Reduction of finite-size effects for second-order M ller-Plesset perturbation theory with singularity subtraction

arXiv:2605.12727 · physics.comp-ph, physics.chem-ph · Submitted 2026-05-12 · Read on arXiv

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: We're looking at the paper: "Reduction of finite-size effects for second-order M ller-Plesset perturbation theory with singularity subtraction".

Mira: Here's the gist of the paper. Second-order Møller–Plesset perturbation theory (MP2) for periodic systems suffers from finite-size errors (FSEs) that have inverse volume scaling due to the Coulomb kernel singularity in reciprocal space The method introduces MP2 singularity subtraction (MP2SS), a systematic approach that applies the singularity subtraction strategy to reduce MP2 FSEs Across a range of extended solids and molecular crystals at large basis sets, MP2SS reduces the prefactor in MP2 FSEs and subsequently the computational resources required to converge to the TDL.

Kai: Theoretical Background The paper reviews periodic MP2 theory in a Bloch-orbital framework, defining pair densities and two-electron integrals using Fourier coefficients The restricted-orbital closed shell MP2 correlation energy at finite Nk is expressed as a triple quadrature over the FBZ of the regular MP2 expression The TDL is achieved as the mesh K becomes dense in omega∗ and each Riemann sum 1/Nk P k∈K converges to the corresponding integral 1omega∗ R3 dk The discussion on FSEs focuses on the part introduced by the quadratures involving the Coulomb-kernel singularity Classification of Singularities The direct term of the MP2 energy at TDL can be expressed as an integral over the FBZ, leading to two types of singularities: a fourth-order (quartic) singularity at ∆G = 0 as q → 0, and a collection of second-order (quadratic) singularities when ∆G ≠ 0 Both types of singularities produce O(N −1 k) contributions to the MP2 FSE Analogously, the exchange part of the MP2 correlation energy at TDL can be expressed as a double integral over the FBZ, where the finite-size error in the exchange term when discretizing over q1 and q2 comes from singularities at q1 = q2 = 0 Numerical Quadrature and Singularity Subtraction (SS) Uniform quadrature is simple but can degrade convergence when integrands contain singularities or cusps Singularity subtraction (SS) reduces quadrature errors of singular integrands by adding and subtracting an auxiliary function that reproduces the leading local singular behavior The goal is to isolate the singular contribution into a term that can be integrated analytically or to very high accuracy, leaving behind a smoother remainder better suited to uniform quadrature Singularity Subtraction Applied to Periodic MP2 SS is applied differently to the two types of singularities in MP2: 1.

Mira: Fourth-order singularity in the direct term: The contribution is approximated by E SS MP2 ← 1/Nk X'q∈Kq+L∗ Sd(q, 0, 0) − Ad(q, 0; α)q4 + 1/omega∗ Z R3 dq Ad(q, 0; α)q4 2.

Kai: Second-order singularities in the direct term: The contribution is approximated by E SS MP2 ← 1/Nk X'q∈Kq+L∗ S˜d,2(q)q2 − fd,2(q; α')q2 + 1/omega∗ X ∆G∈L∗≠0 Z R3 dq hd,2(q; α')hd,2(q + ∆G; α'), where S˜d,2(q) is the “folded” version of the structure factor Sd(q, 0, ∆G) 3.

Title and authors: Kai: So we’re looking at this paper called "Reduction of finite-size effects for second-order Møller–Plesset perturbation theory with singularity subtraction." It tackles those finite size errors that scale inversely with the volume of the system.

Mira: And these errors come from how they discretize integrals over reciprocal space, which is a tricky area in periodic quantum chemistry calculations.

Lev: Basically, if you don't sample the momentum space finely enough, say with N k points, the error in your energy scales as O of n minus one k.

Kai: The authors introduce this method called MP2 singularity subtraction to systematically reduce those errors by subtracting parts that cause problems when the Coulomb kernel becomes singular.

Mira: They define a volume term Omega s which represents the physical size of the supercell corresponding to the sampled k points, and they use that to write down how those two electron integrals are expressed.

Lev: When you look at equation three on page three they show that the integration over reciprocal space is done by tessellating a set K q plus L star across all of R three using this volume term.

Kai: This setup lets them clearly identify where the singularity happens when the momentum transfer vector q plus G equals zero, which is exactly what causes those errors in periodic MP2 calculations.

Mira: The paper explains that they look at two main types of singularities in both the direct term and the exchange part of the correlation energy.

Lev: One type is a fourth order singularity when the energy difference is zero and q goes to zero, and there are also second order singularities when it's not zero.

Kai: And because both these types contribute that O of n minus one k scaling to the finite size error, they show how this scaling limits routine applicability for large systems.

Mira: So the paper sets up the problem by precisely defining those errors based on how many k points you sample and where you are in reciprocal space.

Lev: And then MP2 singularity subtraction is proposed as a systematic way to reduce that prefactor, which means less work is needed to get to the thermodynamic limit.

The paper's summary: Kai: Now we look at what the paper actually summarizes in this paper on "Reduction of finite-size effects for second-order Møller–Plesset perturbation theory with singularity subtraction."

Mira: They summarize that they are dealing with the fact that the restricted orbital closed shell MP2 correlation energy at a finite N k is expressed as a triple quadrature over the first Brillouin Zone of the regular MP2 expression, which is where all these errors originate.

Lev: That triple quadrature looks pretty complicated, and they lay out how to express it using crystal momentum conservation where ka minus ki equals ka minus kb plus q equals zero.

Kai: They define a specific volume term Omega s as the volume of the periodic Born-von Karman supercell corresponding to that discretized Brillouin zone, which is key because it’s what links the k points back to the physical size of your system in three dimensions.

Mira: Then they show how those integrals, specifically the two-electron integrals, can be written using this volume term multiplied by a complicated expression involving structure factors and the Coulomb kernel V over q plus G.

Lev: They focus on how to handle that singularity when q plus G equals zero, which is the point where the Coulomb kernel becomes singular in reciprocal space. This is what creates those errors they are trying to eliminate.

Kai: The paper then shows how their total MP2 energy expression can be written as a triple quadrature over the first Brillouin Zone, and then they explicitly show where those singularities come from in both the direct term and the exchange part of the correlation energy.

Mira: Specifically, for the direct term, they find two types of singularities: a quartic one when the energy difference is zero as q goes to zero, and second order ones when it’s not zero.

Lev: And because those two types contribute an O N minus one k contributions to the MP2 finite-size error. That O N minus one k scaling is what makes routine MP2 impractical for large systems without a fix.

Kai: So in short, they are mapping out the problem of finite-size errors precisely and then proposing a systematic subtraction strategy for them.

Mira: And then they detail exactly how the MP2SS method works by showing its application to both the direct term and the exchange part of MP2, which is a pretty detailed look at the mechanics.

The paper's improvements: Lev: Now that we know what they’ve done, let’s look at what improvements they suggest in this paper. They aren't just saying "we did this," but suggesting next steps for the method itself.

Kai: One big improvement is that they test three different settings for the MP2SS configuration: Gaussian, exponential, and tuned. They found the MP2SS-Gaussian configuration gives them the most robust corrections because of how smooth its auxiliary function behaves near zero.

Mira: The paper implies that while they tested the other two—exponential and tuned—they found them to be less accurate for standard MP2 correlation energy components; they often over-correct standard MP2 results.

Lev: So the recommendation is pretty clear for practical use: if you’re doing routine calculations, stick with MP2SS-Gaussian because it provides the most robust finite-size error correction for the correlation energy.

Kai: That seems like a very practical piece of advice, especially when you’re trying to balance accuracy and computational cost on a real system.

Mira: It also points toward future work being important, suggesting that computing derivatives of those structure factors at the origin could give even better corrections down the line.

Lev: And from an implementation view, they mention that improving how those structure factors are built and sampled would be a key way to push this method further.

Kai: So the overall message is that while MP2SS-Gaussian is great now, the path forward involves getting even better sampling techniques for those integrals.

Conclusion: Kai: So that’s how MP2SS works and what it achieves in this paper on "Reduction of finite-size effects for second-order M ller–Plesset perturbation theory with singularity subtraction." Thanks for listening today.

Mira: So we’ve been looking at how those finite size errors in second order M ller–Plesset perturbation theory scale inversely with volume, and this paper shows a way to fix that by subtracting the singular parts directly.

Lev: Right, so they take the math where the Coulomb kernel blows up when things get too close together in reciprocal space, and they use some singularity subtraction technique to clean up those errors.

Kai: From an error correction standpoint, reducing that prefactor is what matters most for getting results on real hardware; it means less overhead when you try to run these calculations on actual quantum computers.

Mira: They tested three different ways to do this subtraction—Gaussian, exponential, and tuned—and they pointed toward the Gaussian method as being the most robust choice for handling those errors across different crystals.

Lev: Yeah, and it’s important to remember that this method doesn't fix everything; the authors themselves noted a bottleneck where building those structure factors takes a lot of resources, scaling up pretty high with the size of your system.

Kai: So for anyone building simulations for materials that are extended or complex, using that MP2SS-Gaussian setting seems like the sensible default right now.

Mira: Exactly, so while it reduces the error for routine calculations, you still have to deal with that computational cost when you’re trying to model massive systems.

Lev: It's interesting because this same subtraction framework might actually work for other post-Hartree-Fock methods like MP3 or coupled cluster theories later on.

Kai: That potential extension is significant because it would mean we could apply this finite size error correction to even more complex methods.

Mira: The main thing here is that this systematic approach lets us reach the true thermodynamic limit much faster than just using standard MP2 without any corrections.

Lev: It’s a solid piece of work, really showing how you can take a mathematically messy problem and apply a systematic subtraction strategy to make it practical.

Stephen Jon Quiton, Juan D. F. Pottecher, Martin Head-Gordon, Lin Lin

Department of Chemistry, University of California, Berkeley · Department of Mathematics, University of California, Berkeley · Applied Mathematics and Computational Research Division, Lawrence Berkeley National Laboratory · Computing and Mathematical Sciences, California Institute of Technology

physics.comp-ph, physics.chem-ph

Submitted: 2026-05-12

Updated: 2026-09-18

Comments: 49 pages, 53 figures (including supporting information)

DOI: 10.1021/acs.jctc.6c00971

Code: https://github.com/stephen-quiton/fsec

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 85/100

The gist: Second-order Møller–Plesset perturbation theory (MP2) for periodic systems suffers from finite-size errors (FSEs) that have inverse volume scaling due to the Coulomb kernel singularity in…

Key concepts

Finite-Size Errors (FSEs)
These are errors in MP2 calculations for periodic systems caused by using a finite number of unit cells ($N_k$) instead of an infinite system. They arise because the Coulomb kernel has singularities in reciprocal space, leading to errors that scale inversely with the system size.
Singularity Subtraction (SS)
A numerical technique used to improve integration accuracy when functions have singularities or cusps. SS works by adding and subtracting a carefully constructed auxiliary function that mimics the singular behavior, allowing the remaining smooth part of the integrand to be integrated more accurately using standard methods.
MP2 Singularity Subtraction (MP2SS)
The specific application of SS tailored for periodic MP2 theory. It targets two types of singularities: quartic and second-order in the direct term, and singularities when integrating over different momentum points in the exchange term. This systematic subtraction reduces the FSEs significantly.
Thermodynamic Limit (TDL)
The ideal theoretical limit where a system is infinitely large, allowing for exact calculations. MP2SS helps systems converge to this TDL much earlier than standard MP2 because it systematically removes the dominant finite-size error components.

Terminology

Summary

Second-order Møller–Plesset perturbation theory (MP2) for periodic systems suffers from finite-size errors (FSEs) that have inverse volume scaling due to the Coulomb kernel singularity in reciprocal space The method introduces MP2 singularity subtraction (MP2SS), a systematic approach that applies the singularity subtraction strategy to reduce MP2 FSEs Across a range of extended solids and molecular crystals at large basis sets, MP2SS reduces the prefactor in MP2 FSEs and subsequently the computational resources required to converge to the TDL. p. 0

Theoretical Background

The paper reviews periodic MP2 theory in a Bloch-orbital framework, defining pair densities and two-electron integrals using Fourier coefficients The restricted-orbital closed shell MP2 correlation energy at finite Nk is expressed as a triple quadrature over the FBZ of the regular MP2 expression The TDL is achieved as the mesh K becomes dense in omega∗ and each Riemann sum 1/Nk P k∈K converges to the corresponding integral 1omega∗ R3 dk The discussion on FSEs focuses on the part introduced by the quadratures involving the Coulomb-kernel singularity

Classification of Singularities

The direct term of the MP2 energy at TDL can be expressed as an integral over the FBZ, leading to two types of singularities: a fourth-order (quartic) singularity at ∆G = 0 as q → 0, and a collection of second-order (quadratic) singularities when ∆G ≠ 0 Both types of singularities produce O(N −1 k) contributions to the MP2 FSE Analogously, the exchange part of the MP2 correlation energy at TDL can be expressed as a double integral over the FBZ, where the finite-size error in the exchange term when discretizing over q1 and q2 comes from singularities at q1 = q2 = 0

Numerical Quadrature and Singularity Subtraction (SS)

Uniform quadrature is simple but can degrade convergence when integrands contain singularities or cusps Singularity subtraction (SS) reduces quadrature errors of singular integrands by adding and subtracting an auxiliary function that reproduces the leading local singular behavior The goal is to isolate the singular contribution into a term that can be integrated analytically or to very high accuracy, leaving behind a smoother remainder better suited to uniform quadrature

Singularity Subtraction Applied to Periodic MP2

SS is applied differently to the two types of singularities in MP2:

  1. Fourth-order singularity in the direct term: The contribution is approximated by E SS MP2 ← 1/Nk X'q∈Kq+L∗ Sd(q, 0, 0) − Ad(q, 0; α)q4 + 1/omega∗ Z R3 dq Ad(q, 0; α)q4

  2. Second-order singularities in the direct term: The contribution is approximated by E SS MP2 ← 1/Nk X'q∈Kq+L∗ S˜d,2(q)q2 − fd,2(q; α')q2 + 1/omega∗ X ∆G∈L∗≠0 Z R3 dq hd,2(q; α')hd,2(q + ∆G; α'), where S˜d,2(q) is the “folded” version of the structure factor Sd(q, 0, ∆G)

  3. Exchange Term: The contribution is derived as E SS MP2 ← 1/Nk (X'q∈Kq+L∗ S˜x(q)q2 − X'q∈Kq+L∗ fx(q; α')q2) + 1/omega∗ 2 Z R3 dqhx(q; α'') squared

Performance and Recommendations

The paper tests three MP2SS configurations: Gaussian, exponential, and tuned The MP2SS-Gaussian configuration provides the most robust corrections for the correlation energy due to the smoothness of the auxiliary function near the origin MP2SS-Exponential frequently over-corrects standard MP2 correlation energy components, while MP2SS-Tuned often lands between MP2SS-Gaussian and MP2SS-Exponential The results suggest that for routine calculations, MP2SS-Gaussian provides the most robust FSE corrections for the correlation energy due to the smoothness of the auxiliary function near the origin

Computational Procedure and Bottlenecks

The computational procedure involves four steps: (1) periodic HF calculation, (2) ExxSS correction, (3) periodic MP2 calculation, and (4) MP2SS correction The walltime bottleneck of the MP2SS procedure is the building of the MP2 structure factors for fitting auxiliary functions, scaling up to O(N 3 kNGN2 occN2 virt) This cost can be lowered by reducing the number of sampled q + G points or sampling along reciprocal lattice vectors

Results and Conclusion

MP2SS clearly reduces the error for all crystals and allows them to reach the TDL earlier than uncorrected MP2 Among configurations, MP2SS-Gaussian consistently performs the best in this regime The results suggest that MP2SS-Gaussian provides the most robust FSE corrections for the correlation energy due to the smoothness of the auxiliary function near the origin Further improvements could involve computing derivatives of structure factors at the origin and improving how structure factors are built and sampled The singularity subtraction framework developed here can be extended to other post-SCF theories like MP3 and coupled-cluster theories

Supporting Information

The Supporting Information contains detailed discussions on basis set dependence, the Laplace-transform formulation for structure factor build, FSE curves with staggered mesh results, line sampling of the structure factor, and decompositions of the second- and fourth-order contributions to the MP2SS direct term correction The data underlying this study are available in CSV format The MP2SS code is also available as an installable Python package at https://github.com/stephen-quiton/fsec

Data and Software Availability

The data underlying this study are available in the published article and in the Supporting Information, which includes the data used to plot the finite-size error curves in CSV format The MP2SS code is also available as an installable Python package at https://github.

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

  1. Singularity subtraction for second-order Møller–Plesset perturbation theory (MP2SS) can be extended to other post-Hartree-Fock (HF) methods like MP3 and coupled-cluster theories, addressing finite-size errors in those calculations by explicitly accounting for their similar singularities.

  2. The system can perform better at achieving thermodynamic limit convergence for insulators and semiconductors by using MP2SS because the method reduces the prefactor of the finite-size error compared to uncorrected calculations.

  3. The improved AI system can be used to generate accurate ground state energies for large unit cells in machine-learned interatomic potentials (MLIPs) by providing FSE corrections like MP2SS, which are crucial for these potentials.

  4. The AI system can provide a robust default setting by recommending MP2SS-Gaussian as the default configuration due to its performance across extended solids and molecular crystals, as it provides the most robust FSE corrections.

Sources

Related papers