Third-Order Algebraic Diagrammatic Construction Renormalized by Particle-Particle Ladders

arXiv:2608.28424 · physics.chem-ph · Submitted 2026-08-28 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: We're looking at the paper: "Third-Order Algebraic Diagrammatic Construction Renormalized by Particle-Particle Ladders".

Mira: Here's the gist of the paper. The algebraic diagrammatic construction (ADC) of the electronic self-energy is based on Møller–Plesset perturbation theory and is improved by introducing dressed denominators that resum to infinite order diagonal elements of ladder diagrams described by the pair propagator in the particle-particle T-matrix approximation Third-order ADC with EN-dressed denominators leads to an almost fourfold increase in accuracy compared to standard ADC(3) on a challenging benchmark set for IPs of closedshell molecules, reaching an accuracy between EOM-CCSD and EOM-CCSDT Pruning redundant reasoning tokens The algebraic diagrammatic construction (ADC) of the electronic self-energy is based on Møller–Plesset perturbation theory.

Kai: Its vertices contain bare energy denominators that do not capture renormalization of Hartree–Fock excitations through excited-state correlations.

Mira: As shown here, Epstein–Nesbet (EN) partitioning of the electronic Hamiltonian overcomes this deficiency at zero extra cost by introducing dressed denominators that resum to infinite order the diagonal elements of ladder diagrams described by the pair propagator in the particle-particle T-matrix approximation.

Kai: How the training signal is built The EN partitioning renormalizes the energy denominators in the ADC amplitudes with approximate single and double excitations.

Paper summary: Kai: So, we’ve established that this paper is about improving the algebraic diagrammatic construction of electronic self-energy. It starts with Møller–Plesset perturbation theory which uses bare energy denominators that don't account for how Hartree–Fock excitations are renormalized by excited state correlations.

Mira: They tackle this by using Epstein–Nesbet partitioning of the electronic Hamiltonian to introduce dressed denominators at zero extra cost. These dressed denominators resum to infinite order diagonal elements of ladder diagrams described by the pair propagator in the particle-particle T-matrix approximation.

Lev: So they are essentially taking those bare energy terms and replacing them with these more sophisticated, physically dressed versions that capture higher-order interactions through a specific T-matrix route. It sounds like a significant mathematical step for handling excited state correlations.

Kai: Exactly, and this leads to the core claim: third-order ADC with these EN-dressed denominators achieves an almost fourfold increase in accuracy compared to standard ADC(three) on a challenging benchmark set for IPs of closedshell molecules <ref:2608.28424#pg1,an almost fourfold increase in accuracy compared to standard ADC(3) on>.

Mira: They say this new method reaches an accuracy level that sits between EOM–CCSD and EOM–CCSDT, which is quite high for this kind of calculation, but they also acknowledge that open-shell systems see only modest gains in accuracy.

Lev: So the gain is most pronounced when dealing with closed-shell systems, which makes sense if the reference data we have is mostly from those kinds of molecules. But what about open-shell systems where spin contamination gets messy?

Kai: The paper points out that for open-shell systems, the improvements are more modest because of that spin contamination issue. They also highlighted that the spin-summed variant of EN-ADC(three) performs much better than its spin-averaged counterpart <ref:2608.28424#pg1>.

Mira: It seems like they’ve found a way to make this method work better for certain types of systems by focusing on the summation structure rather than just the average over spin states. This is a useful piece of information for modeling things that aren't perfectly closed-shell.

Lev: From an engineering standpoint, if we move this onto real hardware, we need to know how much computational work this extra dressing adds compared to just running standard ADC(three) <ref:2608.28424#pg1>. Is it a small constant factor or something that grows with the size of the system?

Kai: The paper doesn't explicitly quantify the scaling for the dressing itself, but they do mention practical limitations related to reduced-scaling implementations of standard ADC(three) which can be achieved via Laplace-type factorizations, and those aren't available once those denominators are dressed <ref:2608.28424#pg1>.

Mira: That means a potential implementation hurdle: you get higher accuracy, but you might lose the ability to use some of the standard scaling tricks that make ADC(three) computationally feasible on larger systems <ref:2608.28424#pg1>.

Lev: So it’s a bit like trading some computational convenience for higher precision in the final result, which is always a trade-off we have to manage. But what about how this impacts error correction codes?

Kai: It’s a reminder that in our error correction research, finding these ways to systematically improve the underlying self-energy calculation can lead to more robust Hamiltonians than we're currently using.

Mira: Precisely. It shows that renormalizing the ground state amplitudes entering a self-energy is an interesting avenue for improving approximations, regardless of whether you're looking at molecular IPs or quantum hardware fidelity.

Conclusion: Kai: So, we’ve talked about how this paper, "Third-Order Algebraic Diagrammatic Construction Renormalized by Particle-Particle Ladders," uses an Epstein–Nesbet partitioning to dress the energy denominators with information from particle-particle ladder diagrams.

Mira: The main thing is that this dressing leads to a significant accuracy boost on closed-shell molecules, reaching levels comparable to EOM–CCSD and EOM–CCSDT for third-order ADC.

Lev: So when we think about the title, it really emphasizes that this isn't just another correction; it’s a renormalization process that changes the very structure of how we calculate the self-energy.

Kai: Right, so instead of just adding a few terms to fix an error, they are fundamentally changing the energy denominators themselves based on these ladder diagrams.

Mira: The implication is that for molecular properties, especially those involving closed shells, this provides a much more reliable way to get highly accurate electronic structure data using ADC methods.

Lev: For us working in hardware and error correction, it’s a reminder that the theoretical foundation of our models matters immensely; if the ground state amplitudes are renormalized in this way, the resulting Hamiltonians will have different properties.

Kai: Yeah, so this paper gives us a tool to think about how to systematically improve those fundamental calculations we rely on for everything from chemistry to quantum physics.

Mira: It’s about showing that there is a path forward by incorporating these types of physical interactions directly into the self-energy calculation in a way that yields better results, even if it requires more careful implementation.

Lev: So the paper points us toward exploring these amplitude renormalization techniques as a way to build approximations that are fundamentally more robust than our current standard approaches.

Department of Chemistry and Pharmaceutical Sciences, Vrije Universiteit

physics.chem-ph

Submitted: 2026-08-28

Updated: 2026-10-29

Code: https://github.com/ArnoFoerster/MBPTcode

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 76/100

The gist: The algebraic diagrammatic construction (ADC) of the electronic self-energy is based on Møller–Plesset perturbation theory and is improved by introducing dressed denominators that resum to

Key concepts

Algebraic Diagrammatic Construction (ADC): This is a method used to calculate the electronic self-energy, which describes how an electron interacts with the rest of the system. It is based on Møller–Plesset perturbation theory and uses diagrams to represent these interactions.
Epstein–Nesbet (EN) Partitioning: This technique modifies standard ADC by replacing bare energy denominators with 'dressed' ones. These dressed denominators effectively resum infinite orders of ladder diagrams, which are related to particle-particle interactions described by the T-matrix approximation.
Particle-Particle (pp) T-matrix Approximation: The T-matrix describes the repulsion between two particles (like excited holes or particles). In this context, it is used to define the 'double excitations' that are missing in simpler energy calculations, allowing for a more accurate description of electron correlation.

Terminology

Accuracy — 25× in this paper · explained in 1 episode
Benchmark — 20× in this paper · explained in 2 episodes
Polarization — 12× in this paper · explained in 2 episodes
Core — 5× in this paper · explained in 2 episodes
Effective Hamiltonian — 4× in this paper · explained in 3 episodes
Ab Initio — 2× in this paper · explained in 1 episode
ADC in Multi-reference — 1× in this paper
Consistency — 1× in this paper · explained in 3 episodes
Fidelity — 1× in this paper · explained in 2 episodes
Pruning — 1× in this paper · explained in 4 episodes
Quantum Physics — 1× in this paper · explained in 12 episodes
Random Phase Approximation (RPA) — 1× in this paper · explained in 2 episodes
Regularization — 1× in this paper · explained in 3 episodes
computational cost — 1× in this paper · explained in 2 episodes

Summary

The algebraic diagrammatic construction (ADC) of the electronic self-energy is based on Møller–Plesset perturbation theory and is improved by introducing dressed denominators that resum to infinite order diagonal elements of ladder diagrams described by the pair propagator in the particle-particle T-matrix approximation

Third-order ADC with EN-dressed denominators leads to an almost fourfold increase in accuracy compared to standard ADC(3) on a challenging benchmark set for IPs of closedshell molecules, reaching an accuracy between EOM-CCSD and EOM-CCSDT

Pruning redundant reasoning tokens

The algebraic diagrammatic construction (ADC) of the electronic self-energy is based on Møller–Plesset perturbation theory. Its vertices contain bare energy denominators that do not capture renormalization of Hartree–Fock excitations through excited-state correlations. As shown here, Epstein–Nesbet (EN) partitioning of the electronic Hamiltonian overcomes this deficiency at zero extra cost by introducing dressed denominators that resum to infinite order the diagonal elements of ladder diagrams described by the pair propagator in the particle-particle T-matrix approximation.

How the training signal is built

The EN partitioning renormalizes the energy denominators in the ADC amplitudes with approximate single and double excitations. The double excitations are the diagonal elements of the ladder diagrams that arise from the solution of the particle-particle (pp) T-matrix equations and describe the repulsion between two excited holes/particles that is missing in the first-order doubles MP energy denominators. The authors define modified amplitudes t˜ab(1) that only contain the pp ladder terms.

Where the gains come from

The third-order ADC with EN-dressed denominators leads to an almost fourfold increase in accuracy compared to standard ADC(3) on a challenging benchmark set for IPs of closedshell molecules. For open-shell systems, the gains in accuracy are more modest, but standard ADC is nevertheless outperformed. The spin-summed variant of EN-ADC(3) performs much better than its spin-averaged counterpart

Benchmark Performance and Limitations

The EN-ADC(3) methods outperform standard ADC(3) for closed-shell systems, reducing the mean absolute deviation from reference values relative to standard ADC(3) by a factor of almost four. For open-shell systems, improvements are modest, particularly in cases of strong spincontamination. The magnitude of the EN denominator shift remains sizable even for the largest systems considered in series of linear acenes and oligothiophenes. The authors note that standard ADC(3) changes character along both series, moving from overestimating to underestimating the reference IPs.

Theoretical Extensions and Numerical Considerations

The simple amplitude substitutions leading to EN-ADC(3) are only valid for theories that do not already comprise the relevant 2p2h configurations. The authors suggest that the EN-shift damps the amplitudes and therefore reduces the energy of an excitation. Practical limitations include reduced-scaling implementations of standard ADC(3) which can be achieved through Laplace-type factorizations of the energy denominators, which are unavailable once these denominators are dressed by the EN shift. The work suggests that renormalizing the ground-state amplitudes entering a self-energy is an interesting avenue toward improved approximations.

Connections to T-matrix Approximations

The first-order doubles amplitude of MP-PT might be expressed in terms of the irreducible time-ordered pair propagator L(0)pq(ω). The EN double amplitudes can be expressed as t˜abij = ⟨abij⟩ Dabij − ⟨ijij⟩ − ⟨abab⟩. This expression is shown to be equivalent to the EN-denominator restricted to the pp-channel of Eq. (2.14) through residue calculus. The authors show that the pp-EN first-order doubles denominator coincides with the diagonal of the reducible time-ordered pppropagators in the T-matrix approximation.

Improvements for AI systems

Accurate Ionization Potential Prediction for Closed-Shell Molecules: The EN-ADC(3) method reduces errors by a factor of almost four compared to standard ADC(3) and performs almost twice as accurately as EOM-CCSD for closed-shell systems.
Improved Spin Contamination Handling: For open-shell systems, the EN partitioning shows modest but still significant accuracy improvements over standard ADC(3), especially in cases of strong spin contamination, outperforming EOM-CCSD in some benchmarks.
Systematic Error Identification in Self-Energy Approximations: The paper demonstrates that the amplitude damping by the EN-shift that leads to the observed outstanding performance of EN-ADC(3) is the mechanism for its success, suggesting that focusing on amplitude renormalization rather than just amplitude quality is key.
Size Dependence Behavior Prediction: For series of linear acenes and oligothiophenes, the study reveals that the EN correction, which acts in a fixed direction, improves agreement with the reference where ADC(3) overestimates but increases the already-negative error further along the series, allowing for better prediction of crossover behavior.
Enhanced Computational Efficiency via Ladder Resummation: The connection shown between EN double amplitudes and bare scattering vertices dressed by these propagators suggests that using techniques like Tensor Hypercontraction (THC) or Interpolative Separable Density Fitting (ISDF) could allow the full pp ladders to be calculated without raising the canonical O(N5) scaling of ADC(3).

Abstract

The algebraic diagrammatic construction (ADC) of the electronic self-energy is based on Møller--Plesset perturbation theory. Its vertices contain bare energy denominators that do not capture renormalization of Hartree--Fock excitations through excited-state correlations. As we show here, replacing the Møller--Plesset amplitudes by amplitudes obtained from Epstein--Nesbet (EN) partitioning of the electronic Hamiltonian overcomes this deficiency at no extra cost by introducing dressed denominators that resum to infinite order the diagonal elements of ladder diagrams described by the pair propagator in the particle-particle T-matrix approximation. Third-order ADC with EN-dressed denominators leads to an almost fourfold increase in accuracy compared to standard ADC(3) on a challenging benchmark set for ionization potentials of closed-shell molecules, reaching an accuracy between that of equation-of-motion coupled cluster with singles and doubles (EOM-CCSD) and with singles, doubles, and triples. For open-shell systems, EN-ADC(3) halves the error of ADC(3) for weakly spin-contaminated molecules, outperforming EOM-CCSD, and still improves on ADC(3) for strongly spin-contaminated ones, for which EOM-CCSD remains more accurate.