{"id":"bcc524dc-2d10-45b3-9af4-591c600b20c9","arxiv_id":"2608.00254","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"GRISB with up to 21 bath orbitals reproduces DMFT results for spin-orbit-related observables in Sr2RuO4 and shows that U enhances while J suppresses the effective spin-orbit coupling.","lead":"This paper tests a cheaper numerical method, ghost rotationally invariant slave bosons (GRISB), against the gold-standard DMFT approach for a model of the correlated metal Sr2RuO4 with spin-orbit coupling and crystal fields. It finds the fast method matches DMFT for most static and dynamic correlation observables, and that Coulomb U strengthens while Hund coupling J weakens the effective spin-orbit coupling.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline U/J claim rests on N_b=15 finite-bath GRISB at points where no DMFT benchmark exists; since GRISB→DMFT convergence is only conjectured and Z is still drifting at N_b=21, the U-enhances/J-suppresses trend may be a finite-bath artifact.","rationale":"Reader's verdict CONDITIONAL is reasonable. The most load-bearing point is that the central qualitative claim is not directly benchmarked: the finite-bath GRISB results are compared to DMFT at essentially one parameter point, while the U/J scan that produces the headline trend is not. The admitted conjecture in Sec. III.B.3 is therefore not merely a formal gap; it is the hinge of the paper's physical conclusion. Supporting evidence that the finite-bath error is controlled is strong for static quantities (Table I), but Table II shows Z still drifting at N_b=21; since Eq. 26 uses Z, the effective SOC extracted from it may not be converged. This specific concern is consistent with, and slightly sharper than, the reader's weakest assumption. I do not think the paper should be rejected: the method comparisons at the benchmarked point are convincing, and the U/J trends are physically plausible and partly supported by ⟨L·S⟩. But the missing parameter-space benchmark warrants the conditional verdict. No change to the reader's verdict is needed.","tokens_in":20151,"tokens_out":5857,"duration_ms":53994,"concrete_test":"Run GRISB at N_b=21 and DMFT+CTQMC (same Portobello implementation) at N=4 for the same tight-binding model on the U/J grid of Figs. 9–12, e.g. (U,J)=(1.0,0), (2.3,0.4), (3.8,0.4), (2.6,0.2) and the U_2 points near the transition. Extract \\tildeξ_loc from Eq. 26 and ⟨L·S⟩ from the local Green's function in both methods. If the DMFT slopes of \\tildeξ_loc vs U and J do not reproduce the GRISB signs, or if N_b=21 changes the sign/magnitude of the GRISB slopes by >20%, the headline claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline physical result is the U/J dependence of the effective SOC (\"U enhances ξ_eff, J suppresses it\"), shown in Figs. 9–12. These curves are produced by GRISB with a fixed finite bath (mainly N_b=15). The validation against DMFT is limited to a single parameter set for static observables (U=2.6 eV, J=0.4 eV, Table I and Fig. 8) and to one spectral comparison (U=2.3, J=0.4, Fig. 6). The paper itself states in Sec. III.B.3 that convergence of GRISB to DMFT with increasing bath size is only conjectured. Furthermore, Table II shows the quasiparticle weight Z is not converged at N_b=21 (Z_xy still drops 0.25→0.21), so the low-energy quantity \\tildeξ_loc (Eq. 26), which depends on Z, has no demonstrated bath convergence in the same regime. Thus the central trend could be an artifact of the N_b=15 truncation, especially near the metal-insulator transition or half-filling where RISB is shown to fail qualitatively (Sec. IV C). No exact-method check of the U/J slope is provided.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper applies the ghost rotationally invariant slave-boson method (GRISB) to the three-orbital t2g Hubbard-Kanamori model with Sr2RuO4 tight-binding parameters derived from DFT-LDA and LQSGW. It benchmarks convergence in the number of bath ('ghost') orbitals (Nb = 3, 9, 15, 21) against DMFT with CTQMC for spectral functions, off-diagonal self-energies, and static spin-orbit observables, and against ARPES/XAS/quantum-oscillation data. The authors report that most static observables converge quickly, that GRISB reproduces DMFT-like low-energy spectra, and that the effective spin-orbit coupling is enhanced by U and suppressed by J. They also compute Fermi surfaces and the strain-driven Lifshitz transition. Two limitations are acknowledged in the text: the GRISB-to-DMFT convergence is stated to be conjectural (Sec. III.B.3), and the real-frequency Γ-point SOC splitting does not converge with Nb (Table III).","tokens_in":20444,"tokens_out":6837,"duration_ms":67120,"significance":"If the central trends are correct, GRISB would be a valuable low-cost embedding method for spin-orbit-coupling and crystal-field physics, especially for low-temperature parameter scans where DMFT-CTQMC is expensive. The paper's strengths are its independent benchmarks: parameters come from DFT/LQSGW or experiment, results are compared against DMFT/CTQMC and measured Fermi surfaces, and no fitting to the target observables is performed. The DMRG impurity-solver ordering study and the open discussion of QEPack/Portobello reproducibility are useful. However, the abstract's headline 'U enhances, J suppresses SOC' is validated against DMFT at only one or two parameter sets; the U/J curves otherwise rely on a finite bath whose convergence is shown to be incomplete for quasiparticle weights. The physical trend is plausible and testable, but its present support is contingent on additional convergence or DMFT checks.","major_comments":[{"comment":"The claim that 'U enhances the spin-orbit coupling while J suppresses it' is overstated relative to the data. Fig. 9 shows that for N=3 and N=4 the renormalized SOC ξ̃ is nearly unchanged by U, and Fig. 11 states that at the realistic N=4 point U leaves ξ̃ 'insensitive or slightly suppressed'. What survives as a general statement is an increase in |⟨L·S⟩| and in localization, not an increase in ξ̃ itself. Please qualify the abstract and conclusions accordingly, or give direct evidence that at the Sr2RuO4 point the relevant ξ̃ is enhanced by U.","section":"Abstract, Sec. IV.C (Figs. 9 and 11)"},{"comment":"The U/J slopes in Figs. 9–11 are computed at Nb=15 wherever they matter, but no DMFT benchmark exists at those parameters: Table I and Fig. 6 are single parameter sets, and Fig. 8 uses U=2.6 eV, Δ=0 rather than U=2.3 eV, Δ=0.11 eV. Meanwhile Table II shows Z_xy = 0.25→0.21 and Z_yz/xz = 0.33→0.29 between Nb=15 and Nb=21 at U=2.3, J=0.4. Since Eq. (26) involves √Z, ξ̃_loc has not been demonstrated to be converged in the same regime, and Sec. III.B.3 explicitly says GRISB→DMFT convergence is only conjectured. Please provide Nb=21 (or larger) results at two U and two J values along the Fig. 11 curves, or direct DMFT estimates of ξ̃_loc at the endpoints, to rule out a finite-bath artifact in the central trend.","section":"Sec. IV.B/IV.C, Table II, Eq. (26)"},{"comment":"The general statement that 'the addition of ghosts systematically improves the RISB predictions' is not supported for the Γ-point splitting: Table III moves from 0.11 eV (RISB) to 0.04–0.07 eV (GRISB) and does not converge with Nb, whereas DMFT and experiment are near 0.11–0.13 eV. The authors acknowledge this, and I do not treat it as a fatal flaw. However, the abstract's 'GRISB converges quickly for most observables' should be explicitly qualified to exclude real-frequency quantities such as this splitting, which is a central observable of the paper.","section":"Sec. IV.B and Table III"}],"minor_comments":[{"comment":"'the self energy Σ (Eqs. 22 and 22)' should be 'Eqs. (22) and (23)'.","section":"Sec. II, after Eq. (5)"},{"comment":"Typo: 'frequnecy' should be 'frequency'. Also, the caption of Fig. 6 could clarify which GRISB spectra are LDA vs LQSGW-based.","section":"Sec. IV.B"},{"comment":"Please state explicitly in the text that the qualitative U/J trends for realistic N=4 are from Nb=15 GRISB and that Fig. 11 compares RISB (Nb=3) with GRISB (Nb=15); the current captions are ambiguous about the bath sizes used for the trend lines.","section":"Figs. 9–11"},{"comment":"The sentence 'However, at half-filling, the atomic gap is Δat=U+2J, which contradicts our findings' is confusing. Clarify that the contradiction is with the naive atomic-gap expectation, not with the numerical result.","section":"Sec. IV.C"},{"comment":"The row 'ξ̄Γ LDA/GW' is ambiguous: the Nb=3 value is 0.11 and DMFT is listed as ∼0.11. State explicitly whether LDA and LQSGW give identical values or whether only one is shown.","section":"Table III"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is technically careful and unusually transparent about its limitations, but the central 'U enhances, J suppresses SOC' claim needs the additional convergence/DMFT check described in Major Comment 2 before publication. The abstract is somewhat overbroad relative to the presented data. This is a clear revision path, not a rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a careful numerical benchmarking paper. The genuinely new content is the systematic bath-size convergence study for spin-orbit-coupled t2g models, the DMRG bath-ordering analysis, and the strain-driven Lifshitz transition result. The qualitative trend — U enhances effective SOC, J suppresses it — was already in the DMFT literature (Kim et al. 2018, Tamai et al. 2019), so that part is a reproduction rather than a discovery.\n\nWhat the paper does well: the comparisons against DMFT-CTQMC for spectral functions, the off-diagonal Matsubara self-energy, and static quantities like ⟨L·S⟩ are credible and encouraging. Static observables appear converged by N_b=9 (Table I), and the spectral function resembles DMFT by N_b=21. The DMRG orbital-ordering discussion is useful practical work for anyone using tensor-network impurity solvers. The authors are refreshingly transparent about the Γ-point splitting not converging with bath size and about RISB's accidental agreement with experiment. No circular fitting or invented physics.\n\nSoft spots: the headline U/J curves (Figs. 9–12) are mostly computed at N_b=15, with DMFT benchmarks only at one or two parameter sets. Table II shows Z still drifting at N_b=21, so the low-energy ξ~loc defined through √Z could shift with more ghosts. The paper itself states that GRISB→DMFT convergence is conjectured. The stress-test worry about fixed N_b is partly countered by the Fig. 11 caption saying \"under different bath size N_b,\" but the text does not quantify that convergence, and there is no DMFT point at those U/J values to pin down the slope. Adding explicit N_b dependence for ξ~loc and error bars from DMRG/QMC would make the key claims much safer.\n\nMinor: the temperature mismatch (GRISB at 11.6 K vs DMFT at 200–400 K) gets little discussion, and analytical-continuation details are thin. Neither changes the overall picture.\n\nBottom line: the method validation is genuinely useful, the paper is honest, and the central physics isn't new but is correctly attributed. The finite-bath concern is real but not disqualifying. This deserves a serious referee; a revision should add convergence checks and uncertainty estimates for the headline curves.","headline":"Solid GRISB-vs-DMFT benchmark for SOC observables in Sr2RuO4; the U/J trends are reproductions of known DMFT physics, and the finite-bath slope lacks a direct DMFT cross-check.","tokens_in":20971,"tokens_out":4282,"would_cite":true,"duration_ms":39552,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.27.+a","71.70.Ej"],"model":"deepseek-v4-flash","headline":"In a model of Sr2RuO4, Hubbard U enhances the effective spin-orbit coupling while Hund's coupling J suppresses it, and a cheap embedding method (GRISB) reproduces the expensive DMFT results for most static and dynamic observables.","keywords":["ghost rotationally invariant slave-boson method","GRISB","spin-orbit coupling renormalization","Hubbard U","Hund's coupling J","Sr2RuO4","DMFT convergence","Lifshitz transition"],"falsifier":"Compute the effective spin-orbit coupling (e.g., ⟨L·S⟩ or ξ_eff at the lowest Matsubara frequency) with GRISB at N_b=30 or 45 for the same Sr2RuO4 parameters and compare with a converged DMFT-CTQMC solution at the same low temperature; if the difference does not decrease monotonically with N_b, or if the U-enhancement/J-suppression trend reverses at larger bath, the paper's central claim collapses.","tokens_in":20038,"feed_emoji":"⚛️","tokens_out":4417,"duration_ms":38678,"temperature":0.7,"pith_summary":"The paper argues that the ghost rotationally invariant slave-boson method (GRISB) — an embedding scheme that adds 'ghost' bath orbitals to the impurity problem — converges quickly to DMFT-level accuracy for static and many dynamical spin-orbit-related observables in a t2g Hubbard-Kanamori model of Sr2RuO4. Using GRISB to scan a wide parameter range, the authors find a clean dichotomy: the Hubbard interaction U enhances the effective spin-orbit coupling, while Hund's coupling J suppresses it. If correct, this makes GRISB a computationally cheap route to quantitative spin-orbit physics in correlated materials and gives a concrete rule for how interactions renormalize spin-orbit effects.","feed_headline":"Coulomb U enhances spin-orbit coupling; Hund's J suppresses it","feed_subtitle":"Ghost-orbital embedding matches DMFT for Sr2RuO4, showing how U and J shift spin-orbit coupling oppositely.","key_machinery":"Ghost rotationally invariant slave-boson (GRISB) embedding: the lattice problem is represented through a rectangular renormalization matrix R and an impurity model with N_b bath ('ghost') orbitals, solved by exact diagonalization or DMRG at zero temperature. The effective spin-orbit coupling is read off either from the frequency-dependent self-energy (ξ_eff(iω_n)) or from a renormalized local Hamiltonian ĥ_loc built from R and Λ; both serve as diagnostics of how U and J renormalize SOC.","core_discovery":"The central claim is that in the three-orbital t2g model, correlations renormalize the local spin-orbit coupling anisotropically and oppositely with U and J: increasing U enhances the effective SOC (larger |⟨L·S⟩| and larger renormalized ξ extracted from the local Hamiltonian), whereas increasing J suppresses it for every filling studied. GRISB, with 15–21 bath orbitals, reproduces DMFT's static densities, off-diagonal occupancies, ⟨L·S⟩, and the Matsubara-axis self-energy; only real-axis spectral functions at finite frequency remain outside the method's reach. The paper also finds that the Γ-point spin-orbit splitting does not converge monotonically with bath size — RISB happens to match ex","pith_inferences":["If the U/J dichotomy holds generally, it suggests a materials-design knob: in 4d and 5d oxides where SOC is already strong, increasing U will further enhance spin-orbit-driven gaps and anisotropies, while doping or chemistry that raises Hund's coupling will suppress them — a testable trend across the ruthenate and iridate families.","The paper's GRISB-vs-DMFT comparison was done at different temperatures (11.6 K vs 400 K); a direct same-temperature benchmark for the effective SOC would sharpen the method's claimed accuracy for dynamic quantities.","Because the effective SOC is extracted from the static local Hamiltonian ĥ_loc, one could extend the analysis to finite doping or strained heterostructures, mapping how strain and filling modulate the U/J competition in the same framework.","The convergence slowdown for the Γ-point splitting hints that real-axis poles in GRISB are not converged; a practical extension would be to use GRISB to bootstrap a DMFT calculation, initializing the impurity solver with the GRISB self-energy to reduce CTQMC cost."],"forward_implications":["GRISB with a modest number of ghosts can replace DMFT-CTQMC for static SOC observables and for the low-frequency self-energy, cutting computational cost by orders of magnitude in this class of models.","The U-enhances/J-suppresses rule for effective SOC gives a qualitative prediction: materials with larger Hund's coupling will show weaker spin-orbit-induced band splitting and smaller L3/L2 branching ratios at fixed U.","The Lifshitz-transition strain in Sr2RuO4 moves from about -0.9% (noninteracting) to roughly -0.3% in GRISB, approaching the experimental -0.44%, indicating that correlations substantially soften the strain needed to drive the van Hove singularity to the Fermi level.","Static quantities such as orbital occupancies and ⟨L·S⟩ converge with only two ghosts (N_b=9), meaning cheap GRISB runs already capture the correlation-driven redistribution of electrons across orbitals.","The Γ-point splitting in GRISB does not converge to experiment as ghosts are added, so real-frequency quantities should be interpreted with care; RISB's apparent agreement there is coincidental error cancellation."],"fun_headline_variants":["U boosts spin-orbit coupling, J dampens it in Sr2RuO4","Ghost slave-boson method matches DMFT for Sr2RuO4","Correlations tune spin-orbit: U and J act oppositely","Spin-orbit coupling: U enhances, J suppresses in Sr2RuO4","U and J have opposite effects on spin-orbit coupling in Sr2RuO4"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The benchmarking treats finite-bath GRISB as an approximation to DMFT based on a conjecture that GRISB results converge to DMFT as the number of ghosts grows; if that convergence fails for SOC/CFS observables, the U/J renormalization trends lose their reference point.","fun_headline_variants_meta":{"raw":{"variants":["U boosts spin-orbit coupling, J dampens it in Sr2RuO4","Ghost slave-boson method matches DMFT for Sr2RuO4","Correlations tune spin-orbit: U and J act oppositely","Spin-orbit coupling: U enhances, J suppresses in Sr2RuO4","U and J have opposite effects on spin-orbit coupling in Sr2RuO4"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000656,"raw_usage":{"total_tokens":2864,"prompt_tokens":794,"completion_tokens":2070,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":1965}},"tokens_in":538,"tokens_out":2070,"duration_ms":12852,"temperature":1.0,"reasoning_tokens":1965,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T00:53:33.814285+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the effective spin-orbit coupling (e.g., ⟨L·S⟩ or ξ_eff at the lowest Matsubara frequency) with GRISB at N_b=30 or 45 for the same Sr2RuO4 parameters and compare with a converged DMFT-CTQMC solution at the same low temperature; if the difference does not decrease monotonically with N_b, or if the U-enhancement/J-suppression trend reverses at larger bath, the paper's central claim collapses.","supporting_citations":[],"review_version":1}