Constrained field-space estimators for background-field lattice QCD observables

arXiv:2611.10773 · hep-lat, hep-ph · Submitted 2026-11-07 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Lattice. The day in lattice field theory: what the simulations put on the table.

Kai: We're looking at the paper: "Constrained field-space estimators for background-field lattice QCD observables".

Mira: Here's the gist of the paper. field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths This work compares fixed cancellation stencils with covariance-optimized, generalized-least-squares estimators of the same coefficient using archival nucleon double ratios and decuplet correlators The results distinguish statistical precision from robustness to omitted response modes Constrained Estimator Formulation The extraction problem is reorganized from a numerical-analysis perspective, where the desired response coefficient is treated as a derivative or a coefficient of a specified weak-field basis evaluated at zero field This viewpoint makes field redundancy explicit, allowing additional field values to impose further response mode constraints The mathematics involves defining a reduced field response as y(B) = θ + K X−1 k=1 akϕk(B) + r(B) The linear estimator is then given by θb = X i wiyi = wT y Preserving the target coefficient requires X i wi = 1, while annihilating each selected nuisance mode requires X i wiϕk(Bi) = 0, k = 1,..., K − 1 The minimum-variance estimator within a given constraint family is obtained from min w wT Σw subject to A T w = b For nonsingular Σ and full-rank constraints, the resulting best linear unbiased estimator (BLUE) has weights wBLUE = Σ−1A A T Σ −1A − 1 b Role of Covariance Optimization and Omitted-Mode Leverage The constrained formulation exposes the nuisance-mode annihilation conditions, makes the n − K remaining estimator degrees of freedom for n fields explicit, and permits those directions to be allocated independently to further systematic cancellation or statistical optimization If the true reduced response contains an additional mode a⋆ϕ⋆(B), its contribution to the estimator is δθb⋆ = a⋆wT ϕ⋆, (ϕ⋆)i = ϕ⋆(Bi) The scalar wT ϕ⋆ therefore measures the estimator’s leverage on that omitted response mode A statistically stable minimum-variance solution can still be systematically fragile if this leverage is large, which is the mechanism behind Peelle’s pertinent puzzle Fixed Field-Space Hierarchy The fixed estimators are algebraically related to conventional response fits, and the constrained formulation exposes the nuisance-mode annihilation conditions The fixed O4 stencil is obtained by fitting y(B) and y(2B) exactly to θ + aB2 and evaluating the intercept The distinction made here is operational rather than algebraic, as it makes the n − K remaining estimator degrees of freedom explicit Null-Space Specification Tests and Covariance Noise The null-space statistic bounds discrepancies among estimators in the same constraint family For O4, A = (1, x), so the null space has dimension two and the test has two degrees of freedom With the temporal covariance held fixed, O4 is rejected for every ensemble and O6 for every neutron ensemble With that noise included, O4 remains rejected for every ensemble except the lightest neutron (p = 0.053), whereas O6 is rejected for none (p = 0.31–0.65) for the proton Covariance-Optimized Estimators With four fields and the O4 constraints, two weight degrees of freedom remain, and the minimum-variance (BLUE) solution is the correlated linear fit in B2 to all four fields The displacement between any two O4-constrained estimators can be written exactly The optimized BLUE-O4 proton moments lie about one marginal standard error below fixed O4 Known-Answer Controls: The Decuplet The decuplet correlators provide cases in which the answer is known exactly, and on µ∆+ /µ∆++ = 1/2, a higher-order response, plausibly field-induced mixing with the nucleon displaces the lowest-order estimate by up to 25 standard errors The fixed O4–O8 stencils remove it to within errors at the 10−4 level in the production windows, while the correlated linear fit in B2 misses the identity by 3–17% The fixed O(B4)–O(B8) stencils are nearly coincident and cost almost nothing in precision because all fields share the same gauge configurations Weight-free null-space tests resolve departures from the O4 response family in most nucleon ensembles Covariance optimization buys little precision, with 1–2% for the nucleon moments at the five lighter masses and 7–8% at the heaviest The decuplet correlators supply targets known exactly from the lattice contractions Data and Code Availability The derived field-response vectors, the aligned jackknife samples (fixed and nested), the bootstrap and synthetic-study outputs, the field-space covariance matrices and full-precision optimized weights, and the scripts that generate every table and figure are archived at Zenodo The underlying correlators are available from the authors upon reasonable request Acknowledgments We thank Andrei Alexandru for writing the propagator and contraction code with which the correlators analyzed here were produced This work was supported in part by the U.S.

Kai: Department of Energy under Grant No.

Mira: DEFG02-95ER40907 Appendix A: Archival Reconstruction The poly11-small proton archive contains 992 correlator files The original poly0 proton notebook uses t = 8–18 Reconstructing its 90-configuration double-ratio pipeline reproduces all 24 stored central field-fit exponents with maximum absolute discrepancy 1.3 × 10−9 Appendix B: The Decuplet Identity at the Level of the Correlators With S˜ = Cγ3 S Cγ3 and a contraction kernel F(S1, S2, S˜ 3; α, β) that is linear in its weights, the code evaluates C∆++ = F(Su, Su, S˜ u; 2, 4) At zero field Su = Sd = S and the two are equal configuration by configuration In the background field, Sq = S + eq δS + O(e2q) with the same S and δS for both flavors Appendix C: Charged-Particle Response and a Nonanalytic Stencil A charged spin-zero state illustrates why the cancellation basis must follow the physics Preserving a constant while cancelling 1/B, B, and B2 at B, 2B, 4B, 8B yields w = -8/3, 2 3/3, −13/3, 2 cubed This is a Laurent-series stencil in B Appendix D: Dependence on Ensemble Size The magnitude of the leverage grows with N (Fig.

Kai: 2) Larger samples do not drive the covariance optimum toward the low-leverage stencil Appendix E: Known-Covariance Synthetic Study The population covariance is the noise-inclusive field-space covariance of the lightest proton or neutron BLUE-O4 has the same constraints as O4 but is displaced by up to 4.7σ, with a root-mean-square error up to 4.8 times that of O6 Appendix F: Field Design The geometric spacing does not produce the displacement Because Σ = V KV T makes the field-space covariance equivalent across designs by construction, this is a property of the model rather than a test Appendix G: Discussion The field-space viewpoint separates four questions often conflated in a single correlated fit The frontier of Sec.

Title and authors: Kai: We're moving on now into the specific details of this paper, "Constrained field-space estimators for background-field lattice QCD observables." The authors are Scott Moerschbacher and Frank X. Lee.

Mira: They set up this entire framework by treating the extraction of those weak-field response coefficients as a linear estimation problem in field space, which is a key way they organize the analysis.

Lev: I’m looking at their approach to defining that reduced field response as y of B equals theta plus K times X inverse k equals one ak phi k of B plus r of B. That looks like a sophisticated way to handle redundancy.

Kai: Exactly, and they derive the linear estimator, which they call theta b equal x i wiyi, where preserving the target coefficient requires x i w i equals one.

Mira: But annihilating each selected nuisance mode requires x i w i phi k of B equals zero for all those modes k from one to K minus one.

Lev: So they’re trying to find the minimum variance estimator within that constraint family by solving a minimization problem subject to those constraints on the weights w.

Kai: If you have nonsingular sigma and full rank constraints, then this leads to a best linear unbiased estimator, or BLUE, with weights w blue equals sigma inverse A A transpose sigma inverse minus one b.

Mira: That BLUE solution is what they use when they’re doing the covariance-optimized version of the estimator. It's the minimum variance solution given those specific constraints.

Lev: From my perspective, that mathematical setup is how you translate a physical constraint—like keeping a response coefficient fixed—into an optimization problem you can actually solve computationally on a massive dataset.

Kai: And they then spend a lot of time describing the role of covariance optimization and omitted mode leverage in Section III. That’s where the real meat of their analysis lies for us right now.

Mira: They show that if there's an additional mode, say a star phi star, its contribution to your estimator is delta theta b star equal a star wT phi star times the inner product of phi star with B.

Lev: So the scalar wT phi star tells you how much leverage your estimator has on that omitted response mode, and if it’s large, that solution can still be fragile.

Kai: It really highlights Peelle's puzzle; a statistically stable minimum variance solution can still be systematically fragile if that leverage is high.

The paper's summary: Mira: So to summarize the main takeaway from "Constrained field-space estimators for background-field lattice QCD observables," they are comparing fixed cancellation stencils against these covariance-optimized estimators using real data.

Kai: They show that the fixed O(B four)–O(B eight) stencils are nearly coincident, which is a strong result because it means they cost almost nothing in precision across different field strengths because all fields share the same gauge configurations <ref:2611.10773#pg1,the fixed O(B 4 )–O(B 8 ) stencils are nearly coincident>.

Lev: That's good news for systematic error control; if you use these fixed stencils, you don't need to worry too much about the precise field strength used as long as the gauge setup is consistent.

Mira: However, when they look at null-space specification tests, they find that O4 rejections do not uniquely identify a physical B4 magnetic term in most nucleon ensembles.

Kai: They also found that whether those tests reject the three term response model depends on the temporal fits and on the noise of the estimated temporal covariance.

Lev: That suggests that relying solely on a null-space statistic isn't enough; you need to be careful about your temporal analysis systematics, otherwise, you might misinterpret what’s happening.

Mira: And when they look at covariance optimization results, they found that for the nucleon moments at the five lighter masses, the gain is around one point two percent.

Kai: So precision doesn't always increase dramatically just by optimizing your weights; it depends heavily on which mass range you are looking at.

The paper's improvements: Lev: One of the main suggested improvements is the constrained formulation itself, because it reorganizing the problem from a numerical-analysis perspective makes field redundancy explicit.

Kai: That reformulation lets you treat the desired response coefficient as a derivative or a coefficient of a specified weak-field basis evaluated at zero field.

Mira: This viewpoint makes field redundancy explicit, allowing additional field values to impose further response mode constraints on the estimation process.

Lev: It’s like you’re not just fitting one curve; you're using every measurement to place extra restrictions on the shape of that curve, which is a lot more information.

Kai: The fixed O4 stencil is obtained by fitting y of B and y of 2B exactly to theta plus a B two and then evaluating the intercept <ref:2611.10773#pg1>.

Mira: And they show this is algebraically related to conventional response fits, but the authors say the distinction they make here is operational rather than algebraic.

Lev: So it’s about how you actually set up your fit procedure, not some deep mathematical identity that holds true regardless of what kind of fitting you choose.

Kai: The constrained formulation also makes those n minus K remaining estimator degrees of freedom for n fields explicit, which is a huge step in understanding the structure.

Mira: And it permits those directions to be allocated independently to further systematic cancellation or statistical optimization, depending on what you need most.

Conclusion: Kai: So wrapping up this paper "Constrained field-space estimators for background-field lattice QCD observables," the main takeaway is that the fixed O(B four)–O(B eight) stencils are nearly coincident and cost almost nothing in precision because all fields share the same gauge configurations <ref:2611.10773#pg1,the fixed O(B 4 )–O(B 8 ) stencils are nearly coincident>.

Mira: And while null-space tests are useful, they don't uniquely identify a physical B4 magnetic term in most nucleon ensembles, and the results for covariance optimization show that precision gains are modest.

Lev: I think what matters is that the decuplet correlators provide targets known exactly from lattice contractions, which gives us exact identities like muDelta plus over muDelta plus plus equal one over two.

Kai: Those exact answers from the decuplet correlators supply targets for testing higher-order response consistency, and that’s a real win for checking our theory.

Mira: So in short, this paper shows how to use a constrained field-space approach to compare different estimators on real data, giving us more explicit control over statistical stability and systematic stability against missing physics.

Lev: And for me, the noise issue with the temporal covariance means we have to be cautious when using those null-space tests if we want reliable results.

Kai: Thanks for joining us today. We'll be back next time to talk about new arXiv papers on quantum hardware and error correction developments.

Scott Moerschbacher, *, Frank X. Lee, †

Open Inquiry Research Group · Physics Department, The George Washington University

hep-lat, hep-ph

Submitted: 2026-11-07

Updated: 2026-11-07

Comments: 19 pages, 3 figures, 12 tables. Code and derived data: doi:10.5281/zenodo.23869797

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

Importance score: 92/100

The gist: Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths This work compares fixed cancellation stencils with…

Key concepts

Constrained Estimator Formulation
This method treats the desired response coefficient as a derivative evaluated at zero field. By imposing constraints on the field basis, it allows researchers to explicitly see which response modes are being controlled and which remain free, leading to a minimum-variance estimator (BLUE) that is mathematically derived from these constraints.
Omitted-Mode Leverage
This measures how much the chosen estimator depends on an unmeasured or 'omitted' response mode. If this leverage is large, even a statistically stable minimum-variance solution can be systematically fragile, which is a key mechanism explored in the paper regarding systematic errors.
Null-Space Specification Tests
These tests use statistics to bound discrepancies between different estimators within the same constraint family. For instance, testing against the O4 response family often rejects it for most nucleon ensembles, indicating departures from that specific physical model in the data.
Decuplet Identity
These are exact identities derived from decuplet correlators where certain ratios of observables are known exactly (e.g., $ ext{muDelta}^+ / ext{muDelta}^{++} = 1/2$). These serve as precise targets to test the consistency of higher-order response estimates against the lattice data.

Terminology

Summary

Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths This work compares fixed cancellation stencils with covariance-optimized, generalized-least-squares estimators of the same coefficient using archival nucleon double ratios and decuplet correlators The results distinguish statistical precision from robustness to omitted response modes

Constrained Estimator Formulation

The extraction problem is reorganized from a numerical-analysis perspective, where the desired response coefficient is treated as a derivative or a coefficient of a specified weak-field basis evaluated at zero field This viewpoint makes field redundancy explicit, allowing additional field values to impose further response mode constraints The mathematics involves defining a reduced field response as y(B) = θ + K X−1 k=1 akϕk(B) + r(B) The linear estimator is then given by θb = X i wiyi = wT y Preserving the target coefficient requires X i wi = 1, while annihilating each selected nuisance mode requires X i wiϕk(Bi) = 0, k = 1,..., K − 1 The minimum-variance estimator within a given constraint family is obtained from min w wT Σw subject to A T w = b For nonsingular Σ and full-rank constraints, the resulting best linear unbiased estimator (BLUE) has weights wBLUE = Σ−1A A T Σ −1A − 1 b

Role of Covariance Optimization and Omitted-Mode Leverage

The constrained formulation exposes the nuisance-mode annihilation conditions, makes the n − K remaining estimator degrees of freedom for n fields explicit, and permits those directions to be allocated independently to further systematic cancellation or statistical optimization If the true reduced response contains an additional mode a⋆ϕ⋆(B), its contribution to the estimator is δθb⋆ = a⋆wT ϕ⋆, (ϕ⋆)i = ϕ⋆(Bi) The scalar wT ϕ⋆ therefore measures the estimator’s leverage on that omitted response mode A statistically stable minimum-variance solution can still be systematically fragile if this leverage is large, which is the mechanism behind Peelle’s pertinent puzzle

Fixed Field-Space Hierarchy

The fixed estimators are algebraically related to conventional response fits, and the constrained formulation exposes the nuisance-mode annihilation conditions The fixed O4 stencil is obtained by fitting y(B) and y(2B) exactly to θ + aB2 and evaluating the intercept The distinction made here is operational rather than algebraic, as it makes the n − K remaining estimator degrees of freedom explicit

Null-Space Specification Tests and Covariance Noise

The null-space statistic bounds discrepancies among estimators in the same constraint family For O4, A = (1, x), so the null space has dimension two and the test has two degrees of freedom With the temporal covariance held fixed, O4 is rejected for every ensemble and O6 for every neutron ensemble With that noise included, O4 remains rejected for every ensemble except the lightest neutron (p = 0.053), whereas O6 is rejected for none (p = 0. p. 5

Where the paper says it

With that noise included, O4 remains rejected for every ensemble except the lightest neutron (p = 0. 053), whereas O6 is rejected for none (p = 0.

31–0.65) for the proton

Covariance-Optimized Estimators

With four fields and the O4 constraints, two weight degrees of freedom remain, and the minimum-variance (BLUE) solution is the correlated linear fit in B2 to all four fields The displacement between any two O4-constrained estimators can be written exactly The optimized BLUE-O4 proton moments lie about one marginal standard error below fixed O4

Known-Answer Controls: The Decuplet

The decuplet correlators provide cases in which the answer is known exactly, and on µ∆+ /µ∆++ = 1/2, a higher-order response, plausibly field-induced mixing with the nucleon displaces the lowest-order estimate by up to 25 standard errors The fixed O4–O8 stencils remove it to within errors at the 10−4 level in the production windows, while the correlated linear fit in B2 misses the identity by 3–17%

Conclusion

The fixed O(B4)–O(B8) stencils are nearly coincident and cost almost nothing in precision because all fields share the same gauge configurations Weight-free null-space tests resolve departures from the O4 response family in most nucleon ensembles Covariance optimization buys little precision, with 1–2% for the nucleon moments at the five lighter masses and 7–8% at the heaviest The decuplet correlators supply targets known exactly from the lattice contractions

Data and Code Availability

The derived field-response vectors, the aligned jackknife samples (fixed and nested), the bootstrap and synthetic-study outputs, the field-space covariance matrices and full-precision optimized weights, and the scripts that generate every table and figure are archived at Zenodo The underlying correlators are available from the authors upon reasonable request

Acknowledgments

We thank Andrei Alexandru for writing the propagator and contraction code with which the correlators analyzed here were produced This work was supported in part by the U.S. Department of Energy under Grant No. p. 0

Where the paper says it

Constrained field-space estimators for background-field lattice QCD observables Scott Moerschbacher1, ∗ and Frank X. Lee2, † 1Open Inquiry Research Group, Monrovia, MD 21770, USA 2Physics Department, The George Washington University, Washington, DC 20052, USA (Dated: November 10, 2026) Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths.

p. 0
Where the paper says it

Constrained field-space estimators for background-field lattice QCD observables Scott Moerschbacher1, ∗ and Frank X. Lee2, † 1Open Inquiry Research Group, Monrovia, MD 21770, USA 2Physics Department, The George Washington University, Washington, DC 20052, USA (Dated: November 10, 2026) Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths.

p. 0
Where the paper says it

Constrained field-space estimators for background-field lattice QCD observables Scott Moerschbacher1, ∗ and Frank X. Lee2, † 1Open Inquiry Research Group, Monrovia, MD 21770, USA 2Physics Department, The George Washington University, Washington, DC 20052, USA (Dated: November 10, 2026) Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths.

p. 0
Where the paper says it

Constrained field-space estimators for background-field lattice QCD observables Scott Moerschbacher1, ∗ and Frank X. Lee2, † 1Open Inquiry Research Group, Monrovia, MD 21770, USA 2Physics Department, The George Washington University, Washington, DC 20052, USA (Dated: November 10, 2026) Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths.

DEFG02-95ER40907

Appendix A: Archival Reconstruction

The poly11-small proton archive contains 992 correlator files The original poly0 proton notebook uses t = 8–18 Reconstructing its 90-configuration double-ratio pipeline reproduces all 24 stored central field-fit exponents with maximum absolute discrepancy 1.3 × 10−9

Appendix B: The Decuplet Identity at the Level of the Correlators

With S˜ = Cγ3 S Cγ3 and a contraction kernel F(S1, S2, S˜ 3; α, β) that is linear in its weights, the code evaluates C∆++ = F(Su, Su, S˜ u; 2, 4) At zero field Su = Sd = S and the two are equal configuration by configuration In the background field, Sq = S + eq δS + O(e2q) with the same S and δS for both flavors

Appendix C: Charged-Particle Response and a Nonanalytic Stencil

A charged spin-zero state illustrates why the cancellation basis must follow the physics Preserving a constant while cancelling 1/B, B, and B2 at B, 2B, 4B, 8B yields w = -8/3, 2 3/3, −13/3, 2 cubed This is a Laurent-series stencil in B

Appendix D: Dependence on Ensemble Size

The magnitude of the leverage grows with N (Fig. 2) Larger samples do not drive the covariance optimum toward the low-leverage stencil

Appendix E: Known-Covariance Synthetic Study

The population covariance is the noise-inclusive field-space covariance of the lightest proton or neutron BLUE-O4 has the same constraints as O4 but is displaced by up to 4.7σ, with a root-mean-square error up to 4. p. 5

Where the paper says it

Fit-window dependence of the O6 magnetic moment (nuclear magnetons) at the lightest and heaviest masses. Parentheses give marginal jackknife errors in units of the last digit; square brackets give the paired difference from the production window (rows without brackets) in units of its jackknife error.

8 times that of O6

Appendix F: Field Design

The geometric spacing does not produce the displacement Because Σ = V KV T makes the field-space covariance equivalent across designs by construction, this is a property of the model rather than a test

Appendix G: Discussion

The field-space viewpoint separates four questions often conflated in a single correlated fit The frontier of Sec. VII B is descriptive: a leverage is harmful only in proportion to the omitted coefficients it multiplies, since the displacement is c2L4 +a3L6 +· · ·

Appendix H: Limitations

The production windows are not free of excited states The upper ends of several ranges come from the heaviest mass The decuplet identities are exact only for valence-coupled fields with degenerate light quarks

Appendix I: Discussion

The field-space viewpoint separates four questions often conflated in a single correlated fit The frontier of Sec. VII B is descriptive: a leverage is harmful only in proportion to the omitted coefficients it multiplies, since the displacement is c2L4 +a3L6 +· · ·

Appendix J: Data and Code Availability

The derived field-response vectors, the aligned jackknife samples (fixed and nested), the bootstrap and synthetic-study outputs, the field-space covariance matrices and full-precision optimized weights, and the scripts that generate every table and figure are archived at Zenodo The underlying correlators are available from the authors upon reasonable request

Appendix K: Acknowledgments

We thank Andrei Alexandru for writing the propagator and contraction code with which the correlators analyzed here were produced This work was supported in part by the U.S. Department of Energy under Grant No. DEFG02-95ER40907 The most of the calculations were carried out at the DOE-sponsored National Energy Research Scientific Computing Center (NERSC)

Appendix L: Archival Reconstruction

The poly0 proton notebook uses t = 8–18 Reconstructing its 90-configuration double-ratio pipeline reproduces all 24 stored central field-fit exponents with maximum absolute discrepancy 1.3 × 10−9

Appendix M: Normalization

We therefore use the explicitly derived Eq. (17) The difference is an overall rescaling of the moments

Appendix N: Gauge Origin

Moving the origin by δ = x0 − x′ 0 multiplies every y-link by the constant phase e i q η δ, which is the gauge transformation Ωq(y) = e i q η δ y

Appendix O: Decuplet Identity at the Level of the Correlators

The code evaluates C∆++ = F(Su, Su, S˜ u; 2, 4) At zero field Su = Sd = S and the two are equal configuration by configuration

Appendix P: Charged-Particle Response and a Nonanalytic Stencil

Preserving a constant while cancelling 1/B, B, and B2 at B, 2B, 4B, 8B yields w = -8/3, 2 3/3, −13/3, 2 cubed

Appendix Q: Specification Tests

The null-space statistic also bounds discrepancies among estimators in the same constraint family The O4 rejections do not uniquely identify a physical B4 magnetic term

Appendix R: Covariance Noise

The noise that dominates the null contrasts enters through the correlated temporal fits The O6 verdict is therefore conditional on a temporal analysis whose systematics are of the same size as the effect tested

Appendix S: Conclusions

The fixed O(B4)–O(B8) stencils are nearly coincident and cost almost nothing in precision because all fields share the same gauge configurations Weight-free null-space tests resolve departures from the O4 response family in most nucleon ensembles Covariance optimization buys little precision, with 1–2% for the nucleon moments at the five lighter masses and 7–8% at the heaviest The decuplet correlators supply targets known exactly from the lattice contractions

Appendix T: Data and Code Availability

The derived field-response vectors, the aligned jackknife samples (fixed and nested), the bootstrap and synthetic-study outputs, the field-space covariance matrices and full-precision optimized weights, and the scripts that generate every table and figure are archived at Zenodo The underlying correlators are available from the authors upon reasonable request

Appendix U: Acknowledgments

We thank Andrei Alexandru for writing the propagator and contraction code with which the correlators analyzed here were produced This work was supported in part by the U.S. Department of Energy under Grant No. DEFG02-95ER40907 Most of the calculations were carried out at the DOE-sponsored National Energy Research Scientific Computing Center (NERSC)

Appendix V: Fixed Field-Space Hierarchy

Table I summarizes the O2–O8 hierarchy after the normalization in Eq. (17) The estimators are nearly coincident relative to their marginal errors, but this visual agreement is not a specification test because the successive estimators are extremely correlated

Appendix W: Specification Tests and Covariance Noise

The O4 rejections do not uniquely identify a physical B4 magnetic term The O6 verdict is therefore conditional on a temporal analysis whose systematics are of the same size as the effect tested

Appendix X: Conclusions

The fixed O(B4)–O(B8) stencils are nearly coincident and cost almost nothing in precision because all fields share the same gauge configurations Weight-free null-space tests resolve departures from the O4 response family in most nucleon ensembles Covariance optimization buys little precision, with 1–2% for the nucleon moments at the five lighter masses and 7–8% at the heaviest The decuplet correlators supply targets known exactly from the lattice contractions

Appendix Y: Data and Code Availability

The derived field-response vectors, the aligned jackknife samples (fixed and nested), the bootstrap and synthetic-study outputs, the field-space covariance matrices and full-precision optimized weights, and the scripts that generate every table and figure are archived at Zenodo The underlying correlators are available from the authors upon reasonable request

Appendix Z: Acknowledgments

We thank Andrei Alexandru for writing the propagator and contraction code with which the correlators analyzed here were produced This work was supported in part by the U.S. Department of Energy under Grant No. DEFG02-95ER40907 Most of the calculations were carried out at the DOE-sponsored National Energy Research Scientific Computing Center (NERSC)

Appendix AA: Archival Reconstruction

The poly0 proton notebook uses t = 8–18 Reconstructing its 90-configuration double-ratio pipeline reproduces all 24 stored central field-fit exponents with maximum absolute discrepancy 1.3 × 10−9

Appendix BB: The Decuplet Identity at the Level of the Correlators

The code evaluates C∆++ = F(Su, Su, S˜ u; 2, 4) At zero field Su = Sd = S and the two are equal configuration by configuration

Appendix CC: Charged-Particle Response and a Nonanalytic Stencil

Preserving a constant while cancelling 1/B, B, and B2 at B, 2B, 4B, 8B yields w = -8/3, 2 3/3, −13/3, 2 cubed This is a Laurent-series stencil in B

Appendix DD: Dependence on Ensemble Size

The magnitude of the leverage grows with N (Fig.

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. Systematic comparison of estimators: The AI system can perform a systematic comparison between fixed cancellation stencils and covariance-optimized estimators on real background-field correlators, evaluating their behavior on independent data and under the noise of estimated covariances.

  2. Null-space testing diagnostics: The system can use weight-free null-space tests to reject response models, distinguishing between departures from the O4 response family in most nucleon ensembles and not resolving departures from the O6 family unless coupled with a specific temporal analysis.

  3. Covariance optimization leverage assessment: The AI system can quantify the leverage of optimized weights on omitted response modes, noting that the optimized proton moments lie about one marginal standard error below fixed O4 where the response family is misspecified.

  4. Model uncertainty quantification: The system can use known-answer decuplet correlators to establish exact first-order identities like muDelta+ over muDelta++ = 1/2 and muDelta0 over muDelta++ = 0 exactly, which serve as targets for testing higher-order response consistency.

  5. Uncertainty propagation analysis: The system can analyze how the precision gain from covariance optimization depends on the temporal fit, noting that with two-state fits the gain rises to 3–30% because the fitted excited-state term correlates the field responses.

  6. Diagnostic pipeline evaluation: The AI system can evaluate whether a response family is accepted by testing null-space statistics and checking if the O4 rejections do not uniquely identify a physical B4 magnetic term; field-dependent excitedstate, boundary and other effects can project into the same null directions.

Abstract

Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths. We treat this extraction as linear estimation in field space and compare fixed cancellation stencils with covariance-optimized, generalized-least-squares estimators of the same coefficient, using archival nucleon double ratios on about 1000 quenched configurations at six quark masses and decuplet correlators for which exact answers follow from the lattice contractions. The fixed O(B 4) - O(B 8) stencils are nearly coincident and cost almost nothing in precision. Weight-free null-space tests reject the two-term response model in most nucleon ensembles; whether they reject the three-term model depends on the temporal fits and on the noise of the estimated temporal covariance, which raises the null-contrast error by a factor of 1.7-2.4. With that noise included, covariance optimization gains 1-2% in precision at the five lighter masses (3-30% with two-state time fits, which also lower moments by 3-10%) and nothing on independent data halves, while its weights carry three to four orders of magnitude more leverage on omitted response modes than the fixed stencils and move the moments by up to 1.8 marginal standard errors. The decuplet controls show what such leverage does under misspecification: the first-order responses obey μ Δ+/μ Δ++=1/2 and μ Δ 0=0 exactly. The lowest-order estimate misses the first by 10-25 standard errors. The fixed O(B 4) - O(B 8) stencils satisfy it within errors in the production windows, whereas the correlated linear fit in B squared, with the same constraints as the O(B 4) stencil, misses it by 3-17%; the null-space test rejects that fit's model. The results distinguish statistical precision from robustness to omitted response modes; they are not physical-point magnetic moments.

Sources

Related papers