============================================================================

The discussion below is not current.

Refer to the commentary in the header of 

www.perplex.ethz.ch/perplex/datafiles/solution_model.dat 

for discussion of the current implementation and control of subdivision schemes.

============================================================================

John X (name changed to protect the innocent) recently introduced me to your 
thermodynamic modelling package, perple_X, which we have subsequently been 
using to (try to) constrain the P-T conditions of the metamorphic rocks I'm 
studying for my PhD. We have so far mainly calculated pseudosections for the 
bulk composition and parageneses of my rocks and the results are nicely consistent 
with other estimates. However, calculation times are very long because we 
incorporated a large number of solution phases. The number of pseudo-components 
is therefore huge (in the order of 600000) and we were wondering if we could 
restrict this in some way. The program now calculates complete ranges in mineral 
compositions from one endmember to the next, which is not necessary in our 
calculations. In fact even not desired, because we keep getting almost pure 
spessartine garnets in our results down to very low temperatures. Is there a 
way to restrict the range of pseudo-component compositions for the solution phases? 
I've tried modifying the pseudocompound_glossary.dat file, but vertex doesn't allow that 
while running and keeps making a new file, when you restart the program.




(the solution model file is what you have to modify to change pseudocompound 
properties, NOT the pseudocompound_glossary.dat file; the format is 
partially documented within by comments and you might get more info 
from vdoc.pdf and the perplex tutorial) 

in which the mn-endmembers are listed first and therefore the compositional 
range of these endmembers is easily controlled. for example in the Gt(HP) 

Gt(HP) 
2                                     model type: Margules, endmember fractions.                                                                               
1                                     non-reciprocal solution 
4                                     number of endmembers 
spss alm  py   gr                     endmember names 
  1  0  0  0                          endmember flags 
0. 0.2  0.01 
0. 1.   0.02125 
0. 1.   0.02125                 0     subdivision ranges and model 
... etc etc 


the line in red (if the formatting of this message survives) will create pseudocompounds 
with x(spss) = 0-0.2, if you'd like instead to have x(spss) in the range 0.8-1, then you'd 
replace the line with 

0.8 0.1  0.01 

the 2nd and 3rd lines define the ranges of the 2nd and 3rd endmembers, so additionally 
if you know x(py) is never greater than 0.2, you could replace the 3rd line by 

0. 0.2 0.01 

etc. 

you do get an enormous savings in time by playing such games, but you should 
be careful that you don't restrict the compositional range so much that you actually 
eliminate the stable compositions of the solutions you are interested in.