I noticed that setting mf.xc = "wb97x-d4" works in PySCF without raising an error. However, it seems the code splits the xc functional and the D4 dispersion correction into two independent choices. The confusion stems from the fact that both the xc functional and the D4 dispersion correction have more than one common definition:

Base XC functional (wB97X part):

• Original wB97X by Chai and Head-Gordon (J. Chem. Phys. 128, 084106 (2008); doi:10.1063/1.2834918).
• Refitted wB97X-V (without VV10) by Mardirossian and Head-Gordon (Phys. Chem. Chem. Phys. 16, 9904 (2014); doi:10.1039/C3CP54374A).

D4 dispersion parameters:

• Parameters fitted for the refitted wB97X-V (no VV10) by Najibi and Goerigk (J. Comput. Chem. 41, 2562 (2020); doi:10.1002/jcc.26411).
• Parameters specifically optimized within the wB97X-3c composite method by Müller, Hansen, and Grimme (J. Chem. Phys. 158, 014103 (2023); doi:10.1063/5.0133026).

This creates four logical combinations, of which only two are physically meaningful:

Questions:

1. Which combination does PySCF actually execute when I write mf.xc = "wb97x-d4"?
   Does it silently map to the refitted wB97X-V (without VV10) functional, and if so, which of the two D4 parameter sets is applied? I need to know what I am truly computing so I can describe it properly in publications.

2. How can I explicitly select either of the two valid variants?
   If the default is, for example, the Goerigk‑type combination, how should I call the other one without guessing the internal alias?