How CASSCF Works¶

The Complete Active Space Self-Consistent Field (CASSCF) method is a multiconfigurational electronic structure method designed for systems where a single electronic configuration is insufficient to describe the molecular wavefunction.

Unlike RHF, which assumes that all electrons occupy a single electronic configuration, CASSCF allows electrons within a selected active space to redistribute among several molecular orbitals. This enables the method to accurately describe excited states, near-degenerate electronic states, bond-breaking processes, and photochemical reactions.

The overall workflow is illustrated below.

Geometry Optimization
        │
        ▼
   RHF Calculation
        │
        ▼
 Molecular Orbitals
        │
        ▼
  CIS Calculation
        │
        ▼
 Active Space Selection
        │
        ▼
     CASSCF
        │
        ▼
 Optimized Multiconfigurational
      Wavefunction

Step 1 – Start from RHF Molecular Orbitals¶

Every CASSCF calculation begins with molecular orbitals obtained from a previous RHF calculation.

RHF

↓

Canonical Molecular Orbitals

↓

Read using GUESS=MOREAD

These orbitals provide the initial guess for the CASSCF optimization.

Although RHF orbitals are a good starting point, they are optimized assuming that electrons occupy only a single electronic configuration. CASSCF will later refine these orbitals.

Step 2 – Select the Active Space¶

The most important decision in a CASSCF calculation is choosing the active space.

The molecular orbitals are divided into three groups.

Higher Energy

Secondary (Virtual)

────────────────────

Active Orbitals

────────────────────

Inactive (Core)

Lower Energy

Each group has a specific role.

Inactive Orbitals¶

Always doubly occupied.
Do not participate in electronic excitations.
Remain unchanged throughout the calculation.

Active Orbitals¶

Electrons may move freely among these orbitals.
Every allowed electronic configuration is considered.
These orbitals determine the multiconfigurational character of the wavefunction.

Secondary Orbitals¶

Remain unoccupied.
Do not contribute to the CI expansion.

The active space is usually selected using the results of a previous CIS calculation.

Step 3 – Generate Electronic Configurations¶

Once the active space has been defined, GAMESS generates every possible electronic configuration consistent with

the number of active electrons,
the molecular symmetry,
and the spin multiplicity.

Rather than using individual Slater determinants, GAMESS constructs Configuration State Functions (CSFs) using the Graphical Unitary Group Approach (GUGA).

Active Space

↓

Generate CSFs

↓

Spin-Adapted Wavefunctions

Each CSF represents one possible distribution of electrons among the active orbitals.

Step 4 – Solve the Configuration Interaction Problem¶

Each Configuration State Function contributes to the electronic wavefunction with a different weight.

GAMESS constructs the Hamiltonian matrix using all generated CSFs and solves the Configuration Interaction (CI) problem.

Configuration State Functions

↓

CI Hamiltonian

↓

Diagonalization

↓

CI Coefficients

The resulting CI coefficients determine how much each electronic configuration contributes to the final wavefunction.

Unlike CIS, which considers only single excitations, CASSCF includes all possible configurations within the chosen active space, providing a much more complete description of electron correlation.

Step 5 – Optimize the Molecular Orbitals¶

After determining the CI coefficients, GAMESS evaluates whether the molecular orbitals are optimal.

If they are not, the orbitals are rotated to lower the total electronic energy.

Current Orbitals

↓

Orbital Rotation

↓

Improved Orbitals

These updated orbitals change the Hamiltonian, which in turn changes the CI coefficients.

Because the orbitals and CI coefficients depend on one another, this optimization must be repeated until both become self-consistent.

Step 6 – Repeat Until Convergence¶

The optimization cycle continues until the total energy and orbital gradients satisfy the convergence criteria.

Optimize CI Coefficients
          │
          ▼
Optimize Molecular Orbitals
          │
          ▼
Compute New Energy
          │
          ▼
Converged?
      │
 ┌────┴────┐
 │         │
 No       Yes
 │         │
 └────────►Final CASSCF Wavefunction

At convergence, both the molecular orbitals and the electronic wavefunction are fully optimized.

Step 7 – Obtain the Final Wavefunction¶

The final CASSCF wavefunction is a linear combination of Configuration State Functions.

Unlike RHF,

Wavefunction

↓

One Electronic Configuration

CASSCF produces

Configuration 1

+

Configuration 2

+

Configuration 3

+

...

↓

Final Multiconfigurational Wavefunction

This allows the method to accurately describe systems where several electronic configurations contribute significantly to the electronic structure.

Why is CASSCF More Accurate?¶

Many molecules exhibit static electron correlation, where two or more electronic configurations are nearly equal in energy.

Examples include

excited electronic states,
bond dissociation,
conical intersections,
diradicals,
transition-metal complexes,
photochromic molecules.

Because RHF assumes a single electronic configuration, it cannot accurately describe these systems.

CASSCF overcomes this limitation by explicitly including multiple electronic configurations within the active space.

Comparison with RHF and CIS¶

Feature	RHF	CIS	CASSCF
Wavefunction	Single determinant	Excited-state CI from RHF	Multiconfigurational
Orbital Optimization	Yes	No	Yes
Configuration Interaction	No	Single excitations only	Complete within active space
Excited States	No	Yes	Yes
Static Correlation	No	No	Yes
Suitable for Photochemistry	Limited	Moderate	Excellent

Advantages of CASSCF¶

Provides a balanced description of ground and excited states.
Accurately treats near-degenerate electronic states.
Describes bond breaking and bond formation.
Captures static electron correlation.
Forms the reference wavefunction for advanced methods such as XMCQDPT and CASPT2.
Widely used for photochemistry and nonadiabatic dynamics.

Limitations of CASSCF¶

Although CASSCF is a powerful method, it has several important limitations.

The quality of the calculation depends strongly on the chosen active space.
Computational cost increases rapidly with active-space size.
Dynamic electron correlation is not fully included.
Large systems may require subsequent perturbation methods such as XMCQDPT.

Key Takeaways¶

CASSCF begins with molecular orbitals from a previous RHF calculation.
An active space is selected using chemically important orbitals, often identified through a CIS calculation.
GAMESS generates all allowed Configuration State Functions within the active space.
The CI coefficients and molecular orbitals are optimized simultaneously.
The optimization continues until both the energy and orbitals become self-consistent.
The final wavefunction contains multiple electronic configurations, enabling CASSCF to describe systems beyond the capabilities of RHF and CIS.