Some modification after the USPEX 17th workshop, as reported by Dr. Marchal, who performed the GDIS installation in Rennes, France*; and the workshop participants.
Two important changes:
Modification contains a FIX to an openGL issue with software rendering, ill-configure system, or gtkgl flavour that lacks the GDK_GL_MODE_ACCUM capability. The issue was an improper display of the pango text (ie. it only appear briefly on every refresh event), due to a race condition with other openGL draws.
I switch the pango_print function to only use cairo function (the internal GTK drawing function). Unfortunately cairo is not so well integrated with GTK+-2.0, so should the text not being displayed properly, please let me know.
The convex hull COMP graphs for USPEX variable composition calculations was also modified to cope with the ones calculated by USPEX.
I notice however a tiny discrepancy (within the meV order, possibly due to truncation in Individuals file) in the determination of minimum structures for simple elements (ie An or Bm in the following equation). It is of no visible consequence in the display of convex hull, and reported stable composition are correct.
I recommend to create a chem.in file containing reference energies calculated with a good, strong confidence, to eliminate the problem of USPEX internal reference.
The energy is now calculated, for a AxBy structure, as
Ef = E[AxBy] - ( x/n E[An] + y/m E[Bm] ) / ( x + y )
Some other requests have been voiced, but I will need a little longer to investigate ;)
*Dr. Marchal also reported the incorrect behaviour that I fixed in the pull #26
Some modification after the USPEX 17th workshop, as reported by Dr. Marchal, who performed the GDIS installation in Rennes, France*; and the workshop participants.
Two important changes:
Modification contains a FIX to an openGL issue with software rendering, ill-configure system, or gtkgl flavour that lacks the GDK_GL_MODE_ACCUM capability. The issue was an improper display of the pango text (ie. it only appear briefly on every refresh event), due to a race condition with other openGL draws. I switch the
pango_printfunction to only usecairofunction (the internal GTK drawing function). Unfortunatelycairois not so well integrated with GTK+-2.0, so should the text not being displayed properly, please let me know.The convex hull COMP graphs for USPEX variable composition calculations was also modified to cope with the ones calculated by USPEX. I notice however a tiny discrepancy (within the meV order, possibly due to truncation in Individuals file) in the determination of minimum structures for simple elements (ie An or Bm in the following equation). It is of no visible consequence in the display of convex hull, and reported stable composition are correct. I recommend to create a chem.in file containing reference energies calculated with a good, strong confidence, to eliminate the problem of USPEX internal reference. The energy is now calculated, for a AxBy structure, as
Some other requests have been voiced, but I will need a little longer to investigate ;)
*Dr. Marchal also reported the incorrect behaviour that I fixed in the pull #26