/
(wip) theory benchmark planning

(wip) theory benchmark planning

TODO

TODO

1. Submission of constrained optimization w/ (1) MP2/heavy-aug-cc-pVTZ for reference optimized geometries (2) B3LYP-NL/def2-TZVPD (3) wb97x-v/def2-TZVPD
2. search/add more pharma-relevant anions/cations(O-, cyclo-penta-dienyl, deprotonated sulfonamide? etc.)
3. Submission of torsiondrive dataset w/ B3LYP-D3BJ/6-311+g**

 

  • Reference: torsiondrive w/ MP2/6-311+G** B3LYP-D3BJ/def2-TZVPD(calculation done)→ then optimize each frame with MP2/heavy-aug-cc-pVTZ → single point calculation with CCSD(T)/CBS (The CCSD(T) energy with the aug-cc-pvdz basis set is extrapolated to the CBS limit using the MP2/aug-cc-pv[tq]z level of theory)

  •  

  1. accuracy in optimized geometries

    1. method: (1) optimization of each frame in the reference torsion scan with varying combination of methods and basis sets. Then compare RMSD with the reference or (2) run entire torsiondrive for each combination

    2. effect of varying electronic structure theory (b3lyp/b3lyp-d3bj/ b3lyp-nl/ wb97x-v)

      1. list of fuctionals are from conformational energies benchmarking paper. (they were used with def2-qzvp)

      2. basis set: def2-TZVPD? (too expensive?)

    3. effect of varying basis sets: def2 family(DZVP, def2-TZVP, def2-TZVPP, def2-TZVPD, def2-TZVPPD, def2-QZVP, def2-QZVPD), 6-31+g**, 6-311+g**

      1. main purpose: effect of including diffuse function

      2. functional: b3lyp-d3bj

  2. accuracy in energy

    1. method: single point energy calculation at the reference frames and compare the deltadelta energy

    2. effect of varying electronic structure theory (b3lyp-d3bj/ b3lyp-nl/ wb97x-v)

      1. list of fuctionals are from conformational energies benchmarking paper. (they were used with def2-qzvp)

      2. basis set: def2-qzvp

    3. effect of varying basis sets: def2 family(DZVP, def2-TZVP, def2-TZVPP, def2-TZVPD, def2-TZVPPD, def2-QZVP, def2-QZVPD), 6-31+g**, 6-311+g**

      1. functional: b3lyp-d3bj

 

  • Psi4 performance benchmark for a single gradient calculation of FGG114 on i7-6850K 1 core

    • b3lyp-d3bj/6-311+g** (603 s), b3lyp-d3bj/6-31+g**(456 s)

    • mp2/cc-pvtz(2372 s), mp2/heavy-aug-cc-pvtz(10934 s), mp2/6-311+g**(1211 s)

 

Related content

Using RMSD in TorsionProfile fitting
Using RMSD in TorsionProfile fitting
More like this
Theory Benchmarking Set Generation Scheme
Theory Benchmarking Set Generation Scheme
More like this
Fitting torsions to small molecules
Fitting torsions to small molecules
More like this
Report 2 (2021-04-08)
Report 2 (2021-04-08)
More like this
TYK2 Refit Validation
TYK2 Refit Validation
More like this
2022-01-27 Protein FF meeting notes
2022-01-27 Protein FF meeting notes
More like this