Intrinsic Wannier Functions for Hamiltonian downfolding

summary

The gist

Downfolding ab initio material band structure into a low-energy subspace spanned by orbitals of specified atomic character is a common task in the simulation of complex materials and this work…

In short

The Intrinsic Wannier Function (IWF) method is a new, non-iterative technique for downfolding electronic band structures into a low-energy subspace defined by specific atomic characters. Unlike standard methods, IWF produces high-quality Wannier functions that retain the original atomic orbital character of the input basis, offering robustness and compatibility with high-throughput calculations.

Key concepts

Intrinsic Atomic Orbitals (IAO)
These are the reference minimal basis of atomic orbitals used as input for the IWF method. They serve as a foundation to ensure that the resulting localized Wannier functions maintain strong atomic character, making them physically interpretable representations of chemical bonds and orbitals.
Intrinsic Wannier Function (IWF)
This is a construction scheme that creates localized orbitals directly from the molecular band structure without needing iterative protocols. It uses a projection method to map the bands onto a subspace spanned by the chosen atomic characters, providing an efficient way to obtain downfolded structures.
Character-based Smearing
When energy gaps between bands are small, entanglement occurs. To handle this, IWF uses a character-based smearing function instead of energy-based smearing. This procedure helps create a 'diabatic basis' approximation, ensuring that the resulting Wannier functions accurately represent the band structure even when bands are close together.

Terminology used across episodes

This episode discusses

The paper

Intrinsic Wannier Functions for Hamiltonian downfolding · Read on arXiv

Shuoxue Li, Garnet Kin-Lic Chan

California Institute of Technology · Marcus Center for Theoretical Chemistry

Downfolding ab initio material band structure into a low-energy subspace spanned by orbitals of specified atomic character, a procedure known as Wannier downfolding, is a common task in the simulation of complex materials. We introduce the Intrinsic Wannier Function (IWF) method to Wannierize bands with given atomic character. The method is non-iterative and requires only a single dimensionless parameter to disentangle bands. In benchmarks on silicon, graphene, and the three-band model of a mercury cuprate, we show that Intrinsic Wannier Functions provide high quality downfolded band structures compared to those from standard approaches such as Maximally Localized Wannier Functions, the Selected Columns of the Density Matrix method, and Projectability-Disentangled Wannier Functions. Further, their straightforward implementation and robustness position Intrinsic Wannier Functions as a general and useful tool for Wannier downfolding in materials electronic structure and in high-throughput applications.

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Intrinsic Wannier Functions for Hamiltonian downfolding".

Mira: Here's the gist of the paper. Downfolding ab initio material band structure into a low-energy subspace spanned by orbitals of specified atomic character is a common task in the simulation of complex materials and this work introduces the Intrinsic Wannier Function (IWF) method which provides high-quality downfolded band structures compared to standard approaches like Maximally Localized Wannier Functions and the Selected Columns of the Density Matrix method The Intrinsic Wannier Function (IWF) method is presented as a simple alternative scheme to existing downfolding procedures, motivated by the concept of Intrinsic Atomic Orbitals (IAO) The IWF construction creates localized orbitals without the need for any iterative protocol, unlike MLWF Unlike SCDM, the input to IWF is a reference minimal basis of atomic orbitals which ensures that the resulting Wannier functions retain good atomic character and allows for fine control of which bands one wishes to reproduce Theoretical Framework The construction begins with a molecular setting where the molecular orbitals become bands that carry crystal momentum labels, i k⟩, and the B2 basis is a crystalline/Bloch AO basis The Intrinsic Wannier Functions at each k-point are constructed as µ IWF,k⟩ = orth

Oµ˜k: (2) This construction ensures that diagonalizing the projected Hamiltonian PˆIWFhˆPˆIWF will exactly reproduce the band-structure of the selected bands Handling Entanglement If the bands to be downfolded are not well separated energetically, entanglement of the band structure leads to atomic characters being spread out across different numbers of conduction bands at different k points To obtain a 'diabatic basis', a character-based smearing function is applied instead of energy-based smearing The character of the bands i is defined as χki = Pµ′ ⟨i kµ′k⟩2 ∈

0, 1: Comparative Results In benchmarks on silicon, graphene, and the three-band model of a mercury cuprate, the IWF method provides high-quality downfolded band structures compared to those from standard approaches such as Maximally Localized Wannier Functions and the Selected Columns of the Density Matrix method For silicon, IWF functions retain the AO character of the B2 basis, clearly resembling 3s and 3p orbitals In graphene, both IWF and MLWF yield smooth bands identical to the original bands below the Fermi level For the three-band downfolding of HgBa2CaCu2O6, IWF successfully recovers a part of the band structure that MLWF fails to reproduce without requiring an adjustment of an energy window The Intrinsic Wannier Function (IWF) method provides a simple way to downfold band structure given a target set of atomic characters The robustness of the procedure, and the dimensionless nature of the smearing parameter, mean that material-specific adjustments to the calculation are minimal The algorithm is thus compatible with high throughput calculations Acknowledgements We thank Zhi-Hao Cui, Linqing Peng, Petra Shih, Tianyu Zhu and Yuhang Ai for providing enlightening suggestions for the conceptualization of this project We also thank Arash Mostofi, Jaemo Lihm, Vitale Valerio, and Anil Damle for additional insights into SCDM This work was supported by the US Department of Defense Multidisciplinary University Research Initiative (MURI) Grant No.

Kai: W911NF2410292 The source code is available in the Github repository https://github.com/LiShuoxue/iwf-preview.git References

1: N.

Mira: Plakida, High-temperature cuprate superconductors, 2010th ed., Springer Series in Solid-State Sciences (Springer, Berlin, Germany, 2010)

2: T.

Kai: V.

Title and authors: Mira: Now, moving into the title and authors of this paper, "Intrinsic Wannier Functions for Hamiltonian downfolding," we see that the authors are Shuoxue Li and Garnet Kin-Lic Chan.

Kai: I’m curious about who these people are in this context. Are they specialists in the quantum hardware side, or more on the theoretical materials side?

Mira: They’re from a division of Chemistry and Chemical Engineering at California Institute of Technology, which suggests a strong background bridging both areas. Their work is focused on making these downfolding procedures more robust and chemically meaningful.

Lev: I wonder if their focus leans more toward the theoretical modeling aspect, or if they have some insight into how these methods translate to actual simulation overheads on real quantum hardware.

Kai: Since this paper focuses heavily on the construction of the orbitals themselves, I’d expect a strong theoretical materials background. But I want to know if they've considered how easy it is for someone working with a physical system to implement this without getting stuck in unnecessary complexity.

Mira: The paper itself seems very focused on the mathematical construction, emphasizing how the method avoids iteration and requires only one dimensionless parameter, which points toward a strong theoretical focus on simplifying the process.

Lev: That simplicity is what matters for implementation overhead. If it’s just one parameter and non-iterative, that translates to much lower overhead when you are trying to run many different calculations sequentially on real hardware.

Kai: So this paper seems aimed at making the calculation itself more practical, not just finding a new way to calculate something that is computationally intensive for its own sake.

Mira: Exactly. The authors are showing how you can achieve high quality downfolded band structures while maintaining a direct connection to the underlying atomic orbital character.

Lev: That connection is key because when you move from theory to actual hardware, having those physically interpretable functions helps a lot with validation and understanding what’s happening in the simulation.

Kai: So we have the theory here, and Lev is asking how that translates into something tangible for our experimental or simulation pipeline.

Mira: That’s right. The whole point of this work is to provide a method that doesn't require you to guess where to set your energy windows or how many iterations you need to run, which simplifies the workflow immensely.

The paper's summary: Kai: Let’s talk about what the paper actually summarizes. The Intrinsic Wannier Function for Hamiltonian downfolding method is essentially a new way to project bands into a low-energy subspace spanned by orbitals with specified atomic character.

Mira: They are using this to solve the common task of Wannier downfolding in complex materials, and they introduce IWF as an alternative to existing techniques like Maximally Localized Wannier Functions and the Selected Columns of the Density Matrix method.

Lev: The summary highlights that IWF is non-iterative, meaning it doesn't need that iterative protocol you often see with MLWF, which is a huge plus for scalability.

Kai: Right. And they show that they can produce high quality downfolded band structures on benchmarks like silicon, graphene, and the three-band model of a mercury cuprate using this approach.

Mira: They also introduce character based smearing when bands are entangled, which is a specific improvement over standard energy based smearing to keep the character information preserved.

Lev: So it’s not just about getting good results; it’s about *how* they get them—by carefully controlling the construction of the localized orbitals from a minimal basis of atomic orbitals.

Kai: It sounds like the key is using that initial basis, which ensures the resulting Wannier functions retain good atomic character, which is a direct benefit for interpretation.

Mira: And this method allows you to fine control exactly which bands you want to reproduce by selecting them upfront in a way that’s not dependent on energy windows.

Lev: That control over band selection without tuning the energy window is something we really need when dealing with correlated systems where the bands can overlap and become entangled.

The paper's improvements: Kai: Let’s look at the specific improvements they highlight. First, they emphasize that IWF avoids iterative protocols entirely, which means no need for those iterations you’d normally run with MLWF.

Mira: That non-iterative construction is a major step because it simplifies the setup and makes it much faster to run when you are trying to simulate large systems.

Lev: For us, that translates directly to less overhead when we’re trying to push these calculations onto larger, more demanding quantum hardware because we don't have convergence issues slowing us down.

Kai: They also introduce the character based smearing procedure instead of energy based smearing when bands are not well separated energetically. That’s a specific fix for entanglement.

Mira: That’s where I think the paper is strongest—they use a character based smearing function to create a diabatic basis approximation, which is much better than just using energy-based smearing in those situations.

Lev: So they are proposing that this character based selection allows for a diabatic basis approximation when you need it, which should give us more reliable results when bands are crossing over.

Kai: And finally, the fact that the IWF functions retain the input atomic orbital character is a major win for physical interpretation. They clearly resemble orbitals like 3s and 3p in silicon benchmarks <ref:2608.15557#pg2>.

Mira: That retention of atomic character is crucial because it gives us confidence that we are downfolding into something physically meaningful, not just some mathematical artifact.

Conclusion: Lev: So to wrap up on the paper "Intrinsic Wannier Functions for Hamiltonian downfolding," the main implication is that this method provides a simple, robust way to do what was traditionally done with Wannier downfolding.

Kai: It seems like it’s a general and useful tool for Wannier downfolding in materials electronic structure and in high-throughput applications because of its straightforward implementation and dimensionless smearing parameter.

Mira: I think the paper gives us a clear framework: define your target atomic characters, and the method builds those orbitals without needing tedious energy window adjustments or many iterations.

Lev: It’s solid for running on larger systems because it bypasses those convergence headaches, which is something we really need when moving toward real hardware simulations.

Kai: The IWF method gives us a clear path to get those localized orbitals while keeping the physics of the original bands intact in these benchmarks.

Mira: It’s definitely a valuable contribution for anyone needing a more reliable way to go from ab initio band structure to an effective model Hamiltonian without relying on energy window tuning.

Lev: We’ll keep this paper in mind as we look at how it compares when we try to adapt it for simulating the more complex correlated systems we are currently working on.

Kai: Yeah, that sounds like a good plan, let's see what the next arXiv submission is.

More episodes

← Home