Water dimer binding energy — DFT vs. quantum Monte Carlo¶
System(s): Polaris · Code: Quantum ESPRESSO + QMCPACK · Outcome: Partial
The ask¶
(This campaign predates verbatim prompt logging — the request below is reconstructed from the campaign's documented design, not a direct quote.)
"Compute the binding energy of a water dimer with DFT, then refine it with quantum Monte Carlo for higher accuracy."
What happened¶
Trinity computed the water dimer (two water molecules) binding energy via DFT successfully, then attempted to hand the result off to QMCPACK for a more accurate quantum Monte Carlo treatment. The Monte Carlo step was blocked by a technical incompatibility in how the k-point sampling was specified between the two codes — a real, disclosed limitation rather than a silent failure. Trinity correctly recognized and reported this as a partial success rather than claiming full completion.
Results¶
- DFT-PBE binding energy: -0.067 eV, compared to a high-accuracy reference value of -0.215 eV — DFT-PBE is known to underestimate hydrogen-bond energies at this basis-set quality, consistent with expectations.
- Quantum Monte Carlo refinement did not complete, due to a k-point sampling incompatibility between the two codes — flagged as a fixable configuration issue for a future run, not a dead end.