How do static and dynamic correlation differ?

Dynamic correlation arises from the instantaneous avoidance of electrons and is captured by single-reference correlated methods, while static correlation reflects near-degeneracy of several configurations and requires a multireference treatment.

Why is choosing the active space difficult?

The active orbitals must capture the chemistry of interest while remaining small enough to compute; too small a space misses important physics, and too large a space becomes intractable.

Multireference and Multiconfigurational Methods

When a single determinant cannot describe a molecule, multireference methods build the wavefunction from several configurations, capturing the static correlation that defies standard approaches.

Definition

Quantum-chemical methods that represent the wavefunction as a combination of several important configurations rather than a single reference determinant, in order to treat strong electron correlation.

Scope

Covers situations of strong (static) correlation such as bond dissociation, biradicals, transition metals, and excited states; the complete-active-space self-consistent-field (CASSCF) construction; dynamic-correlation corrections such as CASPT2 and multireference configuration interaction; and the challenge of selecting an active space.

Core questions

What is static correlation and when does the single-reference picture break down?
How does the complete-active-space approach select and optimize the important configurations?
How is the remaining dynamic correlation added on top of a multireference reference?
How is a chemically meaningful active space chosen?

Key theories

Complete active space self-consistent field: Performs a full configuration interaction within a chosen set of active orbitals while optimizing the orbitals, providing a balanced multiconfigurational reference for strongly correlated systems.
Multireference dynamic correlation: Adds the remaining dynamic correlation to a multiconfigurational reference, for example through second-order perturbation theory (CASPT2) or configuration interaction, to reach quantitative accuracy.

Clinical relevance

Multireference methods are essential for bond-breaking reactions, excited-state and photochemical processes, and many transition-metal and lanthanide systems where single-reference methods give qualitatively wrong results.

History

Building on early configuration-interaction work, Roos and coworkers introduced the complete-active-space SCF method in 1980; CASPT2 and efficient multireference configuration interaction followed, making strongly correlated and excited-state chemistry tractable.

Debates

Active-space selection and black-box automation: Choosing the active space has traditionally required expert judgment and strongly affects results; whether reliable automated selection is achievable remains an active methodological question.

Key figures

Björn Roos
Per Siegbahn
Hans-Joachim Werner
Isaiah Shavitt

Seminal works

roos1980
szalay2012

Frequently asked questions

How do static and dynamic correlation differ?: Dynamic correlation arises from the instantaneous avoidance of electrons and is captured by single-reference correlated methods, while static correlation reflects near-degeneracy of several configurations and requires a multireference treatment.
Why is choosing the active space difficult?: The active orbitals must capture the chemistry of interest while remaining small enough to compute; too small a space misses important physics, and too large a space becomes intractable.