Utility-Scale Quantum Computation of Ground-State Energy in a 100+ Site Planar Kagome Antiferromagnet via Hamiltonian Engineering

Muhammad Ahsan

REVIEW 4 major objections 5 minor 48 references

Utility-Scale Quantum Computation of Ground-State Energy in a 100+ Site Planar Kagome Antiferromagnet via Hamiltonian Engineering

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that a 103-site kagome antiferromagnet ground-state energy can be estimated on current superconducting quantum processors, and that the per-site estimate $-0.417J$ matches the thermodynamic-limit benchmark after…

desk verdict A real 103-qubit VQE run on a Kagome lattice, but the headline energy is stitched together from fitted couplings, a mismatched ODR reference, and post hoc selection—so the benchmark match is not evidence. read the letter →

arxiv 2507.06361 v3 pith:QX23U5S3 submitted 2025-07-08 quant-ph cs.ET

Muhammad Ahsan This is my paper

classification quant-phcs.ET

keywords kagomeantiferromagnetvariationalquantumeigensolverHamiltonianengineeringground-stateenergyspinliquidnoisemitigationfrustratedmagnetismsuperconductingprocessor

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a 103-site planar kagome antiferromagnetic Heisenberg lattice is within reach of current superconducting quantum processors, and that the measured per-site ground-state energy of $-0.417J$ is consistent with the thermodynamic-limit value $-0.4386J$ once open-boundary effects are removed. The route is a hybrid VQE: small overlapping patches are optimized classically, then stitched into a 103-qubit ansatz whose only trainable parameters are six junction rotations tuned on hardware. The central trick is Hamiltonian engineering: raising one exchange bond on each defect triangle to about twice the natural coupling, which mimics the loop-flip dimer resonances of a spin liquid and lets a shallow single-layer circuit reach accurate energies. If the claim holds, frustrated two-dimensional magnetism becomes a realistic target for near-term quantum simulation despite shallow circuits and limited connectivity.

What carries the argument

The load-bearing object is the engineered Heisenberg Hamiltonian $H_{\rm pert}$ of Eq. (1): on each 'defect triangle' selected by an initial static valence-bond (dimer) covering, one exchange bond is strengthened from $J=1$ to $J'\approx 2$, while all other bonds stay at $J$. The calibration is done on small patches of 6 to 23 sites by matching VQE energies to exact diagonalization, and the same $J'\approx 1.9$ to $2.0$ transfers across patches. Physically, the enhanced bond makes a loop-flip around the defect triangle energetically accessible, so the shallow single-repetition real-amplitude circuit, a chain of $R_y$ rotations and nearest-neighbor CNOT gates, can form a superposition of dimer covers instead of freezing into the static dimer configuration. Around this sits a two-stage scheme: local VQE optimizes 15- to 19-qubit subcircuits classically, stitch junctions add six $R_y(\theta)$ gates, global VQE on hardware optimizes only those six parameters using a truncated cost Hamiltonian $H_{\rm SEL}$ of junction-neighbor terms, and the final noisy expectation is post-processed with the ODR rescaling $\text{noise-mitified}\,\mathrm{Tr}(\rho H_{\rm pert}) = \langle\phi|H|\phi\rangle / \mathrm{Tr}(\sigma H) \cdot \mathrm{Tr}(\rho H_{\rm pert})$, where $|\phi\rangle$ is a Clifford-reachable static dimer state.

What would settle it

Compute the exact ground-state energy, via a high-accuracy tensor-network calculation, of the engineered Hamiltonian $H_{\rm pert}$ with $J'=2$ on a 36- or 48-site planar kagome patch and compare it to the true KAFH ground-state energy on the same patch; if the per-site difference grows with system size, or if dimer correlations change qualitatively, the engineered Hamiltonian is not representing the original model.

Watch

Extended reading notes

Core claim

The paper's central claim is that a spin-1/2 kagome antiferromagnetic Heisenberg model on 103 sites can be treated as a utility-scale quantum-computation target. Using a single-repetition hardware-efficient ansatz on superconducting processors, the paper reports a total energy of $-172.4J$ in its four-times convention, or $-0.417J$ per site. After an open-boundary correction that removes the contribution of the 25 edge sites, the bulk per-site energy becomes $-0.4386J$, the published DMRG thermodynamic-limit value. The paper also reports that quantum-optimized junction parameters produce lower raw energies than parameters from a classical matrix-product-state simulation with circuit cutting on every processor tested, and that two noise-mitigated values, $-168.62J$ and $-172.4J$, lie below the classical baseline $-167.95J$ even after accounting for error bars. Stated on its own terms, the discovery is that a shallow 1D-entanglement ansatz, guided by locally calibrated Hamiltonian engineering, can estimate the KAFH ground-state energy at a size far beyond exact classical treatment.

Load-bearing premise

The argument stands or falls on the assumption that raising one exchange bond on each defect triangle to $J'\approx 2$ yields an engineered Hamiltonian whose low-energy sector faithfully represents the original kagome ground state at 103 sites, even though the calibration is verified only on patches of 6 to 23 sites.

Editorial extensions

If this is right

If the central claim is correct, a single-repetition hardware-efficient ansatz with only six global parameters can reach the KAFH ground-state estimate on a 103-site open-boundary lattice, so circuit depth is not the bottleneck for this class of frustrated magnets.
The same calibrated value $J'\approx 1.9$ to $2.0$ worked on patches from 6 to 23 sites, which implies the defect-triangle engineering recipe transfers to larger lattices without per-geometry retuning.
Because the quantum-optimized junction parameters gave consistently lower raw energies than the classical MPS-optimized set on every processor tested, hardware-in-the-loop parameter optimization can capture correlations that approximate classical simulations miss.
The noise-mitigated values $-168.62J$ and $-172.4J$ sitting below the classical baseline $-167.95J$ mean the quantum result is not merely reproducing a classically computable answer; it is adding information beyond the tensor-network approximation.
At larger lattice sizes the fraction of boundary sites shrinks, so the paper's own scaling argument implies future per-site estimates should approach the thermodynamic-limit value without needing an open-boundary correction.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The paper leaves implicit that the same $J'\approx 2$ prescription could be tested directly on the engineered Hamiltonian itself: a tensor-network ground state of $H_{\rm pert}$ on a 36-site patch would separate 'does the ansatz reach the ground state of the modified model' from 'is the modified model the right physics.'
A natural extension is to apply defect-triangle tuning as a general-purpose variational-error-mitigation tool for frustrated lattices beyond kagome, since the mechanism only needs a local reference configuration whose loop-flip resonances a shallow ansatz cannot express.
The quoted open-boundary correction propagates edge-site energies estimated from 12- to 23-site benchmarks; repeating Eq. (3) with edge energies taken from the 103-site geometry itself would reveal how much of the match to $-0.4386J$ is built into the correction.
The ODR rescaling assumes a depolarizing noise channel; testing whether the ratio $\langle\phi|H|\phi\rangle/\mathrm{Tr}(\sigma H)$ is stable across several Clifford reference states on the same processor would indicate how much of the final estimate inherits that assumption.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The genuinely new thing here is scale: a 103-qubit VQE experiment on a planar Kagome lattice, run on IBM hardware, with the ansatz split into classically optimized local segments and a few tunable junction gates. That is a real engineering achievement, and the open-source code is a plus. The local/global decomposition and the idea of strengthening defect-triangle bonds to mimic loop-flip physics are worth discussing, even if the physical motivation is hand-wavy.

The problem is the central number. The per-site energy -0.417J is not an independent estimate. The J' values are fitted so that small-patch VQE matches exact diagonalization, so the local energies are trained to the answer. The ODR correction in Eq. (2) is applied to Tr(ρHpert), but the reference expectation uses the original H, not Hpert. When Hpert changes 13 bonds, the reference value is not −147 for the engineered Hamiltonian. The ratio is therefore not a controlled rescaling for the quantity being measured. On top of that, the -36 adjustment for the 22 omitted sites appears without derivation, and the final result is the best among many processors and parameter sets, so there is a clear multiple-comparisons bias. The raw hardware values sit 30–40% above the classical benchmark; the ODR factor supplies the entire downward shift. Given all that, the close match to -0.4386J after an open-boundary correction that uses edge energies from the same small fitted systems is exactly what you would expect from a post hoc assembly, not from a measured ground state.

The soft spots are load-bearing, not cosmetic. But the paper is not worthless: it demonstrates that current hardware can run a 100+ qubit variational circuit for a 2D frustrated system and that careful noise mitigation can produce stable, reproducible expectation values for a reduced Hamiltonian. That is useful for the NISQ community, and the authors are transparent about their procedures.

Who should read it? People working on large-scale VQE experiments and error mitigation will get value from the experimental details, but readers should treat the physics claim as unproven. I would send it to peer review because the scale and the method are significant enough to warrant referee time, but the reviewers should push for a major reframing: present it as a hardware demonstration of hybrid local/global VQE, and remove the claim that the DMRG benchmark has been reproduced.

Referee Report

4 major / 5 minor

Summary. The paper reports a hybrid local/global VQE estimate of the ground-state energy per site of a 103-site kagome antiferromagnetic Heisenberg (KAFH) model, using IBM Heron processors. The method splits a hardware-efficient ansatz into classically optimized local segments, recombines them on quantum hardware with six junction angles optimized by global VQE on a truncated Hamiltonian, and applies an ODR noise-mitigation rescaling. The headline result is a per-site energy of -0.417J, which, after an open-boundary correction for a 125-site lattice containing 25 edge sites, is claimed to match the DMRG thermodynamic-limit value -0.4386J. The paper also advertises the approach as a scalable route to utility-scale quantum simulation of frustrated 2D magnets.

Significance. If the central claim were established, this would be a notable experimental milestone: a variational ground-state energy estimate on a 100+ qubit two-dimensional frustrated spin system, with open-source code, measurements on five IBM processors, and a systematic comparison against classical MPS and exact diagonalization. The hybrid local/global decomposition and the bond-strength engineering idea are potentially interesting. However, the headline numerical result depends on several load-bearing elements that are not justified in the manuscript, including an unexplained -36J correction term, an inconsistent site-counting and arithmetic conversion, an ODR reference that is not matched to the target observable, and post hoc selection of the best hardware result. As written, the paper does not provide a controlled estimate of the KAFH ground-state energy.

major comments (4)

[Section VI, Eq. (3)] The reported per-site value -0.417J does not follow from the stated arithmetic. The text gives '-172.4J/4 = -52.1J' and '-521.J/103 = -0.417J', but -172.4/4 = -43.1J and -43.1/103 ≈ -0.418J, while -52.1/103 ≈ -0.506J. The value -0.417J emerges only if one first subtracts the unexplained -36J term (in 4x units) and divides by 125 sites: (-172.4 - 36)/4/125 ≈ -0.417J. The -36J term for the 22 sites excluded from the quantum circuit (Figure 2) is never derived or referenced, and the denominator is used inconsistently (103 vs 125 sites). This correction changes the per-site energy by roughly 0.07J, which is comparable to the claimed agreement with -0.4386J, so the central estimate is currently unsupported.
[Section IV.C, Eq. (2)] The ODR rescaling is not a controlled correction for the target observable. The reference expectation values used in Eq. (2) are ⟨ϕ|H|ϕ⟩ = -147 and Tr(σH), both for the original Hamiltonian H, while the quantity to be mitigated is Tr(ρHpert), where Hpert differs from H on the 13 defect bonds (Eq. (1)). If any enhanced bond overlaps the dimer cover defining |ϕ⟩, as is the case for the engineered defect triangles, then ⟨ϕ|Hpert|ϕ⟩ ≠ ⟨ϕ|H|ϕ⟩ and the scalar ratio is not an anchor for the noise affecting Tr(ρHpert). A valid ODR application would require ⟨ϕ|Hpert|ϕ⟩ and Tr(σHpert). Furthermore, the noise-mitigated values in Table III are reported without error bars, so the statement in Section V.B that -172.4 is 'statistically significant' is not supported.
[Section V, Tables III and V] The best value -172.32 is selected post hoc as the minimum over multiple processors and multiple parameter sets. The text states that among several minima found during global VQE, parameter sets achieving minimal objective values were selected, and Table V lists many candidate parameter sets. Table III then reports, for each processor, the parameter set yielding the minimum noise-mitigated Tr(ρHpert). Taking the minimum of many noisy estimates introduces a downward selection bias, and the comparison to the classical MPS value -167.95 in Section V.B and Figure 5 is a multiple-comparison claim rather than a single pre-specified measurement. The paper should report all runs, state the selection rule before data analysis, or apply a proper multiple-testing correction.
[Section III.B] The extension of the calibrated coupling J' ≈ 2 from small patches to the 103-site system is not quantitatively supported. J' is tuned so that local VQE matches exact diagonalization on subregions of 6-23 sites (Table I), and Section III.B justifies the extrapolation by locality and self-similarity. However, the final energy is an expectation value of Hpert, not of H, and the paper provides no numerical evidence (e.g., comparison of low-lying spectra of H and Hpert on intermediate-size clusters, or DMRG checks on larger patches) that the low-energy sector of Hpert faithfully represents the KAFH ground state at the target size. Without such evidence, the comparison of the measured value to the DMRG benchmark -0.4386J is not physically meaningful.

minor comments (5)

[Section VI] There are arithmetic and sign errors: '-172.4J/4 = -52.1J' should be '-43.1J', and 'Etotal = 52.1J' is missing a minus sign.
[Section IV.B.2] The definition of HSEL contains a duplicate pair (47,48) and the listed set may be missing some intended junction bonds; please check the edge list against the circuit in Figure 3.
[Figure 5] The statistical comparison is unclear: noise-mitigated values are plotted without error bars, while unmitigated values are binned, and the text says '7 out of 10 bars have a mean value lower than -167.95' but does not define the test used.
[References] Reference [25] is a competition blog rather than a peer-reviewed publication; please cite the formal version if one exists.
[Section III.C] The phrase 'Javg =≈ 4/3' is ambiguous and should be rewritten to state the average coupling on defect triangles unambiguously.

Circularity Check

3 steps flagged · score 6.0 of 10

The benchmark match is reached through fitted J' calibrations, an unexplained -36 J offset, and an ODR rescaling whose reference is evaluated on the wrong Hamiltonian.

fitted input called prediction [Section III.A/III.B, Table I]
"Each subcircuit is independently optimized, with J ′ adjusted to match known ground-state energies from exact diagonalization (feasible at these small scales). ... For each subregion size N , classical VQE simulations with J ′ ≈ 2 yield ground-state energies in close agreement with exact diagonalization. ... The data highlights that the same calibrated interaction strength J ′ suffices across different lattice geometries without the need for re-tuning, reinforcing the scalability of the method."

Table I is titled 'Effectiveness of Hamiltonian engineering with J′ ≈ 2 in matching the estimated gs energy to the exact value', and Section IV.A says J′ is 'individually calibrated to ensure that the computed local ground-state energy matches the exact ground-state energy obtained via exact diagonalization.' The close agreement is therefore enforced by the fit, not discovered. The inference that this J′ transfers to the 103-site lattice is an unproved locality/self-similarity assumption, and the small-patch 'validation' cannot support it because the same fitted values are being reported as evidence.
fitted input called prediction [Section VI.A, Eq. (3)]
"To use the eq(3), revert gsenergy to the un-scaled version and estimate the average edge-site energy ¯Eedge using small lattice benchmarks, where exact diagonalization is possible for given Etotal. ... we estimate that ¯Eedge lies within the range: ¯Eedge ∈ [−0.3076J, −0.3296J]. ... For 125-site lattice our gs energy becomes −172.5 − 36 = −208.4J, which gives Etotal = 52.1J. ... eq(2) implies ¯Ebulk(max) = − 0.4386J, ¯Ebulk(min) = − 0.4441J ... closely match the one obtained in the thermodynamic limit −0.4386J."

The edge-site energy is fitted from exact diagonalization energies of 12-, 19-, and 23-site clusters, and the 125-site total is obtained by adding an unexplained '-36' to the 103-site measured total. Inserting these numbers into Eq. (3) produces the literature value -0.4386J. No derivation of the '-36' term is given anywhere; the benchmark match is thus an arithmetic consequence of chosen inputs rather than an independent confirmation of the quantum estimate. Changing that unexplained constant would move the final 'prediction' off the benchmark.

1 more flagged steps

other [Section IV.C, Eq. (2), and Table III]
"According to the ODR formalism, the noise-mitigated expectation value of Hpert for the final VQE state ρ is given by: noise-mitigated Tr(ρHpert) = ⟨ϕ|H|ϕ⟩ / Tr(σH) · Tr(ρHpert) (2). ... the Clifford version of the reference ansatz prepares a static dimer cover state |ϕ⟩. On ideal (noiseless) quantum computer the circuit yields the expectation value ⟨ϕ|H|ϕ⟩ = −147."

The reference ratio uses H, not the engineered Hpert of Eq. (1), even though the target observable is Hpert with 13 defect bonds changed to J' ≈ 1.9-2.0. A correct ODR anchor would require ⟨ϕ|Hpert|ϕ⟩ and Tr(σHpert). As written, the 'noise-mitigated' value is exactly the raw measured Tr(ρHpert) multiplied by a constant derived from a different Hamiltonian; in Table III this constant turns -113.67 into -172.32. The reported estimate is therefore constructed by rescaling the raw value with a mismatched reference, so it does not independently certify that the ground-state energy of Hpert was measured.

full rationale

The paper contains no load-bearing self-citation chain or imported uniqueness theorem, so the circularity is not of the 'self-citation forces the result' type. The central problem is that several quantities that are later presented as validation are actually fitted inputs. J' is calibrated to exact-diagonalization energies on small patches and then reported as if the agreement validated the calibration; the edge-energy correction is extracted from the same kind of small exact diagonalizations; and the 125-site total used in Eq. (3) includes an unexplained '-36' term that is essential for landing on -0.4386J. In addition, the ODR correction in Eq. (2) compares ideal and noisy reference expectation values of H, although the target is Hpert; this makes the noise-mitigated number a rescaling of the raw hardware value rather than a controlled estimate. The hardware execution is real and the raw energies are reported, so the paper is not a pure definitional tautology; however, the headline benchmark match is substantially constructed from the fitted J' values, the fitted edge energies, and the unexplained -36 offset. I therefore score 6: one or more of the 'predictions' reduce, by construction, to earlier fitted inputs, making the central benchmark comparison partially circular. Additional problems such as post hoc selection of the lowest run and multiple-comparison issues are correctness concerns but were not counted as circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on calibrated couplings, an unexplained omitted-site correction, an edge-energy range tied to the same calibration, and a specific noise model. These inputs, not a self-contained derivation, produce the benchmark match.

free parameters (4)

J' enhanced exchange coupling = 1.9 to 2.0 J
Tuned on small Kagome patches so VQE energies match exact diagonalization values (Table I, Table II); used for all defect triangles.
-36 scaled energy correction = -36 (in 4x convention)
Added for 22 alphabetically labeled sites omitted from the quantum circuit; no derivation is given in Section VI.
Eedge edge-site per-site energy = [-0.3076J, -0.3296J]
Estimated from 12-, 19-, and 23-site exact diagonalization cases, the same small systems used for J' calibration; central to boundary correction.
global VQE junction angles theta = e.g., p2 = [5.34, 6.39, -1.24, 5.52, -2.31, 2.03]
Six variational parameters optimized on hardware to minimize truncated HSEL; final reported energy depends on selected parameter set.

assumptions (4)

domain assumption Locality and self-similarity of the KAFH: J' tuned on small patches remains effective for 103-site lattice.
Section III.B: 'we expect the tuning of J' to remain effective as system size increases.'
domain assumption Depolarizing noise model for ODR correction.
Section IV.C: 'Under the assumption of a depolarizing noise model, ODR corrects...'
domain assumption Static dimer reference state with ideal energy -147 is a valid anchor for noise rescaling.
Section IV.C defines reference ansatz; if reference and target noise factors differ, eq (2) is invalid.
domain assumption Shallow 1D CNOT ansatz plus only six junction gates has sufficient expressiveness to represent the relevant ground-state subspace after Hamiltonian engineering.
Sections III and IV.B; no entanglement quantification or expressibility benchmark beyond small-patch energy matching.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Utility-Scale Quantum Computation of Ground-State Energy in a 100+ Site Planar Kagome Antiferromagnet via Hamiltonian Engineering." pith.science (2026). https://pith.science/paper/QX23U5S3

@misc{pith2026250706361,
  author       = {Pith},
  title        = {Pith review of: Utility-Scale Quantum Computation of Ground-State Energy in a 100+ Site Planar Kagome Antiferromagnet via Hamiltonian Engineering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QX23U5S3}},
  note         = {Machine review of arXiv:2507.06361}
}

abstract

We present experimental quantum computation of the ground-state energy in a 103-site flat Kagome lattice under the antiferromagnetic Heisenberg model (KAFH), with IBM's Heron r1 and Heron r2 quantum processors. For spin-1/2 KAFH, our per-site ground-state energy estimate is $-0.417\,J$, which, under open-boundary corrections, matches the energy in the thermodynamic limit, i.e., $-0.4386\,J$. To achieve this, we used a hybrid approach that splits the conventional Variational Quantum Eigensolver (VQE) into local (classical) and global (quantum) components for efficient hardware utilization. More importantly, we introduce a Hamiltonian engineering strategy that increases coupling on defect triangles to mimic loop-flip dynamics, allowing us to simplify the ansatz while retaining computational accuracy. Using a single-repetition, hardware-efficient ansatz, we entangle up to 103 qubits with high fidelity to determine the Hamiltonian's lowest eigenvalue. This work demonstrates the scalability of VQE for frustrated 2D systems and lays the foundation for future studies using deeper ansatz circuits and larger lattices on utility quantum processors.

Figures

Figures reproduced from arXiv: 2507.06361 by the authors.

**Figure 2.** FIG. 2. (Color online) The 125-site planar Kagome lattice showing static dimer coverings (thick black line-segments) and defect [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** FIG. 3. (Color online) Reconstructed ansatz circuit showing: (1) [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: FIG. 4. Global VQE execution on the IBM quantum processors [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

**Figure 5.** Figure 5: FIG. 5. Bar chart showing a statistical comparison of the gs [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: FIG. 6. (Color online) (a) VQE ansatz circuit with qubit indices mapped to the corresponding lattice sites on the Kagome [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]

**Figure 7.** Figure 7: FIG. 7. (Color online) Example subregions of the Kagome lattice used to demonstrate the effectiveness of Hamiltonian engi [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]

**Figure 8.** Figure 8: FIG. 8. (Color online) Subregions of the Kagome lattice of Figure-2 used in local VQE, with corresponding ground-state energy [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 38 canonical work pages

[1]

To implement this approach efficiently, we partition our VQE algorithm into local and global phases

This adjustment compensates the relatively shallow, hardware-efficient ansatz [3] for their lower expressive- ness, particularly our single-repetition real-amplitude cir- cuit supporting only 1D nearest-neighbor entanglement, and helps induce quantum fluctuations in the vicinity of defect triangles, enabling a superposition of dimer covers reminiscent of ...

work page
[2]

Utility-Scale Quantum Computation of Ground-State Energy in a 100+ Site Planar Kagome Antiferromagnet via Hamiltonian Engineering

Note that throughout this manuscript, we assume exchange coupling strength J = 1 and, unless other- wise specified, report four times (4x) the ground-state energy for convenience in calculations and data presenta- tion. This means that singlet energy is taken −3 instead of −3/4. Unlike prior studies constrained to quasi-1D geometries (e.g., cylinders or s...

work page Pith review arXiv 2025
[3]

Local VQE produces pre-optimized ansatz segments for each subregion of the Kagome lattice

Reconstruction of the Ansatz Circuit. Local VQE produces pre-optimized ansatz segments for each subregion of the Kagome lattice. To recover the full 103-qubit circuit, these segments are concate- nated such that all original gate and qubit dependencies of the ansatz are preserved, effectively reconstructing the near original circuit. However, this na ¨ ıv...

work page
[4]

Objective Function Concatenating seven local ansatz segments introduces six free parameters corresponding to the newly inserted Ry(θ) gates on the junction qubits (e.g., qubits 17, 33, 48, 63, 73, 92). The global VQE runs the reconstructed ansatz on IBM quantum processors, with the objective of minimizing Tr(ρHpert), where Hpert comes from eq(1) and ρ is ...

work page
[5]

Bharti et al., Rev

K. Bharti et al., Rev. Mod. Phys. 94, 015004 (2022)

work page 2022
[6]

Cao et al., Chem

Y. Cao et al., Chem. Rev. 119, 10856 (2019)

work page 2019
[7]

Kandala et al., Nature 549, 242 (2017)

A. Kandala et al., Nature 549, 242 (2017)

work page 2017
[8]

IBM Quantum, Device Documentation for Falcon and Hummingbird processors (2025)

work page 2025

Show all 48 references

[9]

Savary and L

L. Savary and L. Balents, Rep. Prog. Phys. 80, 016502 (2017)

2017
[10]

Q. Luo, J. Zhao, H. Y. Kee, and G. Chen, npj Quantum Materials 6, 57 (2021). doi:10.1038/s41535-021-00356-z

2021 doi
[11]

S. Yan, D. A. Huse, and S. R. White, Science 332, 1173 (2011)

2011
[12]

P. W. Anderson, Science 235, 1196 (1987)

1987
[13]

Sachdev, Phys

S. Sachdev, Phys. Rev. B 45, 12377 (1992)

1992
[14]

Cincio, K

L. Cincio, K. Rudinger, M. Sarovar, and P. J. Coles, PRX Quantum 2, 010324 (2021)

2021
[15]

Balents, Nature 464, 199 (2010)

L. Balents, Nature 464, 199 (2010)

2010
[16]

H. C. Jiang, Z. Y. Weng, and D. N. Sheng, Phys. Rev. Lett. 101, 117203 (2008)

2008
[17]

Depenbrock, I

S. Depenbrock, I. P. McCulloch, and U. Schollw¨ ock, Phys. Rev. Lett. 109, 067201 (2012)

2012
[18]

Singh, D

B. Singh, D. Kumar, V. Kumar, M. Vogl, S. Wurmehl, S. Aswartham, B. B¨ uchner, and P. Kumar, Phys. Rev. B 104, 134402 (2021)

2021
[19]

Peruzzo et al., Nat

A. Peruzzo et al., Nat. Commun. 5, 4213 (2014)

2014
[20]

J. R. McClean et al., New J. Phys. 18, 023023 (2016)

2016
[21]

J. I. Colless et al., Phys. Rev. X 8, 011021 (2018)

2018
[22]

G. G. Guerreschi and A. Y. Matsuura, Sci. Rep. 7, 16369 (2017)

2017
[23]

Zhang, L

Y. Zhang, L. Cincio, C. F. A. Negre, P. J. Coles, and S. M. Mniszewski, npj Quantum Inf. 8, 96 (2022)

2022
[24]

M. S. Jattana, F. Jin, H. De Raedt, and K. Michielsen, Front. Phys. 10, 907160 (2022)

2022
[25]

Kattem¨ olle and J

J. Kattem¨ olle and J. van Wezel, Phys. Rev. Research3, 033205 (2021)

2021
[26]

J. L. Bosse and A. Montanaro, Phys. Rev. B 105, 094409 (2022)

2022
[27]

Qian and M

X. Qian and M. Qin, Phys. Rev. B 109, L161103 (2024)

2024
[28]

Javanmard, U

Y. Javanmard, U. Liaubaite, T. J. Osborne, X. Xu, and M. H. Yung, APL Quantum 1, 041101 (2024)

2024
[29]

Weaving, A

T. Weaving, A. Ralli, and V. Wimalaweera (Univer- sity College London), IBM Quantum Open Science Prize 2023, accessed at https://www.ibm.com/quantum/blog/ ibm-quantum-open-science-winners-2023 . Their work achieved sub-1% error in computing the ground-state en- ergy of the spi...

2023
[30]

Weaving, A

T. Weaving, A. Ralli, V. Wimalaweera, P. J. Love, and P. V. Coveney, arXiv:2506.12391

arXiv
[31]

Broholm, R

C. Broholm, R. J. Cava, S. A. Kivelson, D. G. Nocera, M. R. Norman, and T. Senthil, Science 367, eaay0668 (2020)

2020
[32]

Zeng and J

C. Zeng and J. B. Marston, Phys. Rev. B52, 4467 (1995). 12

1995
[33]

Q. Wang, H. Lei, Y. Qi, et al., npj Quantum Mater. 10, 72 (2025)

2025
[34]

R. C. Farrell, M. Illa, A. N. Ciavarella, and M. J. Savage, PRX Quantum 5, 020315 (2024)

2024
[35]

Javadi-Abhari et al., arXiv:2405.08810

A. Javadi-Abhari et al., arXiv:2405.08810

arXiv
[36]

Ahsan, S

M. Ahsan, S. A. Z. Naqvi, and H. Anwer, Phys. Rev. A 105, 022428 (2022)

2022
[37]

Preskill, Quantum 2, 79 (2018)

J. Preskill, Quantum 2, 79 (2018)

2018
[38]

K. M. Nakanishi, K. Fujii, and S. Todo, Phys. Rev. Res. 2, 043158 (2020)

2020
[39]

S. R. White, Phys. Rev. B 48, 10345 (1993)

1993
[40]

A. Lowe, M. Medvidovic, A. Hayes, L. J. O’Riordan, T. R. Bromley, J. M. Arrazola, and N. Killoran, Quantum 7, 934 (2023)

2023
[41]

Ahsan, R

M. Ahsan, R. Van Meter, and J. Kim, ACM J. Emerg. Technol. Comput. Syst. 12, 4 (2015), doi:10.1145/2830570

2015 doi
[42]

Patra, S

S. Patra, S. S. Jahromi, S. Singh, and R. Or´ us,Phys. Rev. Res. 6, 013326 (2024)

2024
[43]

B¨ aumer, V

E. B¨ aumer, V. Tripathi, D. S. Wang, P. Rall, E. H. Chen, S. Majumder, A. Seif, and Z. K. Minev, PRX Quantum 5, 030339 (2024)

2024
[44]

Sachdeva et al., arXiv:2406.01743 [quant-ph] (2024)

N. Sachdeva et al., arXiv:2406.01743 [quant-ph] (2024)

2024 arXiv
[45]

Tindall, M

J. Tindall, M. Fishman, E. M. Stoudenmire, and D. Sels, PRX Quantum 5, 010308 (2024)

2024
[46]

T. A. Chowdhury, K. Yu, M. A. Shamim, M. L. Kabir, and R. S. Sufian, Phys. Rev. Res. 6, 033107 (2024)

2024
[47]

Nishimoto, N

S. Nishimoto, N. Shibata, and C. Hotta, Nat. Commun. 4, 2287 (2013)

2013
[48]

Shibata and C

N. Shibata and C. Hotta, Phys. Rev. B 84, 115116 (2011). Appendix A: Supplementary material TABLE V. Optimal parameters computed using global VQE for different IBM quantum processors. These are obtained from the experiments data of Figure-4. Processor θ1 θ2 θ3 θ4 θ5 θ6 [5.05 -...

2011

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

To implement this approach efficiently, we partition our VQE algorithm into local and global phases

This adjustment compensates the relatively shallow, hardware-efficient ansatz [3] for their lower expressive- ness, particularly our single-repetition real-amplitude cir- cuit supporting only 1D nearest-neighbor entanglement, and helps induce quantum fluctuations in the vicinity of defect triangles, enabling a superposition of dimer covers reminiscent of ...

work page

[2] [2]

Utility-Scale Quantum Computation of Ground-State Energy in a 100+ Site Planar Kagome Antiferromagnet via Hamiltonian Engineering

Note that throughout this manuscript, we assume exchange coupling strength J = 1 and, unless other- wise specified, report four times (4x) the ground-state energy for convenience in calculations and data presenta- tion. This means that singlet energy is taken −3 instead of −3/4. Unlike prior studies constrained to quasi-1D geometries (e.g., cylinders or s...

work page Pith review arXiv 2025

[3] [3]

Local VQE produces pre-optimized ansatz segments for each subregion of the Kagome lattice

Reconstruction of the Ansatz Circuit. Local VQE produces pre-optimized ansatz segments for each subregion of the Kagome lattice. To recover the full 103-qubit circuit, these segments are concate- nated such that all original gate and qubit dependencies of the ansatz are preserved, effectively reconstructing the near original circuit. However, this na ¨ ıv...

work page

[4] [4]

Objective Function Concatenating seven local ansatz segments introduces six free parameters corresponding to the newly inserted Ry(θ) gates on the junction qubits (e.g., qubits 17, 33, 48, 63, 73, 92). The global VQE runs the reconstructed ansatz on IBM quantum processors, with the objective of minimizing Tr(ρHpert), where Hpert comes from eq(1) and ρ is ...

work page

[5] [5]

Bharti et al., Rev

K. Bharti et al., Rev. Mod. Phys. 94, 015004 (2022)

work page 2022

[6] [6]

Cao et al., Chem

Y. Cao et al., Chem. Rev. 119, 10856 (2019)

work page 2019

[7] [7]

Kandala et al., Nature 549, 242 (2017)

A. Kandala et al., Nature 549, 242 (2017)

work page 2017

[8] [8]

IBM Quantum, Device Documentation for Falcon and Hummingbird processors (2025)

work page 2025