In order to compare the calculated cohesive energy with experiment
various correction terms must be added to the solid calculations to
account for: (i) Coulomb finite-size effects; (ii) single-particle
finite-size effects; (iii) the use of a local pseudopotential; and
(iv) zero-point motion. These corrections are discussed in detail in
reference [33]. All these corrections are essentially
independent of the optimisation of the wavefunction and it therefore
suffices to compare directly the VMC and DMC results for the same
system. A DMC calculation for the germanium pseudo-atom gave an energy
of -103.42 
0.03 eV, which is only 0.20 eV below our best VMC
result, while for the solid the VMC calculation (without 
) gave
an energy 0.34 eV above the DMC result. Therefore the VMC cohesive
energy, of 3.80 ev per atom, is only 0.14 eV less than the DMC result,
of 3.85 eV per atom, which amounts to an error of only 4%. As
mentioned earlier we believe that 0.12 eV of the difference in energy
between the VMC and DMC results for the solid is due to the incomplete
basis set used for the single particle orbitals. If this is corrected
for, the VMC cohesive energy differs from the DMC value by only 1%.
This indicates that although there is still a significant difference
in the VMC and DMC energies, the difference in the atomic energies
transfers almost completely into the solid calculations and therefore
cancels out of the cohesive energies.