Python API · stjames models · API examples
Transition-state optimization refines a nearby guess into a first-order saddle point: an energy maximum along the reaction motion and a minimum along the other motions. Use it to locate a proposed reaction barrier. Rowan returns the final geometry, its energy, and the optimization trajectory; frequency calculations help establish whether it is the intended transition state (TS).
TS optimization is sensitive to the initial guess and may find a different saddle point or fail to converge. Check convergence warnings before interpreting the result; a reported final geometry can still be unconverged.
A valid TS should have one significant imaginary vibrational mode. Inspect its motion to confirm that it follows the intended bond changes. Extra tiny imaginary frequencies can arise from numerical noise, but should be checked rather than ignored; no significant imaginary mode or several significant imaginary modes calls for further refinement.
Run an intrinsic reaction coordinate (IRC) in both directions to check which reactant and product structures the TS connects. Use the same level of theory for optimization, frequencies, and IRC so they describe the same energy surface. A single imaginary frequency alone does not establish the reaction mechanism.