Problems with continuous melting along a user-defined PT path

[jakemoore]

Paula,

I think this one is a doozy.  I'm trying to run a MORB composition along a user-defined PT-path with continuous melting and I keep getting minimization errors.  All the run files are at the bottom and the PT-path file is attached. 

Here's the end of the output at failure.  Doesn't seem to be special output beyond the failure of the minimization algorithm:
Adiabat_1ph at: P 35500.000000, T 1586.326000
liquid: 0.525 g 50.46 1.96 14.57 0.58 0.09 10.17 4.87 11.13 6.18 0.00 0.00
Activity of H2O = 0  Melt fraction = 0.00738244
garnet: 36.013265 g, composition (Ca0.27Fe''0.25Mg0.48)3Al2Si3O12
clinopyroxene: 34.592084 g, composition cpx Na0.23Ca0.65Fe''0.12Mg0.53Fe'''0.06Ti0.02Al0.67Si1.72O6
spinel: 0.038230 g, composition Fe''0.52Mg0.50Fe'''0.10Al1.14Cr0.73Ti0.01O4
...-->Linear search: Iteration exceeded - aborting.
...-->Linear search: Iteration exceeded - aborting.
...-->Linear search: Iteration exceeded - aborting.
...-->Linear search: Iteration exceeded - aborting.
...-->Linear search: Iteration exceeded - aborting.
THE QUADRATIC MINIMIZATION ALGORITHM HAS FAILED TO CONVERGE.
...Quadratic convergence failure. Aborting.
Failure in silmin
Not all calculations performed!

The error occurs at 35400 bars.  This is about where the user-defined PT-path would diverge from the PT-profile of the MORB composition as it undergoes adiabatic melting, starting at 40 kbar and 1600 degC.  I though the problem stemmed from the fact that the algorithm can't handle the combination of P, T, and composition (and resulting S, etc.).  However, I can start it at the P and T where the system dies and run it for a while until it dies again.  Although, it look like the mineral modes and chemistries are different when I restart it.

If the PT-path file were constructed at larger/smaller P steps, might that make a difference?  I know running this MORB at all is outside the calibration range, but I'd really like to get this to work.

If more explanation is necessary, just ask!

Thanks,
Jacob


Setting the environment variables:
ADIABAT_VERSION pMELTS
ADIABAT_MODE PTpath
ADIABAT_MAXP 40000
ADIABAT_MINP 25000
ADIABAT_CRASH_FILE adiabat.out
ADIABAT_CELSIUS_OUTPUT true
ADIABAT_PTPATH_FILE composite4025_1600.path
! continuous melting
ADIABAT_CONTINUOUS true
ADIABAT_MINPHI 0.005
! Trace elements
ADIABAT_DO_TRACE true
ADIABAT_TRACE_INPUT_FILE morb_static.trace
ADIABAT_TRACE_OUTPUT_FILE trace.out

Batch file:
1
morb.melts
4
1
14
morb25res10.melts
1
11
adiabat.out
12
1
ME.int
TE.int
0
0

morb.melts:
Title: anhydrous adiabat morb
Initial Composition: SiO2 48.68
Initial Composition: TiO2 1.01
Initial Composition: Al2O3 17.64
Initial Composition: Fe2O3 0.89
Initial Composition: Cr2O3 0.03
Initial Composition: FeO 7.59
Initial Composition: MgO 9.10
Initial Composition: CaO 12.45
Initial Composition: Na2O 2.65
Initial Composition: K2O 0.03
Initial Composition: P2O5 0.08
Initial Pressure: 40000
Initial Temperature: 1600
log fo2 Path: None

[jakemoore]

Paula,
I tried finer increments for the PT-path and am still having the MORB composition crash at ~35000 bars.  It looks like this method isn't working.  I'll have to come up with something else.
Thanks,
Jacob

[Paula]

Hi Jacob,

It looks like the crash is because K₂O and P₂O₅ are both being exhausted at this point.  I tried restarting it at the last good point, i.e. 35500 bars, rather than the point at which it crashed.  I wrote out the residue composition as a melts_file and then used this as input with the same starting conditions (splitting the P-T path into two files starting and finishing with 35500 bars respectively).  K₂O and P₂O₅ were rounded to zero in the residue melts_file.  The calculation ran to the end successfully and the modes and phase compositions varied smoothly across the restart, to within at least couple of decimal places for each oxide.

Altering the step size is definitely a sensible thing to try in situations like this.  Also varying the size of MINPHI / MINF can help sometimes, though not in this particular case.

Hope that helps (belatedly).  If you have any more queries please don't hesitate to post again (I've been on leave but I'll be around more from now on).

Paula