Ticket #33 (closed defect - convergence: fixed)

Opened 1 year ago

Last modified 5 months ago

another H2 advection bug - convergence failure in hmole_step

Reported by: will Assigned to: nobody
Priority: critical Milestone: C08.02
Component: H chemistry (deprecated) Version:
Keywords: dynamics Cc: gary@pa.uky.edu, rjrwilliams@googlemail.com

Description (Last modified by will)

Using the Rauch stellar atmosphere has triggered a new bug :)

This occurs after a few hours, on the start of the 4th iteration. Another model triggers the same problem on the same iteration. Full input and output files attached. Correction: full output file fell foul of the size limit on attachments. 

         Start Iteration Number 4


PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.17355551e+03 n(H) 3.16227783e+03 (old-new)/old -3.566e-03 nH-old 4.413e-04 nH-new -1.128e+01
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16147200e+03 n(H) 3.16227783e+03 (old-new)/old 2.547e-04 nH-old 4.413e-04 nH-new 8.058e-01
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16387637e+03 n(H) 3.16227783e+03 (old-new)/old -5.056e-04 nH-old 4.413e-04 nH-new -1.599e+00
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16053119e+03 n(H) 3.16227783e+03 (old-new)/old 5.522e-04 nH-old 4.413e-04 nH-new 1.747e+00
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16225384e+03 n(H) 3.16227783e+03 (old-new)/old 7.448e-06 nH-old 4.413e-04 nH-new 2.399e-02
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16230392e+03 n(H) 3.16227783e+03 (old-new)/old -8.389e-06 nH-old 4.413e-04 nH-new -2.609e-02
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16221050e+03 n(H) 3.16227783e+03 (old-new)/old 2.115e-05 nH-old 4.413e-04 nH-new 6.733e-02
PROBLEM hmoleee zn 0 proton_sum_old 3.16227739e+03, proton_sum_new 3.16228371e+03 n(H) 3.16227783e+03 (old-new)/old -1.998e-06 nH-old 4.413e-04 nH-new -5.876e-03
 PROBLEM  ConvFail 1, Hmole population not converged iteration 4 zone 0 fnzone 0.00
PROBLEM h1fnd 0 0 0 0 0
 hmole_step: dgetrs finds singular or ill-conditioned matrix
[Stop in hmole]
Command exited with non-zero status 1
12392.36user 5.64system 3:26:42elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+112263minor)pagefaults 0swaps

Attachments

flow10_med_n3bigA.in (1.4 kB) - added by will on 06/22/07 10:37:53.
flow10_med_n3bigA_newmole.in (1.3 kB) - added by rjrw on 12/16/07 07:20:14.
Modify input to use revised newmole "no molecules" command

Change History

06/22/07 10:37:53 changed by will

  • attachment flow10_med_n3bigA.in added.

06/22/07 10:44:28 changed by will

  • description changed.

12/01/07 12:22:50 changed by rjrw

  • milestone set to Bring newmole onto the trunk.

12/10/07 08:05:15 changed by rjrw

It now fails more quickly on the trunk: 

 PROBLEM DISASTER An assert has been thrown, this is bad.
 It happened in the file rt_line_one_tauinc.cpp at line number 51
 This is iteration 1, nzone 45, fzone 45.11, lgSearch=F.

which directs you to the following... 

/* sanity check - total fine opacity should be greater than single ine opacity */
ASSERT( fabs(dTau_line) < 0.01 || 
	rfield.fine_opac_zone[t->Emis->ipFine]*1.1 >= opacity*t->Emis->PopOpc  );

12/16/07 07:17:39 changed by rjrw

For head of newmole (with "no co chemistry molecules" => "no heavy molecules" in deck: 

 PROBLEM DISASTER An assert has been thrown, this is bad.
 It happened at mole_h2.cpp:1657
 This is iteration 1, nzone 90, fzone 90.13, lgSearch=F.

-- the ortho and para densities of H2 don't add up.

12/16/07 07:20:14 changed by rjrw

  • attachment flow10_med_n3bigA_newmole.in added.

Modify input to use revised newmole "no molecules" command

12/18/07 16:20:45 changed by rjrw

  • status changed from new to closed.
  • resolution set to fixed.

Now appears to work at the head of the newmole branch at r1695.