Information Dynamics Explains Chemical Reactions: A Unified Framework from NaI Photodissociation to H2 Formation

Chemical reactions are fundamentally the motion of atomic nuclei on potential energy surfaces. However, there is a methodological gap between quantum chemistry calculations and dynamical simulations: constructing potential energy surfaces is complicated, non-adiabatic couplings are difficult to treat, and different reaction types lack a unified language. In this work we introduce the information dynamics framework, reformulating chemical reactions as a three-element system: virtual space (potential energy surface rules), real space (initial wave packet data), and coupling matrix (projection mechanism). The virtual space defines the absolutely allowed region for nuclear motion, the real space provides the observed distribution of initial positions and momenta, and the coupling matrix drives the system toward a steady-state product according to the virtual space topology (gradient flow for a single surface or Landau‑Zener transitions for two crossing surfaces). We validate the framework with three classic reactions:(1) NaI photodissociation (non‑adiabatic, two crossing surfaces) — simulated dissociation yield 65.3% ± 1.5%, matching Zewail’s femtochemistry experiment (65%);(2) Cl_2 photodissociation (adiabatic, single repulsive surface) — dissociation yield 100%, fragment mean kinetic energy 3.84 eV, in agreement with literature values;(3) H_2 formation (attractive potential) — binding probability remains 1 when total energy is below the dissociation threshold and drops to 0 above it, in agreement with classical and quantum mechanical predictions. The three cases cover typical scenarios: non‑adiabatic/adiabatic, dissociation/formation, single/two potential energy surfaces, all quantitatively described by the same information dynamics framework. This demonstrates that information dynamics can unify not only sequence assembly problems in computational biology but also serves as a fundamental language for chemical reaction simulations. All simulation source code, input parameters, random seeds, and plotting scripts are included in this deposit to ensure full reproducibility of the results reported in the associated manuscript.