Degenerate coupled-cluster theory

This paper introduces a novel, size-extensive, and convergent degenerate coupled-cluster ($\Delta$CC) theory capable of handling arbitrary reference states and a related quasidegenerate variant (QCC) for strong correlation, demonstrating superior accuracy and convergence over existing methods like EOM-CC, CI, and MBGF through high-order implementations.

The Gist

Here is an explanation of the paper "Degenerate Coupled-Cluster Theory" using simple language and creative analogies.

The Big Picture: Fixing the "One-Size-Fits-All" Problem

Imagine you are trying to predict the weather. For a calm, sunny day (a non-degenerate state), you have a perfect, highly accurate model that works like a charm. This is what standard Coupled-Cluster (CC) theory does for atoms and molecules in their ground state. It's the "gold standard" of chemistry, known for being a "black box": you put in the atoms, and it spits out the exact energy and behavior without needing a human expert to tweak the settings.

But what happens when the weather is chaotic? What if there are two storms happening at once, or the wind is blowing from two directions with equal force? In quantum chemistry, this is called degeneracy. It happens when electrons are so confused or "stuck" between different arrangements that no single description works.

Standard CC theory breaks down here. It's like trying to use a weather model designed for a sunny day to predict a hurricane; the math just doesn't hold up.

This paper introduces a new method called $\Delta$CC (Delta-CC). Think of it as upgrading that weather model to handle any condition, whether it's a calm day, a chaotic storm, or a mix of both. It works for excited states, ionized atoms (where an electron is missing), and even weird high-energy states, all with the same "black box" ease.

---

The Core Idea: The "Team of Experts" vs. The "Solo Artist"

To understand the innovation, let's look at how the old and new methods handle a tricky situation.

The Old Way (EOM-CC): The Solo Artist with a Script

The current popular method for excited states is called EOM-CC (Equation-of-Motion Coupled-Cluster).

• The Analogy: Imagine you have a master chef (the ground state) who knows how to cook a perfect steak. To make a dessert (an excited state), you ask the chef to "add a little sugar" or "change the garnish."
• The Problem: This works great if the dessert is just a slightly different steak. But if you want to make a completely different dish, like a soup, asking the steak chef to "add a little water" doesn't work well. The chef is stuck thinking in terms of steaks.
• The Limit: EOM-CC is great for simple changes (one electron moving), but it struggles when the change is massive (two or more electrons moving at once) or when the starting point is already messy.

The New Way ($\Delta$CC): The Flexible Team

The new $\Delta$CC theory changes the approach entirely.

• The Analogy: Instead of asking a single chef to modify a dish, $\Delta$CC says, "Let's hire a whole team of chefs, each specializing in a different starting ingredient."
• How it works: If an atom has electrons that are "degenerate" (confused between two spots), $\Delta$CC doesn't force them into one spot. It acknowledges all the possible starting spots simultaneously. It treats every possible arrangement of electrons as a valid starting point and lets the math figure out the perfect mix.
• The Result: It's a "universal" method. Whether you are looking at a calm ground state, a high-energy excited state, or an atom that has lost an electron, the same mathematical engine runs. You don't need to change the code or the recipe; you just change the input.

---

Why "Black Box" Matters

In science, a "black box" is a tool where you don't need to know the internal gears to get a result. You just put data in, and it works.

• Current Expert Tools: Many advanced methods for complex molecules require a human expert to say, "I think these three electrons are the problem, let's focus on them." This is like needing a mechanic to manually adjust the carburetor every time you drive a car. It's not scalable.
• $\Delta$CC's Promise: This new theory is a true black box. You feed it the molecule, and it handles the complexity automatically. It's "systematically converging," meaning if you ask for more precision, it just keeps getting better until it hits the exact answer (like a GPS that keeps refining your route until it's perfect).

---

The "Sibling" Theory: QCC

The authors also developed a slightly different version called QCC (Quasidegenerate Coupled-Cluster).

• The Analogy: If $\Delta$CC is the reliable, automated self-driving car, QCC is the self-driving car with a "turbo mode" for extreme racing.
• The Trade-off: QCC is even better at handling the most chaotic, "strongly correlated" situations (like a car crashing into a wall and trying to keep driving). However, to get that extra power, you have to tell the car exactly where the trouble spots are. It loses the "black box" simplicity but gains raw power for the hardest problems.

---

What Did They Prove?

The authors didn't just invent the theory; they built the engine and tested it.

1. They built the code: They wrote computer programs that can handle these calculations up to very high levels of complexity (up to 8 electrons moving at once!).
2. They tested it: They ran simulations on small molecules like $CH^+$ (a carbon-hydrogen ion) and $BH$ (boron hydride).
3. The Results:
   • Accuracy: $\Delta$CC was often more accurate than the current gold standards (EOM-CC), especially for difficult "two-electron" jumps.
   • Convergence: While other methods (like standard perturbation theory) sometimes spiral out of control and give wrong answers, $\Delta$CC consistently marched toward the correct answer.
   • Versatility: It handled ionization (removing electrons) and electron attachment (adding electrons) just as well as it handled excited states.

The Bottom Line

This paper presents a universal solver for quantum chemistry.

• Before: We had one tool for calm days (Ground State CC) and a different, sometimes finicky tool for storms (EOM-CC).
• Now: We have a single, robust, "black box" tool ($\Delta$CC) that can handle calm days, storms, and everything in between with high accuracy.

It's a significant step toward making quantum chemistry a truly predictive science, where we can simulate complex chemical reactions and materials without needing a human expert to guess which mathematical tricks to use. It turns the "art" of selecting reference states into a reliable, automated science.

Technical Summary

Here is a detailed technical summary of the paper "Degenerate coupled-cluster theory" by So Hirata.

1. Problem Statement

Single-reference Coupled-Cluster (CC) theory is the gold standard for accurate, size-extensive, and black-box ab initio calculations of ground-state electronic structures. However, it struggles with excited states, ionized states, and electron-attached states when these states are dominated by degenerate or quasi-degenerate Slater determinants (e.g., open-shell singlets, high-spin states, or core-ionized states).

Existing solutions have significant limitations:

• Equation-of-Motion CC (EOM-CC): While effective for one-electron excitations, it often fails for multi-electron processes (e.g., double excitations, satellite states) and lacks roots for excitations exceeding the truncation rank of the linear operator. It relies on a non-degenerate ground state reference.
• Configuration Interaction (CI): While capable of handling degeneracy, it lacks size-extensivity (errors grow with system size) and converges slowly toward the Full Configuration Interaction (FCI) limit.
• Many-Body Green's Function (MBGF) Theory: Often fundamentally non-convergent for satellite states due to the non-analyticity of Green's functions in certain frequency domains.
• Existing Multireference CC (MRCC): Many existing MRCC methods (e.g., Jeziorski-Monkhorst, Li-Paldus) are not "black-box" methods; they require user intuition to select active orbitals or specific reference determinants, and some lack guaranteed convergence to FCI.

The paper addresses the need for a systematically converging, size-extensive, black-box, ab initio method that can treat any stationary state (ground, excited, ionized, attached) starting from any degenerate or non-degenerate Slater-determinant reference, regardless of spin multiplicity or orbital occupancy.

2. Methodology: $\Delta$CC and QCC Theories

The author introduces two primary theoretical frameworks:

A. Degenerate Coupled-Cluster ($\Delta$CC) Theory

This is the central contribution. It generalizes the single-reference CC ansatz to handle $M$-fold degenerate reference determinants ($|I\rangle, |J\rangle, \dots$).

• Ansatz: The theory posits a set of $M$ exponential wave functions $\{e^{\hat{T}_I}|I\rangle\}$ that satisfy the Schrödinger equation within the space of determinants reachable by $(1+\hat{T}_I)|I\rangle$.
• Key Equations:
  1. Projection: $\hat{P}_I \hat{H} e^{\hat{T}_I}|I\rangle = \hat{P}_I \sum_{J=1}^M \hat{P}_J e^{\hat{T}_J}|J\rangle E_{JI}$
  2. C-Condition (Orthogonality): $\langle J | \hat{P}_I e^{\hat{T}_I} | I \rangle = \delta_{JI}$. This ensures that the exponential wave function of one reference does not overlap with other degenerate references, which is crucial for size-extensivity and exact convergence.
• Energy Determination: The energy matrix $\mathbf{E}$ is obtained via $\mathbf{E} = \mathbf{S}^{-1}\mathbf{H}$, where $\mathbf{H}$ and $\mathbf{S}$ are the Hamiltonian and overlap matrices in the internal space. The physical energies are the eigenvalues of $\mathbf{E}$.
• Black-Box Nature: Unlike other MRCC methods, $\Delta$CC does not require the user to specify active spaces. It uses a single set of formalism and code for any reference determinant (Aufbau or non-Aufbau).
• Relation to Perturbation Theory: $\Delta$CC is shown to be the natural coupled-cluster extension of Degenerate Rayleigh-Schrödinger Perturbation ($\Delta$MP) theory, just as standard CC is to Møller-Plesset (MP) theory. It sums infinite diagrams, including "folded" or "backward" diagrams characteristic of degenerate perturbation theory.

B. Quasidegenerate Coupled-Cluster (QCC) Theory

As a byproduct, the author develops a variant called QCC.

• Difference from $\Delta$CC: QCC removes the projectors $\hat{P}_J$ from the right-hand side of the projection equation.
• Trade-off: QCC is not a black-box method (it requires more user input regarding the model space) but is expected to be better suited for strong correlation and bond-breaking problems where the internal space correlation is critical.
• Properties: Like $\Delta$CC, it is size-extensive and converges to FCI.

C. Implementation

Two algorithms were implemented:

1. Determinant-based, General-Order: Uses bit-string representations for determinants. Capable of running up to the FCI limit to generate benchmark data.
2. Algebraic, Optimal-Scaling: Implemented for $\Delta$CCS and $\Delta$CCSD using the Tensor Contraction Engine (TCE). These scale as $O(n^4)$ and $O(n^6)$ respectively (where $n$ is the number of spin-orbitals).

3. Key Contributions

1. Unified Formalism: A single theoretical framework ($\Delta$CC) handles ground states, excited states, ionizations, and electron attachments uniformly. It eliminates the need for separate EOM-CC, IP-EOM-CC, or EA-EOM-CC codes.
2. Proof of Convergence and Size-Extensivity: The paper provides a rigorous proof (via the C-condition and linked-diagram theorem) that the full $\Delta$CC method is equivalent to FCI and that truncated versions are size-extensive.
3. Black-Box Multireference: It achieves the "black-box" ideal for multireference problems, requiring only the reference determinant(s) as input, removing the need for expert selection of active orbitals.
4. Connection to Green's Functions: It proposes $\Delta$CC as a convergent alternative to MBGF theory for computing ionization potentials and electron affinities, avoiding the non-convergence issues of Feynman-Dyson expansions.
5. Projection Hartree-Fock: The single-excitation instance ($\Delta$CCS) is identified as a projection Hartree-Fock theory for degenerate references, offering utility for core ionizations and high-spin states.

4. Results and Numerical Performance

The methods were benchmarked against FCI, CI, EOM-CC, $\Delta$MP, and MBGF for molecules like $\text{CH}^+$, $\text{CH}_2$, $\text{BH}$, and $\text{C}_2$.

• Accuracy:
  • $\Delta$CCSD vs. EOM-CCSD: $\Delta$CCSD shows uniform high accuracy for both one-electron and two-electron excitations. In contrast, EOM-CCSD performs well for one-electron excitations but suffers large errors (>1 eV) for two-electron excitations and lacks roots for higher excitations.
  • Satellite States: $\Delta$CCSD accurately describes satellite ionization and electron-attachment transitions (multi-electron processes) where EOM-CC and CI fail or require very high truncation orders.
  • Core Ionizations: $\Delta$CCS (singles only) performs surprisingly well for core ionizations due to dominant orbital relaxation effects, a regime where standard EOM-CC often struggles without specific modifications.
• Convergence:
  • The $\Delta$MP series (perturbation theory) converges to the correct FCI limits for both Koopmans and satellite states.
  • The MBGF series converges to correct limits only for Koopmans states but fails (converges to wrong limits) for satellite states.
  • $\Delta$CC, being an infinite summation of $\Delta$MP diagrams, is always convergent.
• Performance Order:
  • For transition energies: QCC $\approx$ $\Delta$CC > EOM-CC > CI (at the same excitation order).
  • At the same cost scaling ($O(n^6)$): $\Delta$CCSD > EOM-CCSD > $\Delta$MP3 > CISD.

5. Significance

This work represents a major advancement in electronic structure theory by bridging the gap between single-reference CC and multireference problems.

• Predictive Power: It offers a predictive, black-box tool for complex electronic states (strong correlation, excited states, ionization) without the "chemical intuition" required by current MRCC methods.
• Theoretical Unification: It unifies the treatment of ground and excited states under the exponential ansatz, challenging the necessity of the Equation-of-Motion formalism for many applications.
• Robustness: By avoiding the non-analyticity issues of Green's function theory and the size-extensivity issues of CI, it provides a more robust path to chemical accuracy for a wider range of molecular phenomena.
• Future Outlook: The author suggests this framework could serve as a foundation for finite-temperature CC theory and the development of analytical derivatives and non-iterative corrections (e.g., $\Delta$CCSD(T)) for even greater efficiency.

In summary, $\Delta$CC theory is presented as a superior, general-purpose alternative to EOM-CC for systems where the wave function is dominated by degenerate determinants, offering a unique combination of size-extensivity, systematic convergence, and black-box applicability.