Opened 7 years ago

Closed 6 years ago

#189 closed defect - convergence (fixed)

limited chianti (fe ions) cause blr_n09_p22_Z20 to thrash

Reported by: Gary J. Ferland Owned by: nobody
Priority: major Milestone: database infrastructure
Component: atomic/molecular data base Version: trunk
Keywords: Cc:


this sim has had problems using the CloudyChiantiFe?.ini masterlist for some time now. tests show that disabling ots rates makes it run ok.

are chianti sims being done/not done? is number of levels fluctuating?

Change History (3)

comment:1 Changed 7 years ago by matt

The masterlist CloudyChiantiFe?.ini is now the default (CloudyChianti?.ini) and the convergence problem still exists. The test suite completes successfully with chianti turned on except for this sim (blr_n09_p22_Z20). The sim reaches the default failure limit of 20 before completing. To investigate this further we ran the sim with Fe XV through Fe XXI chianti data, using only one species at a time in the masterlist file. There were convergence failures on each run. Running the sim with chianti turned off gives 2 convergence failures. The fastest run was achieved using a chianti masterlist containing only Fe XVI.

comment:2 Changed 7 years ago by matt

The following is from a ticket created in September 2010. The ticket claims convergence problems running the blr_n09_p22_Z20 sim when all of Chianti is turned on (not just Fe which is currently the default). The specific details included below have not been rechecked since September 2010 and they are being included here so that all reports of this problem are in the same ticket.

testing with all of chianti turned on reveals only one abort, the sim blr_n09_p22_Z20. the symptoms are mysterious getting to zone 16, where the first convergence failure is announced, takes several hours. the code is clearly thrashing. typical ionization at this point is roughly Fe 21 +/-2 there is a thermal front in this model, so I checked whether too large a dr near that front could be a cause. The save dr output was not produced. there were only the first dr choices - the zone dr was never reported. so, "save dr" does not work in this test. the convergence failure is not due to a thermal front - constant temperature also fails the save convergence file shows that the code is struggling to converge the heating and cooling starting early in the calculation. by zone 10 it is having problems. debugging will be faster if earlier zones are examined. The sim now completes without a convergence failure when constant temperature is set. The sim also completes without a convergence failure if "no line ots" is added to the input file.

comment:3 Changed 6 years ago by Ryan Porter

Resolution: fixed
Status: newclosed

This sim has been running fine with CloudyChianti?.ini as the default for months. Closing.

Note: See TracTickets for help on using tickets.