pMELTS bug in Perple_X update log does not apply here

[Paula]

Users of the Perple_X software of James Connolly may have spotted the following message in a recent update log (thanks to Peter Luffi for drawing it our attention):

Jul 12, 2007 pMELTS model reformulation: The implementation of the pMELTS(G) model (Ghiorso et al., G3 3:5, 2002) in Perple_X followed the formulation (equations 1 and 2) of Ghiorso et al (2002) verbatim. However, consideration of the basis of the model outlined in Nichols (CMP, 74:211, 1980) and the subsequent, and supposedly identical, MELTS model description (Ghiorso & Sack, CMP:197, 1995) suggests that equations 1 and 2 of Ghiorso et al. are incorrect. Accordingly, Perple_X '07 has been modified to use the earlier formulation of Nicholls (1980) and Ghiorso & Sack (1995). This modification reduces the excessive stability field for hydrous silicate melts predicted by earlier Perple_X implementations of pMELTS.

Adiabat_1ph differs from other implementations of the MELTS and pMELTS algorithms, such as Perple_X or Alan Boudreau's Pele (a Windows GUI), in that it is based on actual (p)MELTS code of Mark Ghiorso and not reprogramming of published algorithms. Specifically, it incorporates the backend to the Java MELTS version; of course, we are very grateful to Mark for allowing us to use and develop his code in this way.  (For details of a minor way in which the Java/CORBA MELTS versions differ from Mark's standalone MELTS programs please see this thread.)

So, for anyone who is concerned that the bug described above may be present when running Adiabat_1ph in pMELTS mode, please rest assured that it is not. There is a mistake in the printed equations in Ghiorso et al (2002) where the non-regular solution contribution of water is counted twice. However, both the MELTS and pMELTS models use the same expressions (the end-member properties and interaction parameters are different) and the code is definitely correct; the relevant water terms appear only once. The details are in Appendix 3 of Ghiorso et al (1983; CMP v84, p107-145) which describes the extent to which the model follows that of Nicholls (1980).

Paula