Predictive accuracy of ab initio photochemical quantum yields

For benchmark photoreactions (cyclobutanone photodissociation at 200 nm, azobenzene trans-cis isomerization, retinal isomerization in rhodopsin), predict product branching ratios and quantum yields to within 0.1 absolute, and ultrafast electron diffraction or photoelectron signals within experimental error, using electronic structure and nuclear dynamics without fitting to the target data. An answer is a protocol (electronic structure level plus dynamics method) shown to meet this accuracy in blind tests.

Blind prediction challenges in which simulations are submitted before ultrafast measurements are released.