How do I accelerate a geometry optimization using information from a lower (cheaper) level of theory, and does this really help?
It can help a lot and is especially worth doing for most large basis set calculations with correlated wavefunctions. The geometry and Hessian information from a previous optimization are used by default — if you saved them. You should keep all of the files that NWChem puts into its permanent directory. • Set the permanent directory to be somewhere permanent (sic). The default is the current directory, which for a batch job on the EMSL HP is /scratch. If you plan on running both optimizations in the same input then you don’t really need to do this, but if anything goes wrong you can only restart if you have saved the files. • Run the first optimization with a low-level theory. • In the same job, or a subsequent one, specify the new wavefunction parameters and run the second optimization. By default the calculation will restart from the previously converged geometry, but if you can estimate better values you can specify a new geometry (e.g., MP2 often predicts longer bond lengths than
Related Questions
- How do I accelerate a geometry optimization using information from a lower (cheaper) level of theory, and does this really help?
- I have high levels of malic acid in my juice. What yeast strain can I use to help lower the acid level?
- How can making the right food choices help lower my blood cholesterol level?
