Analytic Computation of Vibrational Circular Dichroism Spectra Using Configuration Interaction Methods
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Analytic Computation of Vibrational Circular Dichroism Spectra Using Configuration Interaction Methods".
Mira: Here's the gist of the paper. In this work, we present the first derivation and implementation of analytic gradient methods for the computation of atomic axial tensors (AATs) required for simulations of vibrational circular dichroism (VCD) spectra using configuration interaction methods including double (CID) and single and double (CISD) excitations.
Kai: Methodological Development The authors developed an analytic formulation based on derivatives of excited determinants using a second-quantized approach, which they validated through the finite-difference formulation reducing the computational scaling to O(N5) (Page 3).
Mira: The purpose of this work is to extend these methods for the first time to the configuration interaction singles and doubles (CISD) method in order to examine the significance of singly excited determinants on VCD rotatory strengths (Page 3).
Kai: They also report additional improvements upon previous implementations allowing the use of non-canonical perturbed orbitals, which provide greater numerical stability in the case of near-degenerate orbitals, and frozen-core for greater computational efficiency and for use with basis sets designed for the treatment of valence correlation (e.g., the cc-pVXZ basis sets of Dunning and co-workers) (Page 3).
Paper summary: Kai: To recap, this paper is about developing and implementing a new way to calculate atomic axial tensors needed for simulating vibrational circular dichroism spectra using configuration interaction methods like double and single and double excitations.
Mira: The core thesis is presenting the first derivation of analytic gradient methods for computing these atomic axial tensors. They claim they have a new implementation that uses non-canonical perturbed orbitals for better numerical stability when orbitals are near-degenerate, plus frozen-core capabilities for efficiency with valence correlation basis sets like cc p v x z.
Lev: And they validated this analytic CID and CISD formulation against two new finite difference approaches, which reduced the computational scaling to O N five given the appropriate intermediates. That's a concrete complexity reduction they are claiming.
Kai: The purpose of this work is explicitly stated as extending these methods for the first time to the configuration interaction singles and doubles method in order to examine the significance of singly excited determinants on VCD rotatory strengths.
Mira: They also report improvements that allow the use of those non-canonical perturbed orbitals and frozen-core capabilities, which are designed to give greater numerical stability in cases where orbitals are near degenerate, and for greater computational efficiency when using basis sets designed for valence correlation.
Lev: So what this means is that they're not just doing a calculation; they're building a toolkit that should be more robust and faster than previous approaches when you move into the CISD territory.
Kai: They also formulate the magnetic dipole transition moment as an overlap between ground state wave function derivatives with respect to nuclear displacements and ket state wave function derivatives with respect to the beta-th Cartesian direction of a magnetic field.
Mira: This quantity, I lambda alpha beta, is then approximated using a CISD intermediate form, which leads to an expression for the atomic axial tensor in terms of CI coefficients and their derivatives.
Lev: The final expression they present for the fully normalized atomic axial tensor I lambda alpha beta involves two times the integral over N squared plus several terms involving the derivative of N with respect to R lambda squared times X a i U H beta i alpha plus other terms involving X c a i and derivatives of determinants, all summed up.
Kai: They compare this new approach against Hartree-Fock theory, Møller-Plesset perturbation theory, CID, and CISD theories using the aug cc p v d z basis set on a molecular test set including four of five molecules.
Mira: The findings show sign discrepancies between Hartree-Fock and Møller-Plesset perturbation theory compared to their new CID and CISD methods for four of the five molecules.
Lev: And they also observed similar discrepancies between the CID and CISD methods when looking at one specific molecule, (M) minus one, three dimethylallene.
Kai: The paper also points out that the most notable differences between Hartree-Fock and Møller-Plesset perturbation theory occurred at three hundred ninety-four point twenty-two centimeters minus one and one thousand three hundred nine point ninety-two centimeters.
Mira: These specific differences are related to hydrogen bending motions that are perpendicular and parallel to the O minus O bond axis respectively.
Lev: So, what this implies is that the analytic approach gives you a way to tackle these complex VCD calculations more directly than some of the standard perturbation theories might allow for in those specific regions.
Kai: The paper also shows substantial differences in rotatory strengths between Møller-Plesset perturbation theory and CID methods versus CISD for the nearly degenerate in phase and out of phase hydrogen bond stretches at three thousand seven hundred sixty-seven point forty-six centimeters minus one and three thousand seven hundred sixty-eight point eighty centimeters.
Mira: So, what this means is that while the analytic method gives you a better tool for calculation, you still need to be careful about which level of theory you use when interpreting those specific spectral features.
Lev: And on the limitation side, they note that the analytic formulation is based on derivatives of excited determinants using a second-quantized approach, which means it's an approximation depending on how well that approximation holds up for the specific molecular system being studied.
Conclusion: Kai: So to wrap up, we're talking about this paper "Analytic Computation of Vibrational Circular Dichroism Spectra Using Configuration Interaction Methods" by Brendan M Shumberger and T Daniel Crawford.
Mira: The main contribution is providing the first derivation and implementation of analytic gradient methods for computing atomic axial tensors for VCD spectra using configuration interaction methods, specifically including double and single and double excitations.
Lev: They achieved a computational scaling of O N five when validated through finite difference approaches, which gives them a good performance metric to consider for real hardware implementation.
Kai: The paper shows that by extending the analytic methods to CISD, they can examine the significance of singly excited determinants on VCD rotatory strengths.
Mira: And the implications are that this gives chemists and theorists a more direct path to calculating these spectral features compared to relying solely on standard DFT methods which have been the limitation until recently.
Lev: It’s about having a tool that is designed to handle near-degeneracy better through those non-canonical orbitals, which is crucial for real quantum chemistry work.
Kai: Overall, this work shows a pathway toward using analytic gradient methods to compute VCD spectra more efficiently with configuration interaction methods.
Mira: It provides a framework where researchers can investigate the interplay between different levels of theory, like Hartree-Fock and MP2 compared to CID and CISD.
Lev: And for someone building error correction systems, it means understanding how the overhead of these analytic calculations lands on the actual hardware resources you't need to account for.
Brendan M. Shumberger, T. Daniel Crawford
Department of Chemistry, Virginia Tech
physics.chem-ph
Submitted: 2025-10-30
Updated: 2025-10-30
Code: https://github.com/bshumberger/apyib
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 79/100
The gist: In this work, we present the first derivation and implementation of analytic gradient methods for the computation of atomic axial tensors (AATs) required for simulations of vibrational circular…
Key concepts
- Atomic Axial Tensors (AATs)
- These are quantities derived from the derivative of the wave function with respect to nuclear displacements and magnetic field directions. They are crucial components used in calculating the transition moments that determine the circular dichroism signal in VCD spectra.
- Configuration Interaction (CI) Methods
- These are quantum chemistry techniques used to approximate molecular electronic states by considering linear combinations of excited determinants. The study specifically focuses on single and double excitations (CISD) to model the electronic structure relevant to VCD.
- Analytic Gradient Methods
- This involves deriving a mathematical expression directly from the wave function derivatives rather than relying solely on numerical approximations. This approach reduces computational scaling and allows for more precise calculations of spectroscopic properties like AATs.
- Vibrational Circular Dichroism (VCD)
- VCD is a spectroscopic technique that measures the differential absorption of left- and right-circularly polarized light by molecules. It is sensitive to the rotational strengths, which are related to the transition moments calculated using these AATs.
Terminology
Summary
In this work, we present the first derivation and implementation of analytic gradient methods for the computation of atomic axial tensors (AATs) required for simulations of vibrational circular dichroism (VCD) spectra using configuration interaction methods including double (CID) and single and double (CISD) excitations. p. 0
Methodological Development
The authors developed an analytic formulation based on derivatives of excited determinants using a second-quantized approach, which they validated through the finite-difference formulation reducing the computational scaling to O(N5) (Page 3). The purpose of this work is to extend these methods for the first time to the configuration interaction singles and doubles (CISD) method in order to examine the significance of singly excited determinants on VCD rotatory strengths (Page 3). p. 2 They also report additional improvements upon previous implementations allowing the use of non-canonical perturbed orbitals, which provide greater numerical stability in the case of near-degenerate orbitals, and frozen-core for greater computational efficiency and for use with basis sets designed for the treatment of valence correlation (e. p. 2g., the cc-pVXZ basis sets of Dunning and co-workers) (Page 3). p. 2
Theoretical Formulation
The magnetic dipole transition moment includes the quantity Iλαβ, which is formulated as the overlap between ground-state wave function derivatives where the bra-state wave function is differentiated with respect to nuclear displacements, Rλα, and the ket-state wave function is differentiated with respect to the β-th Cartesian direction of the magnetic field, Hβ (Page 4). By approximating the wave function as a intermediately normalized CISD wave function, ΨG(R, H⃗β) E ≈ Φ0⟩ + Cˆ1 Φ0⟩ + Cˆ2 Φ0⟩ (Page 4), they obtain an expression for the AAT in terms of CI coefficients, derivatives of CI coefficients, determinants, and derivatives of determinants (Page 4). p. 2 The final expression for the fully normalized AAT is given by Eq. p. 5 (10) [Iλαβ]full = N2 [Iλαβ]int + N∂N/∂Rλα 2Xia c a i U Hβ ia + Xia c a i† ∂ca i ∂Hβ + 2Xia c a i Xjb U Hβ jb c ab ij† + 1/4 Xijab c ab ij† ∂cab ij ∂Hβ (Page 7). p. 6
Comparative Analysis of Theories
The study compares Hartree-Fock (HF), second-order Møller-Plesset perturbation (MP2) theory, CID, and CISD theories using the aug-cc-pVDZ basis set for a molecular test set including four of five molecules (Page 10). For four of the five molecules, they noted sign discrepancies between the HF and MP2 methods compared to that of the new CID and CISD methods for four of the five molecules as well as similar discrepancies between the CID and CISD methods for (M)-1,3-dimethylallene (Page 10). p. 0 The most notable differences between HF and MP2 levels of theory occur at 394. p. 722 cm−1 and 1309.92 cm−1 which are related to hydrogen bending motions perpendicular and parallel to the O−O bond axis, respectively (Page 11). p. 5 p. 5
VCD Spectrum Findings
The VCD spectra for (P)-hydrogen peroxide show rather distinct differences between the four spectra even for a small molecule such as (P)-hydrogen peroxide (Page 12). The most notable differences between the HF and MP2 levels of theory occur at 394. p. 722 cm−1 and 1309.92 cm−1 which are related to hydrogen bending motions perpendicular and parallel to the O−O bond axis, respectively (Page 11). Interestingly, we note substantial differences in rotatory strengths between the MP2 and CID methods versus CISD for the nearly degenerate in-phase and out-of-phase hydrogen bond stretches at 3767. p. 246 cm−1 and 3768.80 cm−1 (Page 12).
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Analytic gradient methods for computing atomic axial tensors (AATs) are implemented using non-canonical perturbed orbitals to improve numerical stability in cases of orbital near-degeneracies. This allows AI systems to maintain accuracy when simulating vibrational circular dichroism (VCD) spectra involving complex electronic structures.
-
The implementation includes frozen-core capabilities for greater computational efficiency, particularly when using basis sets designed for valence correlation. This enables the AI system to handle larger molecular systems more efficiently while preserving necessary chemical detail.
-
The analytic formulation scales as O(N5) when utilizing intermediates, which reduces the computational scaling from O(N6) in previous methods. This allows for the simulation of larger molecules or higher-level calculations within a reasonable timeframe.
-
The system can compare Hartree-Fock (HF), second-order Møller-Plesset perturbation (MP2), configuration interaction singles and doubles (CID), and CISD levels of theory using a common basis set like aug-cc-pVDZ. This capability allows the AI to assess the systematic inclusion of dynamic electron correlation on VCD rotatory strengths.
-
The system can detect specific physical effects such as
sign discrepancies between the HF and MP2 methods compared to that of the new CID and CISD methods for four of the five molecules.
This provides a diagnostic tool for identifying where different theoretical approximations diverge in predicting spectroscopic properties. -
The AI system can quantify how much inclusion of singly excited determinants shifts rotatory strengths, noting a shift
from by about 1 to 1.5 times 10 minus 44 esu2 cm2 between the CID and CISD methods
for certain modes in specific molecules like (S)-methyloxirane. This allows for precise assessment of the contribution of higher-order excitations to VCD spectra. -
The system can identify that
the molecules most affected by the inclusion of singly excited determinants into the wave function are those which exhibit axial chirality ((P)-hydrogen peroxide and (M)-1,3-dimethylallene) rather than point chirality.
This suggests a targeted approach for focusing computational resources on chiral systems with specific structural motifs.
Related papers
- A System-Independent Metadynamics Strategy for Reactive Training Data: Application to Gas-Phase Organic Reactions
- A Machine Learning Model for the Chemistry of a Solvated Electron
- Compression of virtual spaces in transcorrelated methods via singular value decomposition: application to the G2 set
- Charge Transfer with a Spin. II: A Framework for Diabatization which Localizes Charge and Spin
- Diabatic Seam Space Sampling for Hydrogen Tunneling Systems with Nuclear-Electronic Orbital Theory
- Augmented Roothaan-Hall Hessian Applied to Spin-Restricted Open-Shell Density-Functional Theory