    is=0x740048, lrw=0x22fa0c, rw=0x700f10, liw=0x22fa14, iw=0x760eb8, 
ierr=0x22fee4, ev_f__=0x402d86 <EV_F>, ev_c__=0x402e30 <EV_C>, 
ev_g__=0x402ddb <EV_G>, ev_a__=0x402e8e <EV_A>,
    ev_h__=0x402f01 <EV_H>, ev_hlv__=0x402fa5 <EV_HLV>, ev_hov__=0x402ff0 
<EV_HOV>, ev_hcv__=0x403034 <EV_HCV>, dat=0x0, idat=0x3d8340) at 
get_scale.F:362
#4  0x004045d7 in ipoptmain_ (norig=0x3d8340, xorig=0x3d85f8, m=0x3d8344, 
nlbo=0x3d8348, ilbo=0x3d83a0, bnds_lo__=0x3d83c0, nubo=0x3d834c, 
iubo=0xbaadf00d, bnds_uo__=0xbaadf00d,
    v_lo__=0x3dc7d8, v_uo__=0x3dc800, lam=0x3dc820, c=0x3dc848, 
lrw=0x22fce8, rw=0x650020, liw=0x22fce4, iw=0x740048, iter=0x22ff64, 
ierr=0x22fee4, ev_f__=0x402d86 <EV_F>,
    ev_c__=0x402e30 <EV_C>, ev_g__=0x402ddb <EV_G>, ev_a__=0x402e8e 
<EV_A>, ev_h__=0x402f01 <EV_H>, ev_hlv__=0x402fa5 <EV_HLV>, 
ev_hov__=0x402ff0 <EV_HOV>, ev_hcv__=0x403034 <EV_HCV>,
    dat=0x0, idat=0x3d8340) at ipopt.F:632
#5  0x00403965 in ipopt_ (n=0x3d8340, x=0x3d85f8, m=0x3d8344, 
nlb=0x3d8348, ilb=0x3d83a0, bnds_l__=0x3d83c0, nub=0x3d834c, 
iub=0xbaadf00d, bnds_u__=0xbaadf00d, v_l__=0x3dc7d8,
    v_u__=0x3dc800, lam=0x3dc820, c=0x3dc848, lrw=0x22feec, rw=0x0, 
liw=0x22fef0, iw=0x0, iter=0x22ff64, ierr=0x22fee4, ev_f__=0x402d86 
<EV_F>, ev_c__=0x402e30 <EV_C>,
    ev_g__=0x402ddb <EV_G>, ev_a__=0x402e8e <EV_A>, ev_h__=0x402f01 
<EV_H>, ev_hlv__=0x402fa5 <EV_HLV>, ev_hov__=0x402ff0 <EV_HOV>, 
ev_hcv__=0x403034 <EV_HCV>, dat=0x0, idat=0x3d8340,
    nargs=0x3d837c, args=0x3d83e8, cargs=0x3d8450, __g77_length_cargs=40) 
at ipopt.F:285
Exception: STATUS_ACCESS_VIOLATION at eip=004321E9
eax=00000000 ebx=10288B60 ecx=10166C50 edx=00000000 esi=101209D0 
edi=00000000
ebp=0022E498 esp=0022E430 program=C:\cygwin\bin\gdb.exe, pid 3724, thread 
main
cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
Stack trace:
Frame     Function  Args
0022E498  004321E9  (10288B00, 101276E0, FFFFFFFF, 10131160)
0022E4B8  004322DE  (0022E670, 00000006, 0022E5F8, 00422B64)
0022E4C8  00422CA1  (10129DC8, 0022E630, 0022E5D4, 0042EFF9)
0022E5F8  00422B64  (00422C90, 10129DC8, 0022E630, 0022E628)
0022E638  00422CFA  (004322A0, 0022E670, 00431D7E, 00000006)
0022E698  00432AEB  (101276E0, 00000006, 00000000, 00000001)
0022E768  0043241A  (101276E0, 00000006, 00000000, 00000001)
0022E7A8  004338FB  (00000000, 00000000, 00000001, 101A13A0)
0022E7E8  00433B7E  (00000000, 00000001, 0022E818, 00407B6C)
0022E7F8  00405879  (10114EC8, 00000000, 00000001, 00000001)
0022E818  00407B6C  (10114EC8, 00000000, 00000001, 00000000)
0022E858  004230FB  (101002AA, 00000001, 0022E888, 005B8594)
0022E888  00484B4B  (101002A8, 10191938, 0022E8B0, 005C852C)
0022E8C8  00484FC5  (1019F220, 1019F220, 0022F0A4, 1019F220)
0022E8E8  005C6ED5  (0022E8A8, 101235E8, 0022E908, 004849AD)
0022E8F8  0048438B  (00000000, 0000000C, 0022E928, 0049D35E)
End of stack trace (more stack frames may be present)
Segmentation fault (core 
dumped)_______________________________________________
Coin-ipopt mailing list
Coin-ipopt at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-ipopt

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/ipopt/attachments/20041206/f6dcb64a/attachment.html 


More information about the Coin-ipopt mailing list