Some effective operators for graphene monolayer superlattices, from variational perturbation theory

arXiv:2602.19185 · math-ph, cond-mat.mtrl-sci, math.MP, quant-ph · Submitted 2026-02-22 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematical Physics. The day in mathematical physics: rigorous models of physical systems.

Kai: We're looking at the paper: "Some effective operators for graphene monolayer superlattices, from variational perturbation theory".

Mira: Here's the gist of the paper. Our goal is to provide precise effective operators for monolayer graphene at Fermi energy by coupling variational approximation, perturbation theory, and a multiscale method The effective operators derived in this document provide more accurate band diagrams and eigenvectors than the traditional massless Dirac operator Effective Operator Formulation The approach involves considering the semiclassical Schrödinger operator hε defined by (1) and coupling it with a macroscopic potential varying at a scale ε−1 ∈ N to create a superlattice At the effective level, the usual massless Dirac operator is replaced by other operators, and simulations are provided for graphene.

Kai: The exact operator Hεk is studied around the Fermi level at the Dirac point K, leading to Hεk:= Q hε K+εk − EF (10) which acts on L2 per(omega).

Mira: Multiscale Modeling and Basis Construction Effective operators are built using a two-scales reduced basis method where a family F of M linearly independent and periodic functions ψ1,..., ψM ∈ H1 per(omega) is used to impose the behaviors of eigenfunctions at the microscopic scale.

Kai: The resulting effective model for Hεk is given by Hεk:= ε−1M ⊗ 1 + L · ⊗(−i∇k + A) + S ⊗ 1/2 ε(−i∇k + A)2 + V (13) as an operator of CM ⊗ C∞ per(omega).

Title and authors: Kai: So, we're looking at this paper, "Some effective operators for graphene monolayer superlattices, from variational perturbation theory". It sounds like they are trying to get really precise descriptions of how graphene behaves when you put it in a superlattice setup.

Mira: Exactly. They’re tackling this by using variational approximation and perturbation theory together with a multiscale method to create these effective operators instead of just sticking with the standard massless Dirac operator.

Lev: From my side, I'm wondering what kind of fidelity we're talking about here on real hardware; these kinds of effective models usually introduce some error overhead that you really have to account for when you try to run them.

Kai: Right, Lev. The paper says their goal is providing precise effective operators for monolayer graphene at the Fermi energy by coupling variational approximation, perturbation theory, and a multiscale method > Page one > <ref:2602.19185#pg1,precise effective operators for monolayer graphene at>

Mira: They introduce a semiclassical Schrödinger operator h epsilon defined as (one) where v and V are periodic potentials having the same periodicity but varying at a scale ε minus one in N to create a superlattice > Page two > <ref:2602.19185#pg1,the same periodicity but varying at a scale ε>

Lev: So they're starting with this semiclassical Schrödinger operator h epsilon, which has that macroscopic potential term εV (εx) and the microscopic potential v(x) as well > Page one > <ref:2602.19185#pg1>

Kai: That macroscopic term is what makes it a superlattice, and when epsilon becomes small, the spectrum of that operator is numerically hard to reach because the periodicity cell is ε minus one omega > Page one > <ref:2602.19185#pg1,is numerically hard to reach because the periodicity cell is ε>

Mira: They are interested in the states at the interface between filled and unfilled bands because those are usually the main ones determining how quantum systems actually behave > Page one > <ref:2602.19185#pg1,interested in the states at the interface between filled and unfilled bands>

Lev: That sounds like they're focusing on band edges, which is important for error correction because you want to know where your low-energy modes sit relative to those gaps > Page two > <ref:2602.19185#pg1>

Kai: The paper then describes the effective level where the usual massless Dirac operator is replaced by other operators, and they provide simulations in the case of graphene > Page one > <ref:2602.19185#pg1,the usual massless Dirac operator is replaced by other operators, and>

Mira: They build this effective model using a two-scales reduced basis method where a family F of M linearly independent and periodic functions psi one through psi M are used to impose behaviors at the microscopic scale > Page three > <ref:2602.19185#pg2>

Title and authors: Lev: A reduced basis method implies they’re reducing the dimensionality, which is good for computation, but I always worry about how much information you lose when you project onto that basis set F > Page three > <ref:2602.19185#pg2>

Kai: The accuracy depends on the choice of this family F, and a larger M improves the approximation but it increases computational cost > Page three > <ref:2602.19185#pg2>

Mira: They build these families using variational perturbation theory where F0 is w one, w two and higher orders involve derivatives with respect to the momentum parameter k > Page three > <ref:2602.19185#pg2>

Lev: So they're adding momentum dependence into those basis functions, which makes sense if you want to describe things away from the Dirac point K > Page three > <ref:2602.19185#pg2>

Kai: For order one, the matrices s, m, l and t are computed based on specific constants derived from Bloch states at the Dirac point K > Page three > <ref:2602.19185#pg2>

Mira: They simplify the structure of these matrices s, m and t by exploiting graphene symmetries to obtain specific forms for them > Page three > <ref:2602.19185#pg2>

Lev: Exploiting symmetry helps a lot with matrix elements, because those involve complex algebraic manipulations based on properties of pseudo-periodic functions and directional derivatives > Appendix B >

Kai: The formal derivation of the perturbation series for the macroscopic functions justifies the choice of adding derivatives with respect to momentum in families F > Page three > <ref:2602.19185#pg2>

Mira: It shows that expanding eigenmodes in epsilon as x equals x q plus ex one q plus O epsilon two, and showing that the resulting equation has a solution if and only if the right-hand side is orthogonal to span g a q > Page three > <ref:2602.19185#pg2>

Lev: So they are bounding the error between exact and approximate eigenfunctions by P perp F x equals P perp F x, which is important for understanding how much deviation we have from reality > Error Bounds >

Kai: They also use Schur reduction on the effective operator h epsilon k, with h epsilon k from (thirteen), y equals s and p equals p two to lead to a simplified form of the operator like (twenty-four) > Page three >

Mira: That Schur reduction allows for reconstructing an approximation of the eigenvector of the original operator by applying one plus r h epsilon k p two > Page three > <ref:2602.19185#pg2>

Lev: That’s nice because transforming it into a simpler matrix-valued operator that has no divergence makes it much easier to work with for finding those eigenvectors > Page three > <ref:2602.19185#pg2>

Title and authors: Kai: The simulation results show that the precision given by the massless Dirac operator can be significantly improved to approach the eigenmodes of the exact Bloch operator > Simulation Results and Accuracy Assessment >

Mira: They found that for a cut-off nu equals two is often a good compromise to avoid spectral pollution of these effective models > Simulation Results and Accuracy Assessment >

Lev: A cutoff nu equals two sounds like a practical choice for simulation parameters, but I always wonder if that choice of nu really holds up when you scale things up to the real hardware implementation > Page three > <ref:2602.19185#pg2>

Kai: They also found that the effective operator F one built with perturbative vectors at first order has most of the time better precision than the effective operator F six coming from adding eigenstates at K > Simulation Results and Accuracy Assessment >

Mira: That's interesting because they are comparing different ways to build these families, and their finding is that perturbing with respect to momentum provides better precision when k equals zero and v equals zero > Simulation Results and Accuracy Assessment >

Lev: So the first-order perturbation based model seems more robust than just adding states at K, but we have to remember that those methods are only good for low orders of perturbation theory > Page three > <ref:2602.19185#pg2>

Kai: The paper concludes by stating the goal was to obtain accurate effective operators for the Bloch eigenmodes of the two-scales exact Schrödinger operator (one) > Conclusion and Future Interest > <ref:2602.19185#pg1>

Mira: They say that using these proposed effective operators could be of interest in some applications to model low-energy Dirac fermions when precision is needed > Conclusion and Future Interest >

Lev: For me, it’s good they show how the computation cost for these effective operators is essentially constant in epsilon while it was exponentially increasing with one over epsilon in the exact model > Final Remarks >

Kai: Yeah, so while we get better accuracy, the cost of running these effective operators stays manageable, which is what we wanted when they developed this framework for graphene monolayer superlattices, as they did > Final Remarks >

Mira: So to wrap up on "Some effective operators for graphene monolayer superlattices, from variational perturbation theory". They show that the precision given by the massless Dirac operator can be significantly improved to approach the eigenmodes of the exact Bloch operator > Conclusion and Future Interest >

Lev: For me, I just want to say that this paper lays out a framework that could be of interest in some applications to model low-energy Dirac fermions when precision is needed > Conclusion and Future Interest >

The paper's summary: Kai: So, this paper is about taking graphene in a superlattice setup and trying to find better mathematical tools than just using the standard massless Dirac operator > Mira, it seems like they’re replacing that standard model with something more complex > Lev, so what’s the actual goal here for us on hardware?

Mira: The main idea is that by coupling variational approximation and perturbation theory with a multiscale method, they can construct these effective operators > They aren't just using one simple formula; they are building a whole system to get a more accurate picture of the band structure > Kai, that means we’re looking at how the electrons actually behave when you introduce those periodic potentials in the superlattice setup.

Lev: When you build something like that, you have to worry about what assumptions you're making about those scales and how much error creeps in > So they are using a two-scales reduced basis method, which is basically trying to simplify the problem by picking a specific set of functions to describe the behavior at different levels >

Kai: Right, so they’re using this family F of functions psi one through psi M to nail down the microscopic scale behavior > Mira, what does that actually achieve in terms of accuracy when you compare it to just using the standard operator?

Mira: It lets them impose those microscopic behaviors on their effective model, which is then described by this operator H epsilon k > Kai, and they found that the precision they get can be significantly better than just using the traditional massless Dirac operator > They’re getting closer to what we expect from the exact Bloch operator.

Lev: That sounds promising for simulation fidelity, but you have to remember that those families F depend on how you build them, so the accuracy really hinges on that choice > Kai, is it true they found a specific way to make this work better than others?

Kai: Yes, they showed that the effective operator F one built with perturbative vectors at first order has most of the time better precision than the effective operator F six coming from just adding eigenstates at K > That’s a concrete comparison between two different construction methods.

Mira: They even found that this momentum-dependent, first-order approach is more accurate when the momentum k equals zero and you have no external potential V > That’s a specific condition where their method really shines compared to the simpler models.

Lev: From an error analysis standpoint, they bound the error between their approximation and the exact eigenmodes by P perp F x, which gives us a measurable way to know how good our model is > So we can quantify exactly how much deviation we’re seeing.

Kai: And they also mentioned that a well-chosen cutoff nu equals two is often a good compromise for avoiding spectral pollution in their simulations > That means if you’re running these models, you need to be careful about picking the right parameters to keep the results clean.

Mira: Exactly, it shows they’ve found a way to manage that trade-off between accuracy and computational cost in a way that keeps things manageable for simulation > So they managed to get a better picture of low-energy Dirac fermions without making the computation prohibitively expensive.

Lev: This whole framework is interesting because it shows how you can build these reduced models, and then use Schur reduction on the effective operator H epsilon k to get a simpler form, which is much easier for finding those eigenvectors > That transformation simplifies things down to a two times two matrix-valued operator.

Kai: So they’re not just giving us a new formula; they’re giving us a whole pipeline—from building the basis functions F, through the perturbation theory, to applying Schur reduction—that leads to these improved operators > It’s more than just one better equation.

Mira: And it suggests that for applications modeling low-energy Dirac fermions where precision really matters, this kind of structured approach is something we should be looking at > It opens up a new way to model these materials accurately.

Lev: We should definitely keep an eye on how this structure translates to real hardware constraints because the cost scaling is good, which is a big win compared to the exact operator which was exponentially increasing with one over epsilon >

The paper's improvements: Tom: So, we’re talking about how this paper suggests we can actually improve our current understanding of graphene in those superlattice setups > Kai, what’s the key suggestion they make for making these models better than just using the basic Dirac equation?

Mira: They suggest moving away from just the standard massless Dirac operator toward these derived effective operators like H epsilon k > It’s about replacing that simple operator with one that incorporates more of the physics you actually have in a real superlattice > Lev, what does that mean for your error correction work?

Lev: It means we can start building models that are closer to the truth, which is good because those simpler models often miss crucial details at the band edges where things get really messy > But those effective operators have their own assumptions baked into them.

Kai: Right, and they address that by showing how using variational perturbation theory lets you build these operators based on specific constants from Bloch states at the Dirac point K > Mira, so you’re not just guessing the structure; it’s informed by known physics.

Mira: They show that this structured approach allows them to get better precision for the eigenvectors than when they just add extra eigenstates at K manually > That comparison between F one and F six shows that building things systematically is more effective than adding random states.

Lev: And that systematic improvement is what matters for us, because if the approximation error P perp F x stays low, then our simulation overhead doesn't explode when we try to scale up the system size > But they still need a good cutoff nu equals two to keep those simulations from getting polluted by artificial bands.

Kai: So it’s about finding a balance—getting that higher precision while making sure the computational cost stays within reasonable limits for what we can actually build in cryogenics > It’s about practical improvement, not just theoretical math on a page.

Mira: They also show that this method can handle non-zero potentials V quite well, which is important because real-world devices always have some external landscape influencing the electrons > So it’s not just for the perfect Dirac point; it works for the more complicated scenarios.

Lev: I’m interested in how they handle those momentum derivatives—those directional derivatives with respect to k—that they build into their basis families F > That incorporation of momentum dependence is what allows them to model things away from the Dirac point K accurately, which is where a lot of interesting physics happens.

Kai: It seems like the main improvement here is moving from a static, simple operator to a dynamic one that respects the underlying symmetries and scale separation of the superlattice > So we’re getting better band diagrams and eigenvectors because we are accounting for more physical reality.

Mira: That’s right, it changes how we visualize those low-energy states; it gives us band diagrams that track how things progress as you change parameters like lambda or the lattice constant a zero > It’s a more detailed map of the system's behavior.

Lev: So, for someone who just listens to the show and doesn't do quantum hardware, this paper suggests that we need to think about these multi-scale modeling techniques when we try to predict how graphene behaves in complex environments > It’s a method for getting higher fidelity results in materials science simulations.

Kai: Exactly. And it sets the stage for us to test if these theoretical improvements can actually be realized on our superconducting qubits, which is the ultimate check on whether this math translates to reality.

Conclusion: Kai: So, to wrap up, this paper on "Some effective operators for graphene monolayer superlattices, from variational perturbation theory" shows that we can significantly improve our mathematical models of these systems > It boils down to using these derived operators instead of just sticking with the standard massless Dirac operator.

Mira: Right. The main implication is that we have a more rigorous way to describe the band structure when you introduce those periodic potentials in a superlattice setup > They’ve shown that this method gives better eigenvectors than traditional models, especially when you look at things away from the simple Dirac point K.

Lev: From an error correction standpoint, it’s good because it gives us a quantifiable bound on the error P perp F x, which tells us exactly how accurate our approximation is > We can use that to determine if the overhead of running this effective model is worth the precision gain.

Kai: It means we have a solid framework now to test these models against what we actually build in our cryogenics labs > So, it’s not just theory; it’s a tool for guiding experimental design.

Mira: The caveat is that the accuracy depends heavily on your choice of the basis family F and whether you use the first-order perturbation or something more complex, like F six > You can’t just plug and play a single formula here.

Lev: And for real hardware implementation, we still need to make sure those cutoff parameters like nu equals two are chosen carefully to avoid spectral pollution in our simulations > Those simulation artifacts have to be minimized when we try to map this onto physical chips.

Kai: So, in short, the paper provides a better way to model low-energy Dirac fermions with higher precision by using sophisticated multiscale modeling techniques > It’s an important step forward for any application where you need a detailed understanding of how these materials behave under external influences.

Mira: This work opens the door for more detailed simulations of moiré bands and other complex graphene systems where high fidelity is required > We should look at how this approach can be used to model things like twisted bilayer graphene in more detail.

Lev: I just want to say that while it’s a great theoretical tool, the next step is seeing if we can implement these effective operators on real hardware with low enough error rates to actually matter for quantum computation > That’s where the real engineering challenge begins after this mathematical foundation.

Louis Garrigue

math-ph, cond-mat.mtrl-sci, math.MP, quant-ph

Submitted: 2026-02-22

Updated: 2026-10-28

Code: https://github.com/lgarrigue/superlattice_graphene

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

Importance score: 70/100

The gist: Our goal is to provide precise effective operators for monolayer graphene at Fermi energy by coupling variational approximation, perturbation theory…

Key concepts

Effective Operator Hεk
This is a derived mathematical operator that approximates the complex behavior of graphene at the Fermi energy. It is constructed by coupling a semiclassical Schrödinger operator with a macroscopic potential varying at a specific scale, offering greater accuracy than the simple massless Dirac model.
Two-scales Reduced Basis Method
This technique builds effective models using a small set of periodic functions (a basis F) that mimic the behavior of microscopic eigenfunctions. By reducing the problem to this smaller basis, researchers create simpler, yet highly accurate, models for the full system.
Variational Perturbation Theory
This method is used to systematically build the families of functions (F) used in the reduced basis. It involves expanding eigenmodes in terms of a momentum parameter and calculating higher-order corrections to improve the approximation accuracy.

Terminology

Summary

Our goal is to provide precise effective operators for monolayer graphene at Fermi energy by coupling variational approximation, perturbation theory, and a multiscale method The effective operators derived in this document provide more accurate band diagrams and eigenvectors than the traditional massless Dirac operator

Effective Operator Formulation

The approach involves considering the semiclassical Schrödinger operator hε defined by (1) and coupling it with a macroscopic potential varying at a scale ε−1 ∈ N to create a superlattice At the effective level, the usual massless Dirac operator is replaced by other operators, and simulations are provided for graphene. The exact operator Hεk is studied around the Fermi level at the Dirac point K, leading to Hεk:= Q hε K+εk − EF (10) which acts on L2 per(omega). p. 3

Where the paper says it

We center the study of h ε around the Fermi level, which we assume to be at a the Dirac point K, presenting a conical intersection in the dispersion relation, see [35] for a precise definition. Hence we work on Hε k := Q h ε K+εk − EF Q −1 = 1 2 ε 2 (−i∇ε−1K+k + A) 2 + Qv + εV − EF, (10) which acts on L 2 per(Ω).

Multiscale Modeling and Basis Construction

Effective operators are built using a two-scales reduced basis method where a family F of M linearly independent and periodic functions ψ1,..., ψM ∈ H1 per(omega) is used to impose the behaviors of eigenfunctions at the microscopic scale. p. 4

Where the paper says it

, ψM ∈ H1 per(Ω), F := (ψj ) 1⩽j⩽M . (11) This family will impose the behaviors of eigenfunctions of the reduced space at the microscopic scale.

The resulting effective model for Hεk is given by Hεk:= ε−1M ⊗ 1 + L · ⊗(−i∇k + A) + S ⊗ 1/2 ε(−i∇k + A)2 + V (13) as an operator of CM ⊗ C∞ per(omega). p. 4
Where the paper says it

Efficient families will • have M small so the resulting effective model will be “computable”, • choose ψj in a way so that the resulting effective operator will be accurate. We then define the operator J : CM ⊗ C∞ per(Ω, C) −→ H1 per(Ω, C) α 7−→ PM j=1 αj (x)ψj x ε , which enables to go from envelop functions to real-space functions.

The choice of the family F determines the accuracy, with larger M improving the approximation but increasing computational cost. p. 5
Where the paper says it

While the exact operator Hε k is only defined for ε ∈ 1/N, the effective operators Hε k are defined for any ε > 0. Moreover, the computation cost of the effective operators is essentially constant in ε while it was exponentially increasing with 1/ε in the exact model.

Perturbative Order and Matrix Simplification

The families Fl,k are built using variational perturbation theory, where F0 is (w1, w2), and higher orders involve derivatives with respect to the momentum parameter k. For order one, the matrices S, M, L and T are computed based on specific constants derived from Bloch states at the Dirac point K. p. 5

Where the paper says it

We recall the Pauli matrices σ1 := 0 1 1 0 , σ2 := 0 −i i 0 , σ3 := 1 0 0 −1 . In our choices of F, for any k we will always have ψa = wa for a ∈ {1, 2}, where wa are the Bloch states at the point K, defined in Section 2.

The structure of these matrices is simplified by exploiting graphene symmetries to obtain specific forms for S, M and T. p. 4
Where the paper says it

The general effective operators In this section we present effective operators for two-scales graphene, in a general way. More specific choices will be made in Section 4.

Simulation Results and Accuracy Assessment

Simulations show that the precision given by the massless Dirac operator can be significantly improved to approach the eigenmodes of the exact Bloch operator. The cut-off ν needs to be well-chosen to avoid spectral pollution of the effective models, with ν = 2 often being a good compromise. p. 0

Where the paper says it

We will often take A = 0 for simplicity. It is well-known that the effective operator modeling particles around the Fermi level, for only one valley, is then a massless Dirac operator in the potential V , i.

Furthermore, the effective operator F1 built with perturbative vectors at first order has most of the time better precision than the effective operator F6 coming from adding eigenstates at K. p. 6
Where the paper says it

Labeling the first index of F by the perturbative order, we define, for any direction k ∈ R 2\{0}, F0 := (w1, w2), F1,k := (w1, w2, RDkw1, RDkw2), (17) F1 := (w1, w2, R(−i∂K,1)w1, R(−i∂K,2)w1, R(−i∂K,1)w2, R(−i∂K,2)w2,), F2,k := {wa, RDkwa,(RDk) 2wa − R 2DkP Dkwa}1⩽a⩽2 F2 := {wa, R(−i∂K,b)wa, R2 (−i∂K,b)wa, R(−i∂K,b)R(−i∂K,c)wa}1⩽a,b,c⩽2. Remark that the vectors of the families Fℓ,k depend on k only via the direction k |k| .

Conclusion and Future Interest

The goal was to obtain accurate effective operators for the Bloch eigenmodes of the two-scales exact Schrödinger operator (1). The derivation of effective operators by using variational perturbation theory led to several models, depending on the perturbation order and on the momentum dependence. p. 1

Where the paper says it

For instance, corresponding to order one in perturbation theory, one can add 4 derivatives (two perturbation directions for two states), resulting in a 6 × 6-matrix valued effective operator presented in Section 4. We introduce the problem in Section 2, then in Section 3 we present the general effective operators.

The use of these proposed effective operators could be of interest in some applications to model low-energy Dirac fermions, when precision is needed. p. 4
Where the paper says it

We study the action of the exact operator Hε k in the projected space (12). We want to associate it a low-dimensional model which will reproduce the Fermi eigenmodes with precision.

p. 4
Where the paper says it

We study the action of the exact operator Hε k in the projected space (12). We want to associate it a low-dimensional model which will reproduce the Fermi eigenmodes with precision.

p. 4
Where the paper says it

We study the action of the exact operator Hε k in the projected space (12). We want to associate it a low-dimensional model which will reproduce the Fermi eigenmodes with precision.

Proof Structure

The proof of Proposition 3.1 involves deriving the effective operator Hεk from the exact operator Hεk using methods similar to those in [21]. p. 1

Where the paper says it

To develop effective models in multiscale situations, reduced basis methods were applied in homogenization [17, 26, 46, 65, 67], and ideas from multiscale finite element method [1, 32, 48] were applied to semiclassial Schrödinger operators [24, 60]. In [21], the authors used a variational approximation which produced an effective operator vFσ · (−i∇) + 1 2 ε(−∆) + V .

This includes proving Lemma 6.1 regarding the decoupling between macroscopic and microscopic scales and applying it to the exact operator's action on test functions. The final step involves applying Schur reduction to the effective operator Hεk, leading to a simplified form that is similar to the one derived in [21]. p. 9
Where the paper says it

We recall the formalism of Schur’s reduction in Appendix C. We formally apply it to the effective operator Hε k , where the Schur projection P2 is the projection onto the first two components of CM, the computations are detailed in Section C.

Computational Tools

The Julia code enabling the production of figures for this document can be found on the Github repository The paper utilizes software like DFTK [45] for simulations and relies on established mathematical techniques such as variational perturbation theory and Schur reduction.

Mathematical Derivations

The computation of matrix coefficients in (15) is achieved by using graphene symmetries to compute scalar products involving derivatives, which are detailed in [83]. Lemma 7.3 provides the necessary relations for computing the matrices S, M and T. p. 1

Where the paper says it

GARRIGUE using a parametrized larger scale potential provides an alternative to moiré systems to obtain precise control over the electronic behavior. Moreover, a macroscopic periodic modulation of monolayer graphene is a first step before a better understanding of bilayer moirés [23], relaxed graphene (see [56, p58], [64, eq.

The derivation of these matrices involves complex algebraic manipulations based on the properties of pseudo-periodic functions and directional derivatives.

Formal Derivations

The formal derivation of the perturbation series for the macroscopic functions justifies the choice of adding derivatives with respect to momentum in families F. This involves expanding eigenmodes in ε as ξεq(x, X) = ξ0q(x, X) + εξ1q(x, X) + O(ε2), and showing that the resulting equation has a solution if and only if the right-hand side is orthogonal to Span(g a q). The error between exact and approximate eigenfunctions is bounded by P⊥Fξεq. p. 3

Where the paper says it

The microscopic exact problem. Let us denote by (E q m, u q m) ∈ R × H1 per(Ω) the Bloch eigenpairs of hq := 1 2 (−i∇q) 2 + v, L − periodic, which is bounded from below, for m ∈ N and q ∈ B, sorted such that E q m is increasing with m.

Schur Reduction Application

The Schur reduction applied to the effective operator Hεk, with H ← Hεk(from 13), Y ← S and P ← P2, leads to a simplified form of the operator (24). This reduction allows for the reconstruction of an approximation of the eigenvector of the original operator by applying (1 + rHεk)P2. p. 9

Where the paper says it

Let us denote by PJ the orthogonal projection onto Im J , and P ⊥ J := 1 − PJ . We recall that we denoted by Φ ε k the eigenvector of the exact operator Hε k defined in (10).

Key Findings from Symmetries

The analysis using symmetries in Section 7.4 shows that for certain combinations of operators, such as when Y1 = Y3 and Y2 = 1, specific terms like χ2 are zero. p. 4

Where the paper says it

The general effective operators In this section we present effective operators for two-scales graphene, in a general way. More specific choices will be made in Section 4.

This leads to simplifications in the resulting matrices, such as F11qjm = χ k−2 cos(2θk) − sin(2θk). p. 7
Where the paper says it

1 (Effective matrices for F1,k). In the case F1,k, we have that r, t′ , r′ , s, s′ , χ, γ1, γ2 ∈ R and the matrices defined in (15) are S = t rei2θk re−i2θk t , M = − t ′ r ′ e i2θk r ′ e −i2θk t ′ (20) T =   (t ′1 − s ′σ2) k |k| r ′ e iθk 1 i r ′ e −iθk 1 −i (t ′1 + s ′σ2) k |k|   , (21) and using σ = (σ1, σ2), L =   χ cos(2θk) − sin(2θk) 2γ1e −iθk cos θk sin θk 2γ1e iθk cos θk sin θk χ cos(2θk) − sin(2θk)   + γ2σ.

The analysis also shows that t and s are real, and r is real if ι1ι2 = 1. p. 4
Where the paper says it

Efficient families will • have M small so the resulting effective model will be “computable”, • choose ψj in a way so that the resulting effective operator will be accurate. We then define the operator J : CM ⊗ C∞ per(Ω, C) −→ H1 per(Ω, C) α 7−→ PM j=1 αj (x)ψj x ε , which enables to go from envelop functions to real-space functions.

Error Analysis

The relative distance d(ψ, ϕ) is used to estimate the error between the exact and effective models. The error analysis shows that the precision given by the massless Dirac operator can be significantly improved to approach the eigenmodes of the exact Bloch operator. p. 1

Where the paper says it

It is now used in the physics literature [31, 39], and a mathematical analysis of this method is provided in [40]. To develop an improved variational space for (1), we enrich the microscale functions by not only inserting the Bloch eigenfunctions at the Dirac point, but also their derivatives with respect to the momentum parameter k.

The cut-off ν needs to be well-chosen to avoid spectral pollution of the effective models. p. 1
Where the paper says it

The mathematics literature also contains works on the study of (1), or in the scaling 1 2 (−i∇ + A(εx))2 + v(x) + V (εx). The derivation of effective models for graphene at the macroscopic scale was done using semiclassics techniques in [20, 30, 41, 43], a dynamical approach in [3, 36, 42, 69], and in a spectral approach in [34] in the strong binding regime.

p. 1
Where the paper says it

The mathematics literature also contains works on the study of (1), or in the scaling 1 2 (−i∇ + A(εx))2 + v(x) + V (εx). The derivation of effective models for graphene at the macroscopic scale was done using semiclassics techniques in [20, 30, 41, 43], a dynamical approach in [3, 36, 42, 69], and in a spectral approach in [34] in the strong binding regime.

The precision is often better for F1 than for F6 when k = 0 and V = 0. p. 0
Where the paper says it

The physics literature investigated those doubly periodic systems, called monolayer superlattices, enabling a better control on the propagation of electronic waves for graphene, with periodic electric potentials [5,18,27,37,38,58, 80,81], magnetic ones [11,12,14,28,29,57,76,84,86] for instance to compute the mini-band structures [4,18,49,59,70,71,85], or transport properties [16,19,25]. We will often take A = 0 for simplicity.

Summary of Key Results

The main conclusions brought by the simulations are that the precision given by the massless Dirac operator can be significantly improved to approach the eigenmodes of the exact Bloch operator. The cut-off ν needs to be well-chosen to avoid spectral pollution of the effective models. We are not retaining the k-dependent families F1,k and F2,k because of non-smooth behaviors and lack of accuracy seen in Figures 4, 5 and 8. As seen in Section 5. p. 1

Where the paper says it

We hence do not obtain a 2 × 2-valued matrix operator but an M × M-valued one, where M is the number of functions in the microscale basis. For instance, corresponding to order one in perturbation theory, one can add 4 derivatives (two perturbation directions for two states), resulting in a 6 × 6-matrix valued effective operator presented in Section 4.

6, the effective operator F1 built with perturbative (with respect to the momentum) vectors at first order has most of the time better precision than the effective operator F6 coming from adding eigenstates at K. p. 6
Where the paper says it

Labeling the first index of F by the perturbative order, we define, for any direction k ∈ R 2\{0}, F0 := (w1, w2), F1,k := (w1, w2, RDkw1, RDkw2), (17) F1 := (w1, w2, R(−i∂K,1)w1, R(−i∂K,2)w1, R(−i∂K,1)w2, R(−i∂K,2)w2,), F2,k := {wa, RDkwa,(RDk) 2wa − R 2DkP Dkwa}1⩽a⩽2 F2 := {wa, R(−i∂K,b)wa, R2 (−i∂K,b)wa, R(−i∂K,b)R(−i∂K,c)wa}1⩽a,b,c⩽2. Remark that the vectors of the families Fℓ,k depend on k only via the direction k |k| .

The use of these proposed effective operators could be of interest in some applications to model low-energy Dirac fermions, when precision is needed.

Error Bounds

The error between the exact and approximate eigenfunctions is bounded (up to a constant independent of ε) by P⊥Fξεq. The error in the case where F = F0 is bounded by ε while in the case where F = F1, since δ+Kg a·(k) ∈ Span F1 we have that P⊥F1δ+Kg a·(k) = 0. The error in the case where V is non-zero is also shown to be well reproduced. p. 0

Where the paper says it

We will often take A = 0 for simplicity. It is well-known that the effective operator modeling particles around the Fermi level, for only one valley, is then a massless Dirac operator in the potential V , i.

Numerical Parameters

The lattice constant used in (2) will always be a0 = 5. The standard choice for ε will be ε = 1/7 and the index of the lower Fermi level will always be mF = 1. The numerical values computed using DFTK [45] provide specific parameters for graphene. p. 8

Where the paper says it

Values of the parameters for graphene. We computed the values of those effective parameters with the software DFTK [45], which uses DFT to obtain the vectors u K m, we took the case of physical standard graphene.

Final Remarks

The paper concludes by stating that the effective operators are defined for any ε > 0, and the computation cost of the effective operators is essentially constant in ε while it was exponentially increasing with 1/ε in the exact model. The use of these proposed effective operators could be of interest in some applications to model low-energy Dirac fermions, when precision is needed. The Julia [13] code enabling the production of figures for this document can be found on the Github repository. p. 0

Where the paper says it

In particular, this applies to graphene considered as a one-body model like in DFT. The physics literature investigated those doubly periodic systems, called monolayer superlattices, enabling a better control on the propagation of electronic waves for graphene, with periodic electric potentials [5,18,27,37,38,58, 80,81], magnetic ones [11,12,14,28,29,57,76,84,86] for instance to compute the mini-band structures [4,18,49,59,70,71,85], or transport properties [16,19,25].

The paper acknowledges Eric Cancès for a useful discussion

Appendix Details

Appendix A reviews first order degenerate perturbation theory when the degeneracy is lifted at first order and when it is exactly 2. Appendix B details directional derivatives and the k · p perturbation method. p. 6

Where the paper says it

We have |F0| = 2, |F1,k| = 4, |F1| = |F2,k| = 6, |F2| = 18. As explained in Appendix B, each Fℓ and Fℓ,k contain the directional derivatives of q 7→ u q m

Appendix C recalls the principle of Schur reduction for self-adjoint operators

References

The paper cites numerous works covering topics such as moiré bands in twisted bilayer graphene [15], effective medium approach to electron waves in graphene superlattices [80], and various mathematical methods for reduced basis methods. The references span from foundational texts on perturbation theory [50] to recent works on topological modes in graphene under external periodic potentials [71]. p. 6

Where the paper says it

An important quantity which we will need is the pseudo-inverse R :=    (EF − hK) P⊥L2 per(Ω)→P⊥L2 per(Ω)−1 on P⊥L 2 per(Ω) 0 on P L2 per(Ω) extended on L 2 per(Ω) by linearity, so (EF − hK)R = R(EF − hK) = P⊥. Variational perturbation theory : Dirac modes and their derivatives up to order ℓ.

Specific Operator Relations

The paper provides detailed relations for the matrices S, M and T derived in (15) through Lemma 7.3. These relations are crucial for computing the elements of the effective operator. p. 0

Where the paper says it

Our approach consists in coupling the variational approximation, perturbation theory together with a multiscale method. At the effective level the usual massless Dirac operator is replaced by other operators, and we provide simulations in the case of graphene.

The derivation of these matrices involves complex algebraic manipulations based on the properties of pseudo-periodic functions and directional derivatives. The computation of matrix elements in (46) is derived from Lemma 7. p. 1
Where the paper says it

In Section 4 we take particular cases for deriving effective operators. Finally in Section 5 we display some simulations for the band diagrams, showing that the effective operators derived in this document provide more accurate band diagrams and eigenvectors than the traditional massless Dirac operator.

4.

Numerical Details

The numerical computations use DFTK [45] to obtain the vectors uK, m and specific values for parameters like r, t, s, and vF. The choice of cutoff ν = 2 is used for simulations to obtain desired bands. The parameter λ varies in the study of how the branches coming from the Dirac cone progressively become inaccurate. p. 4

Where the paper says it

In this case, its band diagram is simply a folding of the band diagram of − 1 2∆ + v, (see [33, 55, 62, 73] for the folding/unfolding operations) and the eigenmodes of H ε,V =0 k can be deduced from the ones of − 1 2 (−i∇p)+v. Let use analyze the branches starting at the Dirac point.

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. Effective operator derivation for monolayer graphene superlattices: The system can be improved by replacing standard massless Dirac operators with derived effective operators like Hεk, which provide more accurate band diagrams and eigenvectors than traditional models.

  2. Multiscale modeling of electronic systems: AI systems can be enhanced to perform two-scale reduced basis methods, allowing them to construct low-dimensional models that reproduce Fermi eigenmodes with high precision by using families F of functions derived from Bloch eigenstates and their derivatives.

  3. Momentum derivative incorporation: The AI system can be equipped to compute momentum dependent effective matrices like S, M, L, and T by incorporating directional derivatives with respect to the momentum parameter k, leading to more accurate descriptions of electronic behavior away from the Dirac point.

  4. Spectral pollution avoidance: The system can be optimized for spectral analysis by implementing a cutoff nu in periodic functions C per nu to prevent artificial bands from polluting the spectrum during simulations.

  5. Parameter sensitivity analysis: The AI can perform extensive simulations by varying parameters like coupling constant lambda and lattice constant a0 to observe how the effective operator reproduces exact bands, identifying regimes where the massless Dirac operator provides significant improvements over other models.

  6. Error quantification for effective operators: The system can be trained to estimate errors between exact Bloch operator eigenmodes and approximate eigenmodes of the effective model by analyzing quantities like d(Phi epsilon k, J alpha epsilon k) and comparing perturbation orders (F1 vs F6).

  7. Schur reduction application: The AI can be used to apply Schur reduction to the effective operator Hεk, transforming it into a simpler 2x2 matrix-valued operator Hεk which has no divergence and is easier to work with for eigenvector approximation.

Abstract

Our goal is to provide precise effective operators for monolayer graphene at Fermi energy. We consider the microscopic potential created by a lattice, and add a macroscopic potential with the same periodicity but varying at a scale epsilon-1 in N, creating a superlattice. Our approach combines a variational approximation with perturbation theory and a multiscale method. At the effective level the usual massless Dirac operator is enriched by other operators. We provide simulations for graphene to show the increased accuracy, and a theoretical result proving the improved validity time scale in the dynamical case.

Sources

Related papers