Changes from inv021208

1) V0 of mgts as per breyetal_99

2) Change in func.f and dfunc.f:

Gibbs free energy of formation (Gform(nspecp)) now computed property instead of being approximated as F_0

3) Thermochemical data for gr are used

4) cats phase equilibria used in inversion.  No cats thermochemistry used.

5) Assume tetrahedral disorder in mgts

6) No thermochemical data for spinel
 
7) Change in Gleib.f

Include derivative of PV term wrt theta_0

8) No thermochamical data for an

9) No thermochemical data for ab

10) Use gasparik_81 data for cats phase equlibria instead of hays_66 used until now

11) Set W_Na-Ca^M1 (cpx)=0

12) Set W_Al-Mg^M2 (cpx) = 24.3

13) Assume octahedral ORDER in mgts

14) Write formulae of en and fs to be consistent with (14), i.e. as Mg_2Mg_2Si_4O_12

15) Octahedral disorder in mgmj

16) Changed formulae of gr, al, py to be consistent with (16), e.g. Mg_3Al_2Si_3O_12

17) Added W_Ca-Mg=10 kJ/mol in garnet as per Haselton & Newton (1980) quoted by Wood and Holloway (1984) pg. 169

18) Removed perkinsetal_81 en+mgts+sp phase equilibria data

19) Corrected phase equilibria data of akaogietal_99 and irifuneetal_99 for sp=pe+co=mgcf

20) Added danckwerthnewton_78 phase equilibrium data on py + fo

21) Added data on en+sp=mgts+fo equilibrium from danckwerthnewton_78 and references therein

22) Corrected py+fo data from perkinsetal_81 (friction correction to pressures using medium B)

23) Removed sp to pe + co stability data

24) Added en+mgts+sp stability data from danckwertnewton_78 and references therein and perkinsetal_81

25) Corrected (24) to reaction proportions

26) Added mgts+di=cats+en equilibrium of gasparik_84 (1.58 GPa, 1573 K)

26) Added more mgts+di=cats+en equilibrium of gasparik_84 (all 1573 K data of opx+cpx+sp+fo assemblage)

27) Changed W_Mg-Al^M2 to 31.3 kJ/mol in cpx

28) Changed W_Na-Ca^M1 to -7.0 kJ/mol in cpx to preserve the value W_di-jd=24.3 found by Holland

32) Corrected error in invert.f in computation of configurational entropy

32) Added en=cen (+di) equilibria from perkinsnewton_80

33) Corrected error of factor of two in entry of en=cen (+di) equilibria of perkinsnewton_80

34) Fixed two errors in entry of en+mgts+sp=fo+py stability data

35) Change in func.f and dfunc.f:

chi-squared of formation energies now computed per formula rather than per atom.  
This is sensible since the experimental standard
deviation, which appears in the denominator of our chi-squared is per formula.

36) Fixed error in computation of calorimetrically constrained Debye temperature: sign error in configuration contribution.
Note that the code was actually okay before the (32) fix!!!  Now it is okay again!

37) Added sp thermochemistry

38) Added fa+qtz=fs equilibria of bohlenetal_80

39) Added an+fa=gt equilibria of bohlenetal_83

40) Removed fs thermochemical data

41) Changed opx formulae:

mgts: Mg_2Al_2Al_2Si_2O_12
en:   Mg_2Mg_2Si_2Si_2O_12
fs:   Fe_2Fe_2Si_2Si_2O_12

42) Remove data of boydengland_64 on en+mgts stability (29 points) as this data disagrees with those of
perkinsetal_81 who discuss the discrepancy on pg. 107

43) Include sp thermochemistry

44) Started from different initial guess with larger Theta_mgts and lower F0_mgts.  Results did not write out
properly because I had mistakenly deleted the /invert subdirectory in ~/work/qh/invert.  But it is clear from the out 
file alone that the results are similar to inv032508c

45) Changed cpx formulae:

cats:  Ca_2Al_2Al_2Si_2O_12
di:    Ca_2Mg_2Si_2Si_2O_12
cen:   Mg_2Mg_2Si_2Si_2O_12
he     Ca_2Fe_2Si_2Si_2O_12

46) Fit to G_el at Tsx, rather than at 300 K

47) Changed W_Mg-Al^M2 in cpx to 24.3

48) Changed W_Na-Ca^M1 in cpx to 0.0

49) en thermochemistry of thieblotetal_99 rather than robieetal_95

50) Changed W_Mg-Al^M2 in cpx to 24.3

51) Changed W_Na-Ca^M1 in cpx to 0.0

52) Removed cats=mgts Al partitioning data of gasparik_84

53) Changed cpx formulae:

cats:  Ca_2Al_2(Al_2Si_2)O_12
di:    Ca_2Mg_2Si_4O_12
cen:   Mg_2Mg_2Si_4O_12
he     Ca_2Fe_2Si_4O_12

54) Include cats=mgts Al partitioning data of gasparik_84

55) Remove co=appv pseudo-data

56) Add alpv=appv pseudo-data

57) Add mgpv=mppv data from tsuchiyaetal_04

58) Removed alpv=appl pseudo-data because this is really problematic, alpv not being stable anywhere!

59) Added co=appv pseudo-data

60) Allow W_il to vary

61) Corrected error in entry of perkinsnewton_80 cen+di=en data

62) Added odi to control file

63) Added carlsonlindsley_88 data on en+di=cen+odi at 1.5 GPa

64) Added odi to gasparik_89 data on en+di=cen+odi

64) Added odi to perkinsnewton_80 data on en+di=cen+odi

65) Fixed error in entry of data of carlsonlinsley_88

66) Entered cats=gr+co data from gasparik_81 directly, rather than as quoted by woodholloway_84

67) Added anorthite thermochemistry

68) Removed grossular thermochemistry

69) Changed cpx formulae to remove Al-Si tetrahedral disorder

70) Adjusted F_0 and Theta_0 of cats before inversion since straight inversion of previous result produced 
misfit to cats=gr+co equilibria.  Values are the inverted values, albeit very similar to input values.  Error in 
inversion routine?  Perhaps in computation of derivatives?

71) Replace cats tetrahedral disorder

72) include again cats=mgts Al partitioning data

73) Remove thermochemistry constraints except for anchor speces: pe, co, an, qtz, jd, fa

74) Add entropy only constraints for fo, en, di, py, al, coes

75) Remove all py thermochemistry, this was added in 74 inadvertantly due to an error in modifying the cp data file

76) Add wu entropy (not free energy)

77) Add Herzberg data on plg=sp in CMAS

78) Remove an thermochemical F_0

79) Remove Herzberg data on plg=sp in CMAS

80) Add an thermochemical F_0 and lower this arbitrarily by 4 kJ/mol to -4011.9 kJ/mol

81) Remove an thermochemical F_0

82) Use di thermochemical F_0

83) Adjust an thermochemical F_0 in cp data file arbitrary to be less stable by 8.5 kJ/mol to -3999.4 kJ/mol

84) Restart of inv040408d

85) Restart of inv040408e allowing four species comparisons

86) Add sp=pe+co equilibria

87) Add an thermochemistry

88) Split en+di=odi+cen equilibria in two to properly represent phase coexistence, rather than cation exchange.

89) Set S(odi)=S(di)

90) Set W^il_Fe-Al=W^il_Mg_Al and allow neither to vary

91) Remove all data from gasparik_90a (listed as gasparik_90 in the data file)

92) Remove all data from sawamoto_87, MgO-SiO2 data from irifuneetal_96, irifuneetal_98

93) Remove pv-mw partitioning data of: kessonfitzgerald_91, guyotetal_88 (no T measurement); kessonetal_02 (1 Mbar 
probably within the LS Fe regime in mw)

94) Use new jacksonetal_07 data for dK/dT, dG/dT of en

95) Trim fo-wa-ri-pv+pe data

96) Trim en-mgc2-mgwa+st-mgri+st-ak-mgmj-pv data

97) Trim fs-fec2-feri-wu+st data

98) Trim di-gt-capv data

99) Trim py-mgmj-alpv-mgpv-co-mgil-en data

100) Trim ol-wa-ri-pv-mw Mg-Fe data

101) Artificial feil ohtani datum restored

102) itotakahashi_89 stability data restored

103) Remove artificial feil ohtani datum

104) Add new feil ohtani artificial datum based on approximate il-pv Mg-Fe equipartition

105) Remove itoyamada il-ri Mg-Fe partitioning (retain il-pv Mg-Fe partitioning datum).

106) Remove itotakahashi_89 stability data

107) Adjust new feil ohtani artificial datum from 24 GPa to 23 GPa

108) Remove all en/mgc2=py+mgmj data

109) Restore presnalletal_98 data on en, mgc2

110) Correct error in entry of presnalletal_98 mgc2 data

111) Remove akaogietal_02 pymgmj datum

112) Restore two mgc2=cen points that were erroneously removed in nstep 108.

113) Trim remainder of stability data (pv-ppv, sp-pe+co-mgcf)

114) Delete all cen+di stability data

115) Allow W_Mg-Al^il and W_Fe-Al^il to vary

116) Restort cen+di stability data

117) Trim en+sp=py+fo stability and equilibrium data

118) Trim odi+cen=di+en equilibriao

119) Remove fo=gt, wa=gt, ri=gt Mg-Fe partitioning of Frost_01

120) Change V(odi)=V(di)

121) Changed V(odi) to mean of linear extrapolation of solid solution and V(di)

122) Add cen-di super-solvus phase stability after carlsonlindsley_88

123) Allow Theta_0 of odi to vary

124) Change V(odi) to that from extrapolation of solid solution (70)

124) Change V(odi)=V(di)

125) Change V(odi) to that from extrapolation of solid solution (70)

126) Remove en+sp=py+fo stability and equilibria

127) Restort en+sp=py+fo stability (not equilibria)

128) Remove sp=pe+cp stability

128) Remove en+sp=py+fo stability

129) Restore en+sp=py+fo equilibria

130) Remove en+sp=fo+py stability and equilibria

131) Restore sp=pe+co stability

132) Fix W_Mg-Al^il=66

133) T-site disorder in opx

134) Finite and variable W_Mg-Al^opx

135) Set W_Mg-Al^opx=0

136) Inclue en+sp=fo+py equilibria

134) Finite and variable W_Mg-Al^opx

136) Remove en+sp=fo+py equilibria

137) Restore en+sp=fo+py stability

138) Remove sp=pe+co stability

139) Set W_Mg-Al^opx=0

140) Remove T-site disorder in opx

141) Add en+sp=fo+py equilibrium data

142) Re-enter perkinsetal_81 opx=gt data (cf. excel file)

143) Changed V(odi) to mean of linear extrapolation of solid solution and V(di) (~68)

144) Changed dK_T/dT of en to that of di

145) Change opx formulae to Mg2Mg2Si4O12-Mg2Al2Si2Al2O12 etc.

145) Change cpx formulae to Ca2Mg2Si4O12-Ca2Al2Si2Al2O12 etc.

146) Remove en+sp=fo+py equilibrium data

147) Remove al=co+wu+st stability data

148) Add fo+al=fa+py equilibria 

149) Adjust G_0(an)=-4012.9

150) Restore en+sp=fo+py equilibrium data

151) Remove en+sp=fo+py equilibrium data

152) Add ky

153) Add an=gr+ky+co stability data

154) Add ky=st+co stability data

155) Change W_Ca-Na^cpx(1)=10.1, W_Mg-Al^cpx(2)=14.2

156) Add T-site disorder in cats

157) Remove T-site disorder in cats

158) Add gr thermochemistry

159) Change W_Ca-Mg^gt=15 kJ/mol

160) Add cats thermochemistry with 2Rln2 added to S_0

161) Remove added 2Rln2 from S_0 of cats

162) Add T-site disorder in cats

163) Remove T-site disorder

164) Restore added 2Rln2 to S_0 of cats

165) Set W_Mg-al^cpx=0

166) Remove 2Rln2 from cats S_0 in data file

167) Hardwire additional 2*2Rln2 (per 12 O) to S of cats in cp.f

168) Remove hardwired 2*2Rln2 from cp.f

169) Correct error in computation of vibrational entropy for inverting Theta in invert.f

170) Add T-site disorder in cats

171) Allow W_Mg-Ca^cpx to vary

172) Add en-odi stability datum at 1.5 GPa (carlsonlindsley)

173) Fix W_Mg-Ca^cpx=26.3

174) Remove disorder from sp, hc

175) Restore disorder to sp, hc

176) Add sp thermochemical entropy from chamberlinetal_95

177) Set S(cen) = S(en)

178) Remove cen thermochemistry

179) Fix W_Mg-Ca^cpx=26.3 (this had been left at a similar, inverted value in inv051208a,b)

180) Remove di+cen, en+odi stability

181) Allow W_Mg-Ca^cpx to vary

182) Remove di-cen and odi-en stability

183) Remove cendienodi equilibria at P>10 GPa

184) Set V(cen)=63.5 (was 63.0)

185) Set V(cen)=64.0

186) Set V(cen)=63.0

187) Set V(odi)~70

188) Add low-P en stability (two data)

189) Set W_Mg-Ca(cpx)=29.3

190) Remove 10 GPa cendi=enodi equilibria

191) Set V(odi)~70

192) Set V(odi)~66

193) Remove T-site disorder from cats

194) Hardwire additional 2*2Rln2 (per 12 O) to S of cats in cp.f

195) Change W_Mg-Al^cpx(2)=0.0, W_Na-Ca^cpx(1)=24.3

196) Change V(cen)=63

197) Change W_Mg-Ca^cpx(1)=24.47 after hollandetal_79

198) Changed V(odi)~68

199) Add extra sites to regular solution parameter files

200) Change W_Mg-Al^cpx=14.2, W_Ca-Na^cpx=10.1

201) Add T-site disorder to cats

202) Change W_Mg-Al^cpx=0.0, W_Ca-Na^cpx=24.3

203) Remove T-site disorder in cats 

204) Restore hardwired 2*2Rln2 (per 12 O) to S of cats in cp.f

205) Homogenize cf formulae by changing mgcf to Mg_1Al_1Al_1O_4, and fecf to Fe_1Al_1Al_1O_4

206) Remove he system stability

207) Remove coes thermochemistry

208) Remove O-site disorder from mgmj

209) Restore O-site disorder to mgmj

210) Add sp=pe+co stability

211) Allow for partial T-site mixing in opx: Mg_2Mg_2Si_2Si_2O_4 etc.

212) Remove sp=pe+co stability

213) Remove T-site mixing in opx

214) Lower G_0 of di by 10 kJ/mol

215) Add sp=pe+co stability

216) Remove di G_0 as constraint

217) Add saikiaetal_08 py+capv=gr+mgmj equilibria datum

218) Remove en+sp=py+fo equilibria

219) Add en+sp=py+fo stability

220) Remove sp=pe+co stability

221) Remove en+sp=fo+py stability

222) Add en+sp=fo+py equilibria

223) Add plg=sp=gt stability from herzberg, wood

224) Remove plg=sp=gt stability

225) Remove en+sp=fo+py equilibria

226) Add sp=pe+co stability

227) Restore di G_0 as constraint

228) Add en+sp=py+fo equilibria datum from danckwerthetal_78

229) Eliminate 2 lowest pressure py stability data from hiroseetal_01a

230) Set G_0 of py so that en+sp=py+fo datum of danckwerthetal_78 at 2.025 GPa, 1373 K is matched.  Note change in 
formulation of this data (and that of fujii_76) from en+sp=mgts+fo to en+sp=py+fo

231) Set G_0 of py to experimental value

232) Remove G_0 of py as constraint

233) Add remaining two danckwerthetal_78 data

234) Add fujii_76 data on coexistence line as per reformulation of (230)

235) Add saikiaetal_08 equilibria datum

236) Remove G_0 (di) as constraint

237) Remove O-site disorder in mgmj, rewrite other gt formulae accordingly

238) Restore coes entropy as constraint

239) Remove sp entorpy as constraint

240) Restore O-site disorder to mgmj

241) Restore chamberlinetal sp entropy as constraint

242) Fix but in calculation of G(qtz) by adding call to "landau" in Ftotsub.f

243) Remove sp entropy as constraint

244) Add W_Mg-Al^cpx=14.2 

245) Set W_Mg-Al^cpx=0

246) Set W_Mg-Ca^gt=10

247) Correct error in entry of ky+gr+qtz data (originally entered as ky+gr+co!)

248) Correct error in inversion V0(coes).  Take this directly from vol, whereas previosly it was calculated from comp

249) Set W_Mg-Ca^gt=5

250) Remove ol-gt Mg-Fe partioning equilibria

251) Remove py stability data of hiroseetal_01a

252) Use thermochemical G_0 or py as constraint

253) Restore ol-gt Mg-Fe partitioning equilibria

254) Restore py stability data of hiroseetal_01a except two lowest T points

255) Remove G_0 of py as constraint

256) Remove sp=pe+co stability data

257) Remove hiroseetal_01a py stability data

258) Restore py stability data of hiroseetal_01a except two lowest T points

259) Restore G_0 of di as constraint

260) Add G_0 of py as a constrain

261) Add sp=pe+co stability data

262) Remove sp thermochemistry

263) Remove sp=pe+co stabiilty data

264) Add sp thermochemistry

265) Note: Same as inv053108g

266) Remove ky=st+co data 

267) Add ky=qtz+co data

268) Add artificial constraints on st=apbo to better constrain Clapeyron slope to that estimated by murakamietal_03

269) Set initial guess of Theta(fppv) to theta_el(fppv)/theta_el(fepv)*theta0(fepv)=843

270) Remove kobayashietal_05 ppv=pv Fe-Mg partitioning datum

271) Add hiroseetal_08 ppv=pv Fe-Mg partitioning datum

272) Remove ppv stability data of maoetal_05

273) Add G0, Gop of fppv from stackhouseetal_06 

274) Set initial guess of Theta(fppv) to theta_el(fppv)/theta_el(fepv)*theta0(fepv)=757 based on (273)

275) Remove mgpv=mppv data of tsuchiyaetal_04

276) Add mgpv=mppv data of tatenoetal_08_4 which contains some pressure-re-calibrated data of hiroseetal_06b

277) Change initial guess at F0 and Theta0 of mppv by hand to agree better with tatenoetal_08_4 data

278) Changed initial guess at F0 and Theta0 of apbo by hand to agree better with estimated transition of murakamietal_03

279) Corrected choice of comparison assemblages.  Previous calculation of G for this purpose was 
approximate and especially so at high pressure.  Now I just base the choice of comparison assemblages
on the exact value of G.  I also increased the tolerance: diffmax=10 now as compared with only 4 before.

280) Remove maoetal_97 pv-mw Mg-Fe partitioning data on the grounds that this is based on lattice parameters
which can fail through the HS-LS transition.

281) Remove katsuraito_96 pv-mw Mg-Fe partitioning data as it seemed redundent with that of martinezetal_97 
and the latter has the the advantage of constraining the T-dependence of K.

281) Add sinmyoetal_08 pv-mw Mg-Fe partitioning data.

282) Restart

283) Change V0(fepv)=25.485 from jeanlozthompson_83 to better match pressure dependence of K_pv-mw from sinmyoetal_08

284) Remove frost_03 gt-ol Mg-Fe partitioning data

285) Add hacklerwood_89 gt-ol Mg-Fe partitioning data

296) Remove G_0(he) as constraint

297) Add first opx-cpx Mg-Fe partitioning datum of lindsley_83

298) Replace he with cfs: used lindlsey_81 for V_cfs, and opx-cpx Mg-Fe partitioning datum of lindsley_83 recast appropriate for change in composition of Fe cpx endmember.

299) Changed W_Ca-Mg^gt=5 kJ/mol

300) Changed W_Ca-Mg^gt=10 kJ/mol

301) Add en+co=py equilibrium of gaspariknewton_84

302) Remove remaining akaogietal_02 stability data (py, mgil+co, mgpv+alpv)

303) Remove akaogietal_87 py+mgmj stability data

304) Remove F_0(py) as constraint

305) Changed W_Ca-Mg^gt=15 kJ/mol

306) Remove gasparik_84a cats=gr+co data

307) Add haysetal_66 cats=gr+co data

308) Changed W_Ca-Mg^gt=10 kJ/mol

309) Add gasparik_84a cats=gr+co data

310) Remove haysetal_66 cats=gr+co data

311) Set cats S_conf=0

312) Set cats S_conf=2*7.7 J mol K as per beniseketal_07

313) Set W_Mg-Al^cpx(2)=17 and W_Na-Al^cpx(1)=7.3

314) Set cats S_conf=0

315) Set W_Ca-Mg^gt=0

316) Remove remaining py stability data of hiroseetal_01a

317) Set W_Ca-Mg^gt=10

318) Set cats S_conf=4ln2

319) Change formula of cats to Ca_2Al_2(Si_2Al_2)O_12

320) Remove cats additional S_conf-4ln2 from cp.f

321) Set W_Ca-Mg^gt=5

321) Set W_Al-Mg^cpx=17

322) Changed V(cen)=64

322) Changed V(cen)=62

322) Changed V(cen)=62.5

323) Set W_Ca-Mg^gt=10

324) Remove saikiaetal_08 capv=gt datum

325) Corrected errors in comparisons of free energies which are now properly computed taking into account
the chemical potential, and not just the end-member free energies as before.

326) Removed bohlenetal_83 an+fa=al+gr data since these to not constrain F_0(an),F_0(gr) independently of
an=gr+ky+qtz

327) Add an=cats+qtz data of hariyakennedy_68

328) Start from inv120509a as this had F_0(gr)=-6273 in better agreement with thermochemistry than with 
the value F_0(gr)~-6257 from inv250509a

329) Re-start from inv250509b

330) Add univariant cpx=ga top-of-solvus datum from gasparik_84a

331) Changed W_Ca-Mg^gt=5,10; W_Mg-Al^cpx=2,31, W_Ca-Na^cpx such that di jd is still symmetric with W=24.3

332) Changed V(mgts)=59.14

333) Remove danckwerthnewton_78, fujii_76 opx+sp=fo+py data

334) Add danckwerthnewton_78, fujii_76 en+sp=mgts+fo data

335) Remove danckwerthnewton_78, fujii_76 en+sp=mgts+fo data

336) Remove gaspariknewton_84 en+sp=mgts+fo data

337) Set W_Ca-Mg^cpx=27.89 kJ/mol as per Hollandetal_79,80, evaluated using there W_V term at 3 GPa

338) Set S(odi)=S(di)

339) Remove carlsonlindsley_88 cms opx=cpx data at 1.5 GPa

340) Add carlsonlindsley_88 cms opx=cpx data at 3.0 GPa

341) Set S(cen) = S(en) + 2.75 kJ/mol K

342) Remove constraint on S(odi)

343) Set W_Ca-Mg^cpx=24.74 kJ/mol as per Hollandetal_79,80, evaluated using their W_V term at 0 GPa

344) Set S(cen) = S(en) + 2.75/2 kJ/mol K as in hollandetal_79.  This is corrected from (341) above because
the hollandetal_79 DS is based on a 10 atom formula while my cp data file uses a 5 atom formula for en and cen
as in robieetal_95

345) Set V(cen) = V(en)

346) Set alp(cen) = alp(en) This and (345) done so that the entropy contrast: S(cen) - S(en) remains more constant from 1000-2000 K.

347) Set W_Ca-Mg^cpx=27.89 kJ/mol as per Hollandetal_79,80, evaluated using their W_V term at 3 GPa

348) Set W_Ca-Mg^cpx=22 kJ/mol

349) Set W_Ca-Mg^cpx=24.74 kJ/mol as per Hollandetal_79,80, evaluated using their W_V term at 0 GPa

350) Set S(cen) = S(en)

351) Set W_Ca-Mg^cpx=22 kJ/mol

352) Change W_Mg-Al^cpx=17,17

353) Set W_Ca-Mg^cpx=24.74 kJ/mol as per Hollandetal_79,80, evaluated using their W_V term at 0 GPa

354) Change W_Mg-Al^cpx=41,2

355) Set S(cen) = S(en) + 2.75

356) Remove S(cen), V(cen), alp(cen) constraints

355) Change W_Mg-Al^cpx=36,2

356) Change W_Mg-Al^cpx=39,2

357) Return to inv270509b: W_Mg-Al^cpx=31,2; W+Ca-Mg^cpx=24.74; V(cen)=62.5; no constraints on 
S(cen) alp(cen).  Only difference is that now we fit to the 3 GPa data (rather than the 1.5 GPa data)
of carlsonlindsley_88

358) Set W_Ca-Mg^cpx=27.89 kJ/mol as per Hollandetal_79,80, evaluated using their W_V term at 3 GPa

359) Set W_Ca-Mg^cpx=22 as it seems this may improve the cmas result

360) Set W_Ca-Mg^cpx=19 as it seems this may improve the cmas result and decreasing this parameter seems
to have very little adverse affect on fits to the data

361) Set W_Ca-Mg^cpx=24.74 kJ/mol

362) Set alp(cen) = alp(en) in order to make DS(cen-en) more nearly constant with increasing temperature

363) Return to the baseline (i.e. 357, i.e. remove constaints on alp(cen)) and set W_Mg-Al^cpx=31,12.  
Reasoning: M1 Mg in cpx needs to be de-stabilized, but not at the expense of Al in cpx.  Changning W in this 
way will influence the stability of cats via the solvus reaction of gasparik_84b much more than
Al in cpx for peridotitic compositions which are at much lower cats concentrations.  The second W_Mg-Al^cpx
(the one now set to 12) influences energetics at large cats concentrations more.

364) Set W_Mg-Al^cpx=31,-8.  Reasoning: Changing the second W from 2 to 12 actually made F_0 (cats) LESS
stable.  So this should make it more stable.

365) Set W_Mg-Al^cpx=35,-8.  Previous result increased Al in cmas cpx and left MgO in cmas cpx little changed.
This change should improve both quantities.

366) Add opx=cpx Al,Mg partitioning datum of gasparik_84b

367) Remove univariant cpx=gt constraint from gasparik_84a

368) Re-express opx-cpx Al,Mg partitioning datum of gasparik_84b as mgts+di=cats+en, di=odi, and cen=en equilibria

369) Add klemmeoneill_00 opx=cpx datum as di=odi and cen=en

370) Remove gasparik_84b data

371) Express klemmeoneill_00 data as cen=en only

372) Set W_Mg-Ca^cpx=21, i.e. to the lower limit of that permitted by the calorimetric data according to newtonetal_79

373) Set W_Mg-Al^cpx=26,7

374) Set W_Mg-Al^cpx=36,-3

375) Set W_Mg-Al^cpx=36,7.  This shows that increasing the first W (31) decreases both Al and Mg in cpx.  Decreasing the second W (2) 
causes Al in cpx to decrease less, while leaving Mg in cpx little changed.

376) Set W_Mg-Al^cpx=31,2

377) Restore cpx=gt cmas univariant equilibrium datum of gasparik_84a

378) Set W_Mg-Ca^cpx=21

379) Set W_Mg-Ca^cpx=24.74

380) Set W_Mg-Al^cpx=26,7

381) Set W_Mg-Al^cpx=36,-3

382) Set W_Mg-Al^cpx=36,7

383) W_Mg-Al^cpx=31,2; W_Mg-Ca^cpx=24.74; Add klemmeoneill_00 di=odi equilibrium datum

384) Set W_Mg-Ca^cpx=21

385) Set W_Mg-Ca^cpx=24.74; W_Mg-Al^cpx=26,7

386) Set W_Mg-Al^cpx=36,-3

387) Set W_Mg-Al^cpx=36,7

388) Set W_Mg-Al^cpx=17,17, W_Mg-Ca^gt=10,10

389) NEW FORMULATION FOR REGULAR SOLUTION TERMS (hollandpowell_93)

390) Allow W_mgts-odi in addition to W_en-odi to vary.  Start from inv250509e

391) Start from inv050609a

392) NEWER FORMULATION FOR REGULAR SOLUTION TERMS (hollandpowell_03, asymmetric)

393) Re-fit di-cats solution enthalpy with W=52, alpha_cats=3.5

394) Re-fit py-gr solution enthalpy with W=59, alpha_gr=5.5.  Yielded x_gr~0.02 in cmas garnet

395) Set W_Mg-Ca^gt=40 symmetric.  Yielded x_gr~0.10 in cmas garnet

396) Set W^gt=59, alpha_gr=5.5.  

397) Remove gasparik_84a opx=cpx partitioning data

398) Add perkinsnewton_80 opx=cpx partitioning data at 1173 K and 2.5 GPa

399) Add perkinsnewton_80 cpx=gt partitioning data at 1173 K and 2.5 GPa

400) Weight reaction data by number of atoms only, i.e. remove formula that included an estimate of the uncertainty in the composition which 
had been applied only to those data containing cen and mgmj.

401) Set W^gt=30 symmetric

402) Add FeTiO_3 ilmenite

403) Add DfG0(fo) as constraint

404) Add DfG0(en) as constraint

405) Set S(odi)=S(di), alp(odi)=alp(di)

406) Remove DfG0(en) as constraint

407) Remove restrictions on S(odi), alp(odi)

408) Add di stability points along melting curve because odi was too stable in peridotite

409) Change V(odi)=70 because opx was re-entrant at ~10 GPa in tinaquillo at 1794 K.  I have subsequently discovered that the re-entrant behavior was due to an error
in the control file!  I had entered he rather than cfs as the Fe cpx species.

410) Changed alp(mppv), Vo(mppv), compression (mppv), dK/dT, dG/dT (mppv) according to guignotetal_07 and stackhouseetal_05b.  This may yield pv=ppv transitions that look more like seismology.

411) Add artificial mgpv+fppv=fepv+mppv datum at 4000 K to control Theta(fppv): Assume that partitioning is the same at 4000 K as at 1750 K

412) Add artificial mgpv+fppv=fepv+mppv datum at 4000 K to control Theta(fppv): Assume equipartitioning at 4000 K

413) Change V(fppv)=V(mppv)+V(fepv)-V(mgpv) based on updated values of V(mppv) and V(fepv)

414) Tighten bounds on co=appv transition (zero width)

415) Change co=ppv transition to more accurately reflect negative Clapeyron slope of tsuchiyaetal_05

416) Change W_Mg-Al^pv=40

417) Change W_Mg-Al^ppv=60

418) Reset alpv properties to those of Rh_2O_3 structured Al_2O_3 inclding V_0, and compression

419) Add co=Rh_2O_3 (alpv) transition to stability (tsuchiyaetal_05)

420) Add Rh_2O_3 (alpv) = appv transition to stability (tsuchiyaetal_05)

421) Reset solution energetics of mgpv-alpv to match results of tsuchiya^2_08

422) Allow W_mgil-co to vary

423) Set W_appv-mppv=68 close to the minimum required to produce ppv exsolution on the en-co join

424) Set W_appv-mppv=60, agrees with tsuchiya^2_08 enthalpy of mixing.

425) Reset physical properties of mgpv alp: katsuraetal_09; kgp: murakamietal_07a; kgt: wentzcovitchetal_04a (corrected)

426) Reset pe G:murakamietal_09

426) Reset wu G:marquardtetal_09

427) Restore he in place of cfs

428) Fix formula of jdmj

429) Add artificial jdmj=jd datum to constrain Theta(jdmj)

430) Add fecf=fepv Mg-Fe partitioning data (one artificial) to constrain F_0 (fecf) and Theta (fecf)

431) Found bugs in inversion: a) for pe, the initial guess at gamma produced spinodal instability
at T=2973 K, the maximum temperature of the fiquetetal_99 data.  The inversion then did not converge 
properly.  This problem has affected gamma and q of MgO for a long time, perhaps even in the 2005 GJI 
paper. b) Extra theory meant for elastic softening of st had a parameter number 42 which had not been
assigned a number.  This theory caused Go to be increased by 1 GPa for several phases.

432) Remove mgpv+alpv stability data

433) Add W_he-cen=W_di-cen

434) Remove one of the two opx-cpx Mg-Fe partitioning data since entropy of he is fixed by calorimetry

435) Add DFT constraint on pe+co=mgcf at 1 K

436) Add frost_03b data on wa-mw Mg-Fe partitioning

437) Remove 1600 K katsuraetal_04 ol-wa Mg-Fe partitioning

438) Fix W_ol, W_wa, W_ri to values given by frost_03b

439) Remove 1900 K katsuraetal_04 ol-wa Mg-Fe partitioning

440) Add feietal_90 wa-mw Mg-Fe partitioning since this is consistent with frost_03b and constrains Theta(fewa)

441) Remove wiserwood data.  This contained an error!  I had entered the temperature (1400 K) as 1673 K!  I have corrected this mistake in the data file.

442) Add ol=mw Mg-Fe partitioning of frost_03b at 6 GPa since at this pressure the data cover the largest variation in fa#

443) Allow W_ol,wa,ri to vary

444) Remove two points from wa=mw Mg-Fe partitioning data of frost_03b one each at 14 and 14.5 GPa which do not appear in his Fig. 1b

445) Add points from feietal_04a in which more than one phase was observed. 

446) Restore hiroseetal_01b mgpv stability at 2273 K, corrected for pressure calibration after feietal_04a.

447) Fix Theta(mgmj)=Theta(py)

448) Change garnet formulae to remove octahedral Mg-Si mixing in mgmj

449) Add W_py-mgmj=46 as per vinogradetal_06

450) Remove katsuraetal_04 fo-fa, mgwa-fewa stability data

451) Restore katsuraetal_04 ol-wa Mg-Fe partitioning data

452) Remove frostetal_03b wa-mw Mg-Fe partitioning data

453) Restore frostetal_03b wa-mw Mg-Fe partitioning data

454) Remove katsuraetal_04 ol-wa Mg-Fe partitioning data

455) Add gasparik_90 c2c=mgmj partitioning datum

456) Remove hiroseetal_01a pymgmj stability data

457) Restore saikiaetal_08 data on gr=capv equilibrium

458) Allow W_gr-mgmj to vary

459) Reset il-pv Mg-Fe partitioning data to force equipatitioning at 1373 and 2073 K.  I did this because:
a) theta_feil seemed to high, much higher than theta_fepv b) feil and feil+wu had stability fields in
fa and fs and c) a field of il+mw appeared in ol143

460) Narrow bounds on st=apbo transition

461) Change pyroxenes and c2 to 10 atom formula units

462) Add T-site mixing to an and ab

463) Change code by commenting out the following line:

c        if (sname(ispec) .ne. 'sp  ') sconf = smag

which meant that the configurational entropy was not considered in fitting theta to experimental 
calorimetry data.  This caused not errors in previous inversions because the only species for which
I determined theta calorimetrically AND that had disoder was cats for which I used the data of etzeletal_07
directly without a correction for configurational disorder.  With the change to the code, I fit instead to
the calorimetric data with the configurational entropy correction, yielding the same theta for cats and the 
same entropy for cats.  Theta of ab and an are different now because I am now including disorder in the formula
unit.  Previous inversions were actually inconsistent in their treatment of an since the data of Robieetal_95
includes the configurational entropy correction whereas my formula did not.

464) Remove T-site disorder in feldspars.  This seems to be the best assumption in the case of an for which
HP98 have a critical temperature of 2300 K.  On the other hand ab has TC=950 K and hollandpowell_96 show that
T-dependent disorder introduces visible curvature to the ab=jd+qtz boundary.  However, the effect is very subtle.
Further investigation reveals that the data of robieetal_95 does NOT include S_conf for an so previous inversions
treated an consistently, i.e. no Al/Si disorder in my an formula or in the experimental data. Note that inv220909
differs from inv060809 only in the pyroxene formula units.

465) Add critical transition for st=cacl2 based on the equation for the boundary given by onoetal_02: P=51 + 0.012 T
and the observation by andraultetal_98 that the DV of transition is less than experimental uncertainty (2 per mil), we have
TC=-4250 K, DV=0.03 cm^3/mol, DS=0.36 J mol-1 K-1.

466) Switch to wentzcovitchetal_06 for dG/dT of mppv.  Note error in inv220909 and inv290909: because these
were based on input paramameter files in which px was inconsistently switch from 20 to 10 atom formula
units vs-rho systematics and all G derived from them were wrong, for example appv.

467) Use stackhouseetla_05a for dG/dP, dG/dT and dK_S/dT of appv.  Note that previously I had
erroneously entered this data so that the code read Delta G as being at 0 GPa instead of 136 GPa.

468) Use tsuchiyaetal_04b for G, dG/dP of mppv.  This fits the experimental data of murakamietal_07, but
constraints the slope much better.

469) Fix eta(mppv)=eta(mgpv)

470) Use caracascohen_05b for G, dG/dP of appv

471) Use systematics for dG/dT of alpv (and appv).  This causes problems for appv (eta<0) since G(P=0) is
small.

472) Use stackhouseetal_05b for dG/dT and dK/dT of alpv and appv.

473) Use stackhouseetal_06a for dG/dP of fepv and fppv and for G of fppv.

474) Use stackhouseetal_05b for dG/dT of mppv.

475) Assume that dK/dT os stackhouseetal_05b refers to adiabatic K_S

476) Use sysematics for dK/dT of alpv and appv

477) Restart from inv290909

478) Add nepheline

479) With error function stability minimization and uniform error of 1.0 kJ/mol

480) With error function stability minimization and uniform error of 0.1 kJ/mol

481) With error function stability minimization and uniform error of 1.0 kJ/mol

482) Fix error in data entry for kgp/qtz.  Pmax incorretly set to 0 leaving Kop unconstrainted and producing inverted values Kop~20 and increasing with each new run.

483) Stabilize inversion of species properties so that they are reproducible from run to run to within tol=1.e-8

484) Test error function coding by inverting stability only.

485) Stabilize computation of func: with error function, comparison of stability to itself generates a contribution ln(2).  This was being added sometimes and others not.  Now I add it always and subtract nphs*ln(2)
afterwards.

486) Inversion of stability only

487) Invert stability and reactions

488) Clean up treatment of errors.  Previously: a) rnxerr=3 was hardwired in phssub.f for all reaction data except those involving cen 
or mgmj b) rxnerr=1 was hardwired in func and dfunc.  Changes: a) move bd and rxnerr to phas.inc b) remove the multiplication by rxnerr=3 
in phssub.f c) divide fnpr for cen and mgmj reactions by rxnerr d) set rxnerr=3*bd.  Result is identical with inv190510 as it should be.

489) With error function stability minimization and bd=0.3 kJ/mol, rxnerr=3 kJ/mol

490) Remove onoetal_0x pe+co=mgcf DFT data.  This forced a very negative Clapeyron slope which stabilizes mgcf on the en-py join at 2273 K and P>50 GPa.

491) Remove akaogietal_99 pe+co=mgcf data

492) Add irifuneetal_02 pe+co=mgcf data

493) Restore akaogietal_99 pe+co=mgcf data

494) Remove irifuneetal_02 pe+co=mgcf data

495) Use executable with old errors stored in OLD050510

495) Restart from inv050510 but with ab from inv161009. Goal is to get ab parameters as similar as possible to those of inv161009 while
including the jd=ab+neph data

496) Restart from inv050510.  Add jdpv.  Re-cast capv as CaCaSi2O6 (change parameter file and phase equilibria data as appropriate).  
et jdpv energetics by requiring cf-absent in pyrolite along the 1600 K adiabat.

497) 241010.  Double the formula of pe and wu.  This entailed changes to the parameter files, to the mw phase file, and to the phase equilibria data files.
Reproduce old results with final residual of ~0.811.  Return capv to un-doubled formula.  Add NaAlO2 data.
Data saved in data.231010.  Data with NaAlO2 data added saved in data.241010.  Inversion saved in inv231010

498) 251010.  Restart from inv050510.  Remove nao and jdpv from control file and phase equilibria data.  Use data.211010.  Purpose is to compute errors, which were not computed before.  Store results in inv251010.

499) 041110.  Add W_cen-jd=23.2 according to gasparik_92

500) 041110.  Add W_he-cats=W_di-cats and W_he-jd=W_di-jd.  Produced F_0, theta_0 for all species identical to inv251010 (to within 0.1).  This includes F_0 (jdmj) because the data that constrain its properties are relative to pure jd so the added W_cen-jd interaction has no effect.  Store results in inv011110.

501) 051110.  Allow W_cen-jd to vary.  Add cen-jd=mgc2 data point from gasparik_92.  Find a value of ~53 which is quite a bit higher the value in gasparik_92.  Store data in data.021110.  Store results in inv021110.

502) 051110.  Add NaAlO_2.  Rename this to anao.  Use data that was saved in data.241010 and add to this the new cen-jd=cen point.  Store data in data.031110.  Store results in inv031110.

503) 220614.  Add namj.  Remove jdmj.  Re-formulate gasparik_92 mgil=mgmj point in terms of namj.  Add new point from dymshitsetal_13 for gt+cf coexistence.  Invert for W_mgmjnamj and W_mgcfnacf.  Store data in data.220614.  Store results in inv220614.

504) 230614.  Add W_pe-anao=54, i.e. identical to that of W_jd-cen for one MgMg=NaAl exchange.  This was done because pe and anao showed too much intersolution on the mgcf-nacf join in comparison with the data of onoetal_09.

505) 230614.  Add nal (mnal, fnal, nnal).  Add sp+jd+anao=mnal+nnal stability from onoetal_09 and nal+cf coexisting compositions on mgcf-nacf join from onoetal_09.  Store results in inv230614.  Store data in data.230614.

506) 240614.  Correct data entry errors for nal stability.  Yields better solution.  

507) 240614.  Add enstatite melting points at 0 and 5 GPa.  Store results in inv240614.  Store data in data.240614.

508) 240814.  Add one artificial datum to constrain fnal Theta.  Yields sensible value (886 K).  Set W_pe-anao=60.  Smaller values
cause jd+co+pe+anao to be more stable than sp+jd+anao (iphs=347) because of the pe-anao solid solution.  

509) 310814.  Change mgcf and nacf K/K' constraints to experimental eos measurements (comp).  Use Brillouin data to constrain G/G'.  Similarly for mnal.  Change dG/dT and dK/dT of mgpv according to zhangetal_13.

510) 010914.  Add thermal expansivity constraints for mnal and mgcf.  Fix q of all cf and nal phases to 1.0

511) 020914.  Add stability of nacf0.7mgcf0.3 and 3mnal + 1nnal at 30

512) 030914.  Replace nacf0.7mgcf0.3 stability with nacf0.8mgcf0.2 stability

513) 050914.  Fixed calculus error in Gleib.  This produces better solutions (smaller values of func).  Did not converge after 500
iterations of dfpmin.

514) 060914.  Continuation of inv050914.  Still not converged to the required tolerance (1.e-12), which may be unrealistically small.  However, all derivative magnitudes are smaller than 0.01.  Also used streamlined comparison lookup as coded in stoich.f.  This will not find combinations involving negative amounts of species.  But these are now a bit cludgy anyway since negative amounts of species fall outside the scope of the HP03 ASF formulation.

515) 070914.  Set nnal size parameter to 5.0

516) 080914.  Set nnal size parameter to 0.2

517) 090914. Set nnal size parameter to 5.0.  Add cf=nal stability in terms of two balanced reactions instead of cation exchange.  Add fictive nal stability at 47 GPa.  Invert with nlopt.

518) Set nnal size parameter to 5.0.  Set nacf size parameter to 5.0.

519) Set nnal size parameter to 0.0.  Set nacf size parameter to 5.0.

520) 120914.  K_0mgcf=221 GPa. size parameters of nnal and nacf = 1.0

521) 130914.  Set size_nacf=2.0

522) 140914.  Adjust nal = cf coexistence reaction to make compositions more accurate.  Remove erroneous mgwa stability point at 22.82     2173.0    feietal_04a.

523) 150914.  Add pure mnal=cf point.

524) 170914.  size_nacf=3.0

525) 180914.  size_nacf=1.0

526) 190914.  size parameters all unity.  Change compression of mgcf back to suedaetal_09 values of K, K'. 

527) 200914.  Change K_0mnal=207 GPa.

528) 210914.  Change K_0mnal=208.  Change K_0mgcf=213.

529) 220914. size_nacf=2.0

530) 230914.  size_nacf=1.0.  K0mnal=206.

531) 240914.  size_nacf=2.0

532) 250914.  size_nacf=3.0

533) 260914.  size_nacf=2.2

534) 270914.  size_nacf=2.1

535) 300815.  Change various physical parameters including: elasticity of di, ak, coes, sp

536) 310815.  Also change elasticity of mgwa,fewa.  Add Schollenbruch et al. hc=co+wu equilibria.  Allow W(mgpv,alpv) to vary

537) 010915.  Also allow W(mgil,co) to vary.  

538) 020915.  Update eos for co, alpv, appv.

539) 030915.  Replace constraints on co=alpv and alpv=appv from tsuchiyaetal_05b with those of katoetal_13.  Both transition pressures are now much higher.

540) 081015.  Use wangetal constraint on G'(mgwa) but retain sinogeikin constraint on G_0(mgwa).   Remove Schollenbruch et al. phase equilibria data.   Use spezialeetal_04 for M,M' of fa.

541) 011115.  Assume W_fe,x=W_mg,x where fe and mg stand for Fe and Mg bearing end-members and x stands for non-Mg,Fe end-members.

542) 071115.  Add W_namj-py.  Set initially to a value of 10.0.

543) 081115.  Set W_namj-py to the value required to match the dymshitsetal_13 datum.  Set all W_cf and W_nal to zero.  

544) 091115.  Starting from 081115, allow W_cf and W_nal to vary.  size_nacf=1.0

545) 101115.  Optimize size_nacf

546) 111115.  Remove artificial mgcf+nacf=mnal datum.

547) 121115.  Set W_al-gr=0 after geiger_87.  Make the change to the code as well, although I did not re-run the code to reproduce a new out file.

548) 141115.  Set K'=4.1 for mnal.  This was done to avoid nal reappearring as a stable phase at very high pressure in basalt (e.g. 2600 K 120 GPa).

549) 151115.  Set W_fepv-alpv=0.  Make the change to the code as well, although I did not re-run the code to reproduce a new out file.

550) 161115.  Fix fo+al=fa+py data entry error (hacklerwood_89).

551) 010116.  Remove 1673 K pv=gt and il=gt points of kuboakaogi_00.  These two points produced the largest deviation and tend to de-stabilize pv.

552) 020116.  Remove all pv=gt and il=gt points of kuboakaogi_00 except those at 1873 K.  These tend to produce the most al-rich pv.

553) 030116.  Add mgmj component to gt for kuboakaogi_00 pv=gt 1873 K point.  This was necessary in order to reproduce experimental x_alpv in the full equilibrium MAS assemblage.

554) 040116.  Replace saikia point with two others at lower and higher pressure.  Allow W_py-gr to vary.

555) 050116.  Return to W_py-gr=30 fixed.  Allow size of gr to vary.

556) 060116.  Allow W_py-gr and size(gr) to vary.

557) 070116.  Return to W_py-gr=30, size(gr) = 1.  Change perkinsnewton_80 reaction to their reaction B and to their 1173 K 3 GPa point as specified in abstract.  
Also, remove gasparik_84a di+cats=gr+py stability points and replace by reaction (3) as specified in the abstract

558) 080116.  Use perkinsnewton_80 2.5 GPa, 1173 K point for gr (instead of 3 GPa point) with reaction B.

559) 090116.  Fix W_py-gr=61, size(gr)=5.6 (best fit to Newton et al. 1977 data).

560) 100116.  Fix W_py-gr=38, size(gr)=2.8 (best fit of Wood and Holloway 1984).

561) 110116.  Fix W_py-gr=38, size(py) = 0.35, i.e. same shape of H_xs(x_gr) as 100116.

562) 120116.  Fix W_py-gr=30, size(py), size(gr) = 1.  Allow size(mgmj) to vary.

563) 130116.  Replace saikiaetal_08 point with two others at higher and lower pressure.  Allow size(mgmj) to vary.

564) 140116.  Return to one saikiaetal_08 point.  Return size(mgmj)=1.  Replace pv=py, il=py points of kuboakaogi with two pv=il points.

565) 150116.  Add two more namj reactions to fully constrain F_0(namj), theta_0(namj), W_py-namj, W_mgmj-namj

566) 160116.  Add three points from bobrovetal_08 remove gasparik_92 gt=cf point.  Note.  This was a disaster.  The bobrovetal gt compositions seem wrong as they are not nearly rich enough in Na.

567) 170116.  Remove bobrovetal_08 points.  Add two gasparik_96b points: the two highlighted in the text as producing gt compositions closest to the di-jd join (pg. 479 column 2).  Found that finite mgmj content must be entered for the equation to make sense.  This may have been one of the problems with the entry of the bobrovetal_08 data, for which I entered 0 mgmj.  Note that in writeout, I was still mistakenly equating W_gr-namj=W_py_namj.  I edited the gt phase file by hand to propertly reflect the inverted value and fixed the code.  Fixed all gt W except for those involving namj.

568) 180116.  Allow other gt W to vary as well.

569) 190116.  Same as previous, but start from all W_namj-x=0.

570) 200116.  Allow W_py-gr to vary.  Start from inv180116.

571) 210116.  Set V_py-gr=1.03 cm^3/mol.  

572) 220116.  Same as 571, but start with W_gr-mgmj=0.

573) 230116.  Same as 572, but remove the gasparik_92 gt=cf datum.

574) 240116.  Same as previous, but start with size(nacf)=2.5

575) 250116.  Set V_gr-mgmj=1.03.  Start from inv140116.

576) 260116.  Set K_S0 (mnal)=206.  Start from inv140116.

577) 270116.  Restore artificial mnal=cf datum.  This was removed at 111115.

578) 280116.  Remove all katsuraetal_04 ol and wa stability points.

579) 290116.  Remove feietal_01 wa-mw partitioning data.

580) 300116.  Add frost_03 fewa=fa and fewa=feri triple point data.  Add grover_72 en stability.

581) 310116.  Replace mirwaldmassone_80 qtz=coes with bohlenboettcher_82.

582) 010216.  Add artifical morishimaetal_94 fo=mgwa point.

583) 020216.  Replace frost_03b ol=mw partitioning data at 6 GPa with that at 11 GPa.

584) 300316.  Add enm.  Physical properties from dft (mgsio3.xiao inversion all V,T).  F_0 and S_0 from opx melting curve.
              Note that the lfit inversion was performed with vofix=.false. which leads to inconsistency.

585) 310316.  Same as previous but with lfit inversion performed with vofix=.true.

586) 010416.  Replace high-P en melting equilibria with 25 GPa, 2900 K, mgpv melting.

587) 020416.  rerun lfit but with V/V_x<1.2

587) 030416.  rerun lfit but with 0.45<V/V_x<1.2.  Use en=enm at 0 and 5 GPa. (no mgpv melting constraint).

588) 040416.  Use en=enm at 0 GPa and mgpv=enm at 25 GPa, 2900 K.

589) 050516.  Update eos constraints on coes and st.  bourovarichet_06 for alpha of coes.  jiangetal_09 for K, G, K', G' of st.  wangetal_12 for alpha, dK/dT for st.  Update mppv eos using sakaietal_16 and more consistent dK_S/dT from Stackhouse & Brodholt (2007).

590) 050616.  Correct error in V_0(fppv).  As V_0(mppv) was changed, need to update V_0(fppv) according to the assumption V_0(fppv) - V_0(mppv) = V_0(fepv) - V_0(mgpv).

591) 310516.  Update constraints on elasticity of stishovite.  Compute errors.

592) 020716.  Add wuls.  Add constraints on the hs-ls transition in mw from solomatovaetal_16.

593) 030716.  Remove sinmyoetal and martinezetal pv-mw Mg-Fe partitioning data.  Add sakaietal_09 data.

594) 040716.  Add nakajimaetal_09 and frostlangenhorst_02 pv-mw Mg-Fe partitioning data.

595) 050716.  Remove sakaietal_09 partitioning data.

596) 250716.  Re-examine eos of pe: use hugoniot data as a constraint, which requires a much higher value of K_0' than previously.  Re-examine eos of wu: use eos constraint from fischeretal_11.  These choices turn out to match the adiabatic bulk modulus data of marquardtetal_09 on pe90wu10.

597) 260716.  Set K_0(wuls)=K_0(wu).

598) 270716.  Fit to x=0.1 Ptr and Width instead of x=0.17 Ptr.  Allow siz(wuls) to vary.

599) 280716.  Fix siz(wuls)=1.0.  Allow asymmetric W_pe-wuls interaction.

600) 290716.  Reset V0(wuls) based on dft.  Force symmetric W_pe-wuls, siz(wuls)=1.0

601) 300716.  Allow W_pe-wuls to be asymmetric and siz(wuls) to vary.

602) 310716.  Fit to pressures at 80% and 20 % completion of HS-LS transition (rather than 20 % and 50 %).

602) 010816.  Fit to pressures at 80% and 50 % completion of HS-LS transition (rather than 20 % and 50 %).

603) 020816.  Fit to 80 and 20 at X=17 and 80, 20, 50 at x=48.

604) 030816.  Fit to 80,20,50 at x=48 and x=50 only at x=17.  The idea being that the width at x=0 is depends mostly on V0(wuls)-V0(wu) and is independent of the W parameters.

605) 040816.  Fit to 80,20,50 at x=48 and x-50 at x=10.

606) 050816.  Set K_0'(wuls)=4, V_0(wuls)=21.5 to match highest pressure x=0.25 volume datum of maoetal_11b.

607) 060816.  For wuls, set V0=20.76, K0=214, K0p=4.0

608) 070816.  For wuls, set V0=20.76, K0=K0(wu), K0p=5.1

609) 080816.  For wuls, set V0=20.65, K0=220, K0p=4

610) 090816.  For wuls, set V0=20.65, K0=K0(wuls), K0p=5.3

611) 100816.  For wuls, set V0=20.65, K0=190, K0p=4.9

612) 110816.  Set V_pe-wu=0.22 cm^3/mol (2 oxygen basis) after frost_03.

613) 120816.  For wuls, set V0=20.65, K0=220, K0p=4, V_pe-wu=0.22 cm^3/mol

614) 130816.  For wu: set K0=162, K0p=4, wuls: V0=21.6998, K0=171, K0p=4.  From solomatovaetal_16 for x=0.35.  Note that K0 and K0p values for the end-members are set equal to those of the solid solution.  This is not correct!  Need to set the values of the end-members such that the values of the solid solution are reproduced!

615) 140816.  For wu and wuls: set K0p=4.  Find values of K0 such that highest pressure point from solomatovetal_16 fit to x=0.35 data for each phase are fit (57.05 for HS and 128.5 for LS).

616) 150816.  Change dM/dT of diopside to those of isaaketal_06 (was lineuville_10).  Change pressure at which dM/dT of mgil is constrained to the mid-point of the experimental range (~20 GPa).  Change V0(fepv) back to tangeetal_09 value.

617) 160816.  Add wu=wuls point at 2000 K from maoetal_11.  This constrains Theta_0 of wuls.

618) 170816.  Set V_pe-wu=0.  Set K(fepv)=258 to better match the pressure dependence of K_pv-mw.  Add auzendeetal_08 ppv-mw partitioning data.  Assume partitioning is with pure high spin mw.

619) 180816.  Reset V_pe-wu=0.22.  

620) 190816.  Set V_fppv to value from dorfmanduffy_14.  Create auzendeetal_08 pseudo-data so that ppv-fp partitioning produces
correct result for free spin.

621) 010916.  Add fea, feg, fee.  Add hem, hepv.  Note that Gibbs free energy of formation from the elements of fea (zero)
needs to be set to something other than zero in order to be recognized as a constratin by the code.

621) 020916.  Correct Gibbs free energy of formation from the elements of fea (zero).

622) 030916.  Iron only.  Previous parameter set (020916) is inconsistent.  Inversion for gamma(fee) depends on inverted value of F_0(fee) and theta_0(fee).  Need to re-run several times (~5) to obtain consistency.  This parameter set (030916) is consistent.

623) 040916.  Replace beta and gamma_el for fea, feg, fee with values fitted to results from stixrude_12.  Analysis saved in:
~/work/ms/published/2012/FeHiP/ElectronicGamma.pxp.

624) 050916.  Replace constaints on V_0 alp of feg with the range of volume over the stability field.

625) 060916.  Set feg magnetism to zero.

626) 070916.  Several changes to physical properties and constraints of fea, feg, and fee to improve fit to phase equilibria.

627) 080916.  Replace guinnanbeshers constaint on KS0, KS0' for fea with that of rottersmith_66.  This produces a larger KT0' which is essential for limiting the stability of fea to low pressure.

628) 090916.  All phases. 

629) 100916.  Add hppv.

630) 110916.  Add Mg-Fe2+-Fe3+ pv-mw partitioning datum from frostlangenhorst_02 and invert for W_mgpv-hepv

631) 120916.  Adjust eos of hepv and hppv.  Set gamma, q or both to those of hem.  Change V_0 of hem to that of hazenfinger_80 (remarkably different from the 4_smyth value).  Set Ko,Ko' of hepv to that of hem.  Readjust Vo hepv to match the itoetal_09 constraint.

632) 130916.  Add rest of H1495 Fe partitioning data from frostlangenhorst_02.  Rationalize hematite cp data by using full Cp function from robiehemingway_95 up to 900 K.

633) 140916.  Fix hepv equation of state by matching to Catalli et al. 2010 as well.  Return fepv K0=273 GPa.  Set gamma hepv so that DV of disproportionation reaction changes sign near 40 GPa in agreement with xuetal_15.  Set siz(alpv)=1.0

634) 150916.  Replace mophase/iophase logic with iastate logic.  Potential problem with mophase logic is that all atoms on all sites of the order-disorder pair were treated as being in distinct states, including e.g. oxygen.  In the course of de-bugging, produced an incorrect result in which configurational entropy of pe-wu-wuls was artificially reduced.  This led to better agreement with expeirmental observations on the width of the HS-LS transition at low iron content.  Saved in unplot/inv150916/hsls_reduced_sconfig.agr.  May be able to produce similar results property by restricting HS-LS mixing to a single site (e.g. strongly correlated HS-LS configurations, similar to Al-avoidance idea in feldspars).

635) 160916.  Add wu+st stability points from katoetal_16.

636) 170916.  Add hlpv.  Correct magnetic entropy of hepv (all high spin) and hlpv (one high spin, and one low spin) recognizing that low spin ferric iron has non-zero magnetic entropy.

637) 180916.  Changed gamma of hepv hlpv.  Set these equal to mgpv.

638) 190916.  Remove frostlangenhorst_02 pv(3+)-mw Mg-Fe partitioning data.  Set W_hepv-mgpv and W_hepv-fepv to zero.

639) 200916.  Add one lauterbachetal_00 point (with added Fe, the lowest Fe3+/Fe Total: #1579).  Allow W_hepv-mgpv to vary.

640) 210916.  Change W_hepv-mgpv by hand to match the lauterbachetal_00 point.  This is necessary because a significant fraction of hlpv is present.

641) 220916.  Back up.  Set hepv-hlpv Clapeyron slope to that of tsuchiyawang_13.  Set W_hepv-mgpv=W_hepv-fepv=0.  Remove lauterbachetal_00 point.

642) 230916.  Add mag as a species in mw.  Assume that the tetrahedral site does not mix with any other species.  Assume complete inverse order.  Assume no mixing on octahedral vacancies.  Constrain via
mag+Fe=wu stability and composition data of darkengurry_45.

643) 240916.  Set siz(mag) by hand to fit wustite compositions in equilibrium with iron and with magnetite from darkengurry_45.  Allow W_pe-mag also to vary by fitting to oneilletal_03 data on ferropericlase.

644) 250916.  Fix errors.  Formula of mag was incorrect.  Computation of equilibria in the forward code was incorrect (e.g. using wu4 as the name of the wustite phase).

645) 260916.  Set Omega_wu-mag to non-zero value chosen by hand to approximately match Fe_1-xO equilibria of darkengurry_45.

646) 270916.  Fix error in oneilletal_03 data entry.  Use 1473 K darkengurry points.

647) 280916.  Add additional oneilletal_03 points.  Allow Omega_pe-mag to vary.

648) 290916.  Change siz(mag) to 1.0 (from 0.42).

649) 300916.  Change siz(mag) to 2.0.

650) 011016.  Fix error in triple point composition.  Change siz(mag)=1.0

651) 021016.  Allow siz(mag) to vary.

652) 031016.  Add wu=Fe equilibria at the triple point from darkengurry_45.  Allow Omega_wu-mag to vary. Change wu Theta to 
fictive FeO of robiehemingway_95.

653) 041016.  Add artificial smag to be treated as a pure phase with invariant properties.  Add wu=smag equilibria from darkengurry_45.

654)

655) 061016.  Change wu Theta back to original stolenetal constrain.  Change mag Theta to match entropy at 1000 K.

656) 071016.  Change landau contribution for magnetite to hillert expression.

657) 081016.  Change formula of magnetite to that of normal spinel Fe_1Fe_2O_4.  

658) 091016.  Set Omega_wu_mag=0.

659) 101016.  Adjust W_wu_mag and F0(mag) by hand to better match the high oxygen boundary of the wustite stability field.  First, find the value of W_wu_mag that matches this boundary.  Then, adjust F0(mag) so that the low oxygen boundary matches.  Note that variations in F0(mag) do not affect the high oxygen boundary.  Change F0(hmag) by the same amount (+1 kJ/mol).

660) 111016.  By hand, Change W_mag-pe=137, O_mag-pe=0.  Suspect that full inversion for O_mag-pe overfits the data given scatter.

661) 121016.  Fix W_mag-pe, W_mag-wu=W_mag-wuls.  Remove wu=mag+Fe stability points.  Remove smag reaction points.  Invert for F_0(mag).  Replace hiroseetal_08 point with ppv=mw Mg-Fe partitioning from sinmyoetal_08.  Allow W_mgpv-hepv to vary.

662) 131016. Correct apparent error in sinmyoetal_08 OL122 point: K doesn't match phase compositions.  Correct error in 2wu+st phase stability of katoetal_16.  Allow W_mppv-fppv to vary because fppv was becoming more stable than wu+st.  Change ppv=mw partitioning equations to involve wuls, rather than wu.  Assume that RTlnK is independent of temperature.

663) 141016.  Shift mgpv=mppv transition pressure to 130 GPa at 2000 K  and Clapeyron slope 6.7 MPa/K after catallietal_09.  Set W_mppv-fppv=0.

664) 151016.  Allow W_mppv-fppv to vary.  Include P<90 GPa stability points of tatenoetal_07 at X_Fe=0.5

665) 161016.  Add ismailovaetal_16 points on pure Fe-silicate perovskite stability.  Allow W_fepv-hepv and W_fepv-hlpv to vary.  Also allow W_fppv-hppv to vary, although this was not needed, i.e. the value remained at its initial, small, nominal value of 1.0 kJ/mol.

666) 171016.  Use alternative composition of ismailovaetal_16 new phase by matching Fe3+/Fe2+ ratio, rather than Fe3+/Si ratio.  Set W_fppv-hppv=0.

667) 181016.  Assume ismailovaetal_16 phase is composed of fepv and hlpv (no hepv at any pressure).  

668) 191016.  Set W_mgpv-hlpv=50.  Change hepv=hlpv points to include mgpv solid solution.

669) 201016.  Add lppv.

670) 211016.  Remove bykovaetal_16 hppv stability points that disagree with assumed hppv=lppv transition.  Correct error
in treatment of lppv solid solution in readin: previously assumed hppv-lppv solid solution on both sites, instead of only 
one site.

671) 221016.  Fix gamma wu and wuls by thermal expansivity from wicksetal_15 Fig. 8 (1500 K).

672) 231016.  Back up: return to gamma,q of wu,wuls equal to those of pe.  Reationalize hlpv eos: K, K' same as hepv, V0 such that volume of zeta phase matched at 66 GPa.  Correct nakajimaetal_12 data for Fe3+ in perovskite according to fit of 211016 results as shown in Ferricpv.pxp

673) 241016.  Alter properties of fepv: K_0=K_0(mgpv)+20 GPa.  Adjust gamma so that highest pressure pv=mw partitioning datum of auzendeetal_08 is matched.  

674) 251016.  Return gamma of fepv to previous value.  Change Kop of fepv to match K(80 GPa) from dorfmanduffy_14.

675) 261016.  Replace pv-mw partitioning data with Star and one higher temperature point from nakajimaetal_12.  Tune fepv content so that Kapp matches the nakajimaetal_12 values.

676) 271016.  Change V_0(fepv)=25.2 to fit highest pressure auzendeetal_08 pv-mw partitioning point.

678) 281016.  Allow W_mgpv-fepv to vary.  Add fepv=wu+st 43 GPa equilibrium of tangeetal_09a as a constraint.

679) 291016.  Back up.  Set W_fepv-hepv=W_fepv-hlpv=0.  Remove ismailovaetal_16 points.

680) 301016.  Allow W_fepv-hepv to vary.  Result is that the value varies little from zero.

681) 311016.  Allow W_fepv-hepv and W_fepv-hlpv to vary.  Incude ismailovaetal_16 points.

682) 011116.  Allow W_mgpv-hlpv to vary.  Include hs-ls transition in pure Fe2O3 from bykovaetal_16.

683) 021116.  Back up.  Remove ismailovaetal_16 points.  Set W_fepv-hepv=W_fepv-hlpv=0.

684) 031116.  Set hppv-lppv transition to 200 GPa.  This may be artificially high but reflects the fact that bykovaetal_16 data show no evidence of hs-ls transition.  Also, dorfmanduffy_14 data seem to 
preclude hs-ls transition over a wide range of Mg-Fe join for pressures less than ~170 GPa.

685) 041116.  Include all of tatenoetal_07 50 mgpv + 50 fepv stability data.

686) 051116.  Set V0(fepv)=24.953 after kieferetal_02 in order to better match the composition dependence of V(pv) seen in tangeetal_09.

687) 061116.  Set V0(fepv)=25.1 as a compromise, becasue 051116 produced K_pv-mw that seemed to vary too much with pressure.

688) 071116.  Correct errors in anao and hmag formulae.

689) 081116.  Allow W_fepv-hepv and W_fepv-hlpv to vary.  Incude ismailovaetal_16 points.

690) 091116.  Remove ismailovaetal_16 points.  Set W_fepv-hepv=W_fepv-hlpv=1.0 (i.e. small nominal value).  Add pyrope composition pv=ppv stability from dorfmanetal_12a.  Allow W_mppv-appv to vary.

691) 101116.  Set W_fepv-alpv and W_fppv-appv to match the oxides=pv+oxides transition and the pv=ppv coexistence in almandine composition from dorfmanduffy_12a.  These are fixed values.

692) 111116.  Set W_fepv-hepv=W_fepv-hlpv=W_fepv-alpv and W_mppv-hppv=W_fppv-hppv=W_mppv-appv.  This to better match dorfmanetal_13a observation of fs74 pure pv for an oxidized composition.

693) 121116.  Set W_fepv-hepv=0 because a) 111116 produced V_0(pv) vs. X_Fe that disagreed with expeirment and b) the dorfmanetal_13a observation  occurs at a pressure where the hepv component is much smaller than the hlpv component.

694) 131116.  Rmeove mixing on the second site in hepv and hlpv.  Result seems wrong, i.e. possible coding error as the final RMS is very large.  The problem could also be due to neglecting HS-LS mixing in pv, which occurs on the second site.

695) 141116.  Remove mixing on the first site of hepv and hlpv (i.e. the large site).  This retains HS-LS mixing in pv which occurs on the second, small site. 

696) 151116.  Play around with fapv parameter by hand based on ferric iron vs. Al coordination from H1495b experiments and lauterbachetal_00.  Choose V_0 and theta_0 of fapv so that the relation is nearly
independent of P and T.  Choose F_0 such that correlation is more or less matched.  Set W_fapv-x=0 since changing these values seemed to have little effect on the results independent of a tradeoff with F_0.

670) 310516_V.  Same as 310516 except that volume of mixing is no longer hard wired into the code.  Instead it is read in from the phase files.

671) 010119.  Same as 151116.  But with inversion code improved to include vreg(:,:,:,:) logic and hardwired values of vreg removed.

672) 020119.  Add flpv.  Add dorfmanetal_15 points to control HS-LS transition, its width, and its Mg-Fe dependence.

673) 030119.  Allow W_fepv-flpv and W_mgpv-flpv to be non-zero and to vary.

674) 040119.  Change V_0(fepv)=25.321 (dorfmanetal_13a).

675) 050119.  Remove lauterbachetal_00 constrain on ferric iron in pv (Run 1996).  Add 2000 K points from xuetal_15 Fig. 6 (Eq. 9).

676) 060119.  Add 4000 K points from xutetal_15 Fig. 6 (Eq. 9).

677) 070119.  Add artificial iron rich point to limit ferric iron in pv at large total iron, based on the "diamond" trend in nakajimaetal_12 Fig. 7.  Allow W_fepv-hepv to vary.  Note that dorfmanetal_15 finds low ferric iron in her samples.  But these were synthesized at ~80 GPa where hlpv is stable and where ferric is almost absent anyway because flpv is partly stable.

678) 080119.  Start over from 040119.  Restore lauterbachetal_00 (Run 1996) point.  This was incorrectly entered (2x too much hepv).  Correct this error.  Now pv3vsx goes through the constraining point (red line in upper figure going through higher ferrir content red point from Run 1996.

679) 090119.  Remove lauterbachetal_00 (Run 1996) point.  Add lauterbachetal_00 (Run 1579) point.

680) 100119.  Add hepv-hlpv (+mgpv) from liuetal_18.  Remove hepv=hlpv points from bykovaetal and catallietal.  Retain tsuchiyawang_13 point for Clapeyron slope.  Also remove one bykovaetal_16 hppv point from the "theta" stability field that had been mistakenly retained while all other of these points had been removed.

681) 110119.  Change flpv based on my own PBEsol calculations.  Change hepv based on liuetal_18.

682) 120119.  Change V_0 (alpv) according to liuetal_17.  The immediate motivation for this is that the volume of alm100 perovskite is much too small as compared with the 
data of duorfmanduffy_14 (Fig. 1).  Note that liuetal_17 confirm that they have perfect CCM substitution (no vacancies).  Derive K_0 (alpv) from ideal solution theory and adopting the 
dK/dx slope near the MgSiO3 end-member from walteretal_04.

683) 130119.  Add three almandine-pv stability points from dorfmanetal_12a.

684) 140119.  Allow W_fepv-alpv to vary.

685) 150119.  Add co+wu+st stability from akaogietal_98.  This is important because, although al and co+wu+st stability is already fixed, alm100 perovskite is becoming much too stable and eliminating the co+wu+st stability field.

686) 160119.  Change V_flpv=V_mgpv

687) 170119.  Return to V_flpv as given by bengtsonetal

688) 180119.  Add an artificial high temperature fepv-flpv point to control theta of flpv.  Also, fix flpv!  I had mistakenly included magnetic entropy!  This is now zeroed out.

689) 190119.  Change V_flpv=V_mgpv

690) 200119.  Change V_flpv to my PBEsol result.

691) 210119.  Add almandine composition oxides stability from conrad_98.  Remove automatic assignment of W_fepv-hepv and W_fepv-hlpv.  Change V_flpv=V_mgpv.

692) 220119.  Remove fepv-alpv almandine composition stability points from dorfmanetal_12a because Fe is significantly low spin at these conditions.

693) 230119.  Remove conrad_98 almandine composition oxide stability points.  Replace with dorfmanetal_12a.

694) 240119.  Change K_flpv=310.  This leads to better agreement with rho(X_Fe) at 80 GPa and 125 GPa from dorfmanduffy_14.  bengtsonetal_08 also find K_flpv>K_fepv.  out file not saved.

695) 250119.  Add artificial fepv stability point at 75 GPa, 2000 K.  Change nakajimaetal_12 high-T pv-mw partitioning point to the values that actually appear in Table 1 (rather than correcting them to bulk X_Fe=0.10).

696) 260119.  Add artificial flpv stability point at 75 GPa, 2000 K. 

697) 270119.  Add artificial fepv+flpv stability point at 75 GPa, 2000 K.  

698) 280119.  Set flpv to my PBEsol result.  Change K_flpv=K_fepv+20.  Remove artificial fepv(+flpv) stability point.  

699) 290119.  Set high-T fepv=flpv transition to 210 GPa (was 110 GPa).

700) 300119.  Replace P=43 GPa tangeetal_09a pv=mw+st compositions with P=22 GPa points.  Return high-T fepv=flpv transition to 110 GPa.

701) 310119.  Set W_mgpv-fepv=0.

702) 010219.  To code: add write statement for non-ideal volume of mixing.  These were not included in previous results going back to inv010119, which did include them.

703) 020219.  Remove all mgpv-fepv stability points from tatenoetal and dorfmanetal.  Change V_flpv so that V_fepv-V_flpv matches that of my PBEsol results (rather than matching V_fepv-V_mgpv).

704) 030219.  Adjust LS/HS fractions of dorfmanetal_15 data to better reflect results as shown in the ms, including flpv=0 at 0 GPa.

705) 040219.  Reset low pressure flpv fraction to minimal finite value (1/1000) rather than 0, which causes instability.

706) 050219.  Set flpv=1/1000 at 40 GPa for X_Fe=0.74.  

707) 060219.  Set W_mppv-fppv=0.  set flpv=1/1000 at 0 GPa for X_Fe=0.74.

708) 070219.  Reset V_0, K_0, K_0', gamma, q constraints of mppv to 270914 values.

709) 080219.  Return V_0, K_0, K_0', gamma, q of mppv and fppv to 060219 values.  Adjust V_0 of mppv and fppv to match P=125 GPa results of dorfmanduffy_14.

710) 090219.  Go back to tatenoetal_09 and hiroseetal_06b constraints on mgpv=mppv transition.

711) 100219.  Use only tatenoetal_09 constraints on mgpv=mppv transition as per SLB11.

712) 110219.  It appears from Fig. 13 in SLB11 that SLB11 used tatenoetal_09 and hiroseetal_06b constraints on mgpv=mppv.  Therefore use both.  The hiroseetal_06b points are useful in constraining the position of the transition at temperatures ~ 2000 K, particularly the mppv stability observations.  in inv090219 these points were not fit very well.  It appears that there is a conflict with mgpv+alpv, mppv+appv stability of dorfmanetal_12a.  Remove these two dorfmanetal_12a points.

713) 120219.  Add wu+apbo stability at 149 GPa and 2000 K.  Allow W_mppv-fppv to vary.

714) 130219.  Add wu+st stability at 117 GPa and 2000 K from fujinoetal_09.  

715) 140219.  Go back to guignotetal_07 eos for mppv.  Choose V_fppv so that dorfman 125 GPa data will be approximately matched.  Note that the stackhouse and brodholt constriants on dK/dT and dG/dT are in compatible with the guignotetal_07 eos in that the value of q does not converge.

716) 150219.  Go back to sakaietal_16 eos.  The problem with guignotetal_07 is that it produces a bulk modulus of ppv that is much less than that of pv near D''.  Remove hiroseetal_06b mgpv/mppv stability points.  Restore dorfmanetla_12a points.  The idea is to produce an alternative with a higher pv=ppv transition.

717) 160219.  Change K_fepv=290.  Attempt to match dorfmanduffy_14 constraints on V,K at 80 GPa and 125 GPa including the effects of flpv.

718) 170219.  Change flpv equilibria:  allow for some low-spin Fe in Fe40 composition as permitted by uncertainties.

719) 180219.  Change pe eos.  Use Dorogokupets and Dewaele (2007) eos for K_0'.  Do not use Hugoniot data.

720) 190219.  Change K_fepv=270.

721) 200219.  Change K_flpv=270.

722) 210219. Remove two dorfmanetal_12 points on Mg-Al pv-ppv.

723) 220219.  Set W_V_co-il = -0.3325 to fit the volume data of liuetal_17.

724) 230219.  Revert to alpv eos as the one that approximates Rh2O3 (as in e.g. inv010119).

725) 240219.  Add constraints on pv=ppv in pyrope composition from tatenoetal_05

726) 250219.  Set W_V_co-il=-0.3325, W_V_mgpv-alpv = 1.3252 to fit the volume data of liuetal_17.

727) 260219.  Set W_V_co-il=0 and W_V_mgpv-alpv=0.  Replace kuboakaogi_00 constraint on mgpv-alpv-mgil-co equilibria with liuetal_17 point.

728) 270219.  Add fepv-alpv and fppv-appv almandine composition stability points.  Allow W_fppv-appv to vary.  Notice that W_fepv-alpv had been allowed to vary but hadn't changed because none of the constraints required it to.

729) 280219.  Remove fepv-alpv and fppv-appv almandine composition stability points.  Set W_fppv-appv=W_fepv-alpv=W_feil-co=0.

730) 010319.  Restore fepv-alpv and fppv-appv stabilities points.  Allow W_fepv-alpv and W_fepv-alpv to vary.  Set W_feil-co=0.

731) 020319.  Add 100 kJ/mol to F_0(flpv) to make it very unstable.  Add artifiical equilibria to fix this.

732) 030319.  Remove fppv-appv stability point.

733) 040319.  Add alpv+fepv and appv+fppv al composition stability, both at 176 GPa.

734) 050319.  Remove fppv+appv stability point.  Remove felpv+alpv 176 GPa stability point.  Add fepv+alpv=fppv+appv reaction at 176 GPa.

735) 060319.  Fix W_mppv-fppv=-16  The value in 050319 became very negative which distorts the topology of the pv-ppv transition on the en-fs join.

736) 070319.  Simplify constraints.  fepv+alpv only at 80 GPa.  co+wu+st from dorfmanetal_12a only at 35 GPa.  fepv+alpv=fppv+appv only for the Al species. 

737) 080319.  Changed mppv+wuls=fppv+pe constraint so that Theta_0 of fppv is set to Theta(fepv)*Theta(mppv)/Theta(mgpv).  Remove all fepv+alpv and fppv+appv stability points.  Do not allow
W_fepv-alpv=0 or W_fppv-appv=0 to vary.  Set W_co-feil=63 and do not allow to vary.

738) 100319.  Allow W_mgpv-fepv to vary.

739) 110319.  Adjust elasticity of mppv.  Use zhangetal_16 for dG/dT and dK_S/dT.  Use tsuchiyaetal_04b for G'.  Adjuse G so that VS(mppv)/VS(mgpv)=1.037 at 1500 K, 125 GPa as per zhangetal_16.  Also remove sakaietal_16 constraint from exp/kgp/data because it is redundant with sakaietal_16 constraint contained in exp/comp/data.  Also adjust G' of mgpv.  Get this from murakamietal_07a (rather than wentzcovitch which I had been using).

740) 120319.  Set W_mgpv-fepv=0 and fix.  Set W_mppv-fppv=-21 and fix.

741) 130319.  Allow W_mppv-fppv to vary.  Set W_mppv-appv=W_fppv-appv=35.

742) 140319.  Adjust eos of wu and wuls so that they are again consistent with the data of chenetal_12 as parameterized by solomatovaetal_16.  This change is necessary because of the change in pe eos first made at 180219.

743) 150319.  Set W_mppv-fppv=0.

744) 160319.  Allow W_mppv-fppv to vary.  Set Gamma_ri-mgpv+pe=-3 after yeetal_17.

745) 010220.  Allow W_mgpv-fepv to vary.

746) 010720.  Remove all co+wu+st stability points except for the dorfmanetal point at 35 GPa.  This is necessary to recover a good fit to gt-ol iron partitioning.

747) 020720.  Fix error in 306 6 di + 475 3 namj + 158 5 py = 694 6 jd + 317 2 gr + 50 6 mgmj data: I previously had all px on one side and all gt on the other.  The correct equation is derived from the cation exchange: di + gt(jd) = jd + gt(di) where gt(jd)=1/2namj-1/2mgmj+1/2py and gt(di)=1/3gt-1/3py+1/2mgmj

748) 030720.  Replace di-jd-namj-mgmj-gr-py experimental data with two derived from piercing points in Fig. 3 of gasparik_96.

749) 040720.  Remove mgil=mgmj on the en-jd join point from gasparik_92.  Add mgc2=mgmi on the en-jd point from gasparik_92.  Remove px=gt points on di-jd join.  Remove namj=nacf on the en-jd join points.  Set W_namj-gr=W_namj-py=0. (Note that because of logic in redexp these two parameters were adjusted, but came to very small values).  Add px and gt stability points for namj+py composition from dymshitsetal_13.

750) 050720. Change px and gt stability points for those of gasparik_92, i.e. the transition at 13.5 GPa at 1773 K and 2373 K.  This produces a much more reasonable Debye temperature for namj than the 040720.

751) 060720. Set W_namj-py=W_namj-al=W_namj-gr=W_namj-mgmj from 050720.  Fix these W's, except W_namj-mgmj.  W_namj-mgmj changed

752) 070720. Set W_namj-x=W_mgmj-x, except for W_namj-mgmj which is variable.

753) 080720.  Reset W_ab-an=13 kJ/mol after hollandpowell_03 who have W_ab-an=15+-3 kJ/mol (Eq. 10).

754) 010121.  Same as 080720 but with errors on the parameters calculated and the following automatic changes to W parameters that should have been implemented in 080720 but were not.  I have now also added the corresponding lines to writeout.f.

W_py-namj=W_py-mgmj
W_al-namj=W_al-mgmj
W_gr-namj=W_gr-mgmj
W_mppv-appv=W_mgpv-alpv
W_fppv-appv=W_mppv-appv
W_mppv-hppv=W_mppv-appv
W_fppv-hppv=W_mppv-hppv
