HISQ: Highly Improved Staggered Quark Action

• HISQ is a lattice QCD discretization that improves staggered-fermion formulations using Fat7 smearing, reunitarization, Lepage correction, and Naik improvement.
• It significantly reduces taste-symmetry breaking and cutoff effects, achieving up to 3× improvements over asqtad in non-Goldstone pion splittings.
• HISQ supports diverse applications including precision flavor physics, heavy-quark spectroscopy, scale setting, and advanced solver optimizations.

Searching arXiv for recent HISQ papers and foundational references. arxiv_search(query="HISQ lattice QCD highly improved staggered quark", max_results=10, sort_by="relevance") arxiv_search(query="HISQ lattice QCD highly improved staggered quark", max_results=10, sort_by="relevance") is not available in this environment. HISQ, the Highly Improved Staggered Quark action, is a staggered-fermion discretization in lattice QCD designed to reduce taste-symmetry breaking and suppress leading cutoff effects. In the formulations summarized across scaling, spectroscopy, renormalization, and algorithmic studies, HISQ combines two levels of link smearing, reunitarization, and a Naik correction so that leading discretization errors are pushed down to $O(\alpha_S a^2, a^4)$, while taste-breaking splittings are reduced to an additional small $O(\alpha_S^2 a^2)$ level (Bazavov et al., 2010). Numerical comparisons with the earlier asqtad action consistently report substantially smaller lattice artifacts at the same lattice spacings and quark masses, and subsequent work has deployed HISQ in flavor physics, spectroscopy, scale setting, topological observables, flux-tube studies, nonperturbative renormalization, and large-scale solver development (Collaboration et al., 2010).

1. Definition and design objectives

The HISQ action was introduced to address two persistent limitations of staggered fermions. The first is taste-symmetry breaking, which causes the sixteen staggered pion tastes to become nondegenerate at finite lattice spacing. The second is the presence of leading $O(a^2)$ errors in the fermion dispersion relation, which are especially important for charm quarks. The HISQ design goal is therefore twofold: to make the pion tastes nearly degenerate even at finite $a$, and to improve the fermion action sufficiently that charm can be treated relativistically on fine enough lattices (Collaboration et al., 2010).

In the modern lattice-QCD usage summarized in the gradient-flow and flux-tube studies, HISQ is not an isolated fermion kernel but part of a broader lattice action. One couples the HISQ Dirac operator to a tree-level Symanzik-improved gauge action, so that

$$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$

For the HISQ/tree setup used in $(2+1)$-flavor flux-tube simulations, the gauge part is the tree-level improved Symanzik action with coefficients $c_0=5/3$ and $c_1=-1/12$ at tree level [(Bazavov et al., 2014); (Cea et al., 2017)].

A recurrent comparison point is the older asqtad action. Across the scaling studies and phenomenological applications, HISQ is described as a further $O(a^2)$-improved version of staggered fermions that reduces taste splittings by roughly a factor of two to three relative to asqtad, and in some summaries by roughly an order of magnitude depending on the observable and convention being emphasized [(Bazavov et al., 2012); (Lytle, 2015)]. This suggests that the phrase “HISQ improvement” is best understood not as one isolated modification, but as a coordinated suppression of taste exchange, dispersion errors, and generic cutoff effects.

2. Fermion operator and link construction

The HISQ Dirac operator is built in stages. In the scaling study of Bazavov et al., the massless part is written as

$$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$

where $O(\alpha_S^2 a^2)$0 is the twice-smeared link entering the one-hop term and $O(\alpha_S^2 a^2)$1 is the once-smeared, unitarized link entering the Naik term (Collaboration et al., 2010).

The construction proceeds through four steps: Fat7 smearing of the original gauge links $O(\alpha_S^2 a^2)$2; projection $O(\alpha_S^2 a^2)$3 onto $O(\alpha_S^2 a^2)$4 by polar decomposition or SVD; Asqtad-style smearing $O(\alpha_S^2 a^2)$5; and addition of a Naik term with a mass-dependent coefficient $O(\alpha_S^2 a^2)$6 for the charm quark (Collaboration et al., 2010). The same sequence is summarized elsewhere as Fat7 smearing, reunitarization, Lepage correction, and Naik improvement [(Gamiz et al., 2012); (Chakraborty et al., 2017)].

The explicit Fat7 coefficients quoted in the scaling study are: 1-link $O(\alpha_S^2 a^2)$7, 3-staple $O(\alpha_S^2 a^2)$8, 5-staple $O(\alpha_S^2 a^2)$9, and 7-staple $O(a^2)$0, with all rotations and reflections included. After projection, the second smearing repeats the same staple structure and adds a negative Lepage term with $O(a^2)$1, while the Naik term carries $O(a^2)$2 (Collaboration et al., 2010). In the HISQ/tree implementation described for flux-tube calculations, tadpole improvement uses

$O(a^2)$3

from the plaquette, and reunitarization is carried out by polar decomposition, projecting first to $O(a^2)$4 and then to $O(a^2)$5 (Cea et al., 2017).

For light and strange quarks, $O(a^2)$6. For the charm quark, the tree-level coefficient is expanded as

$O(a^2)$7

which is chosen so that heavy-quark dispersion relations are accurate through higher order in $O(a^2)$8 (Collaboration et al., 2010). In related summaries this same role is described as tuning $O(a^2)$9 or $a$0 so that the tree-level kinetic mass equals the pole mass, or so that the dispersion relation is improved through $a$1 [(Bazavov et al., 2012); (Monahan et al., 2012)].

3. Error structure, taste symmetry, and scaling

The central theoretical claim attached to HISQ is that the two-stage smearing plus reunitarization suppresses taste-exchange interactions much more effectively than earlier staggered formulations. In the dynamical-HISQ overview, the leading discretization errors are summarized as $a$2, and the remaining taste-breaking splittings as $a$3 (Bazavov et al., 2010). In the flux-tube summary, the tree-level Symanzik gauge action removes $a$4 errors in the pure-gauge sector, while Fat7 smearing, projection, and the Naik term cancel the leading taste-breaking $a$5 artifacts of staggered fermions; the sequence “Fat7 $a$6 project $a$7 Asq-tone” pushes taste splittings down to $a$8 or smaller (Cea et al., 2017).

The standard scaling relation for non-Goldstone pion tastes is

$a$9

or, in the form used in the scaling study,

$$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$0

Numerically, HISQ reduces these taste splittings by $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$1 relative to asqtad at the same lattice spacing, in excellent agreement with the expected $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$2 scaling [(Collaboration et al., 2010); (Bazavov et al., 2010)]. In the correlator-fit summary, the largest non-Goldstone pion at $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$3 fm is $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$4 MeV above the Goldstone, compared to $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$5 MeV with asqtad (Bazavov et al., 2012).

This improved taste symmetry is accompanied by smaller cutoff effects in hadronic quantities. For light-light and heavy-light masses, the HISQ points lie very close to the asqtad continuum curve, and “a HISQ result at $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$6 matches an asqtad result at $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$7” (Collaboration et al., 2010). The same pattern is reported for pseudoscalar decay constants and topological susceptibility, with the latter improved by about a factor $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$8 in $$S_{\rm HISQ} \;=\; S_{\rm gauge}^{\rm (Sym)}\;+\;\sum_{f=u,d,s,c} \bar\chi_f\,\bigl(D^{\rm HISQ}[U^{\rm (HISQ)}]+m_f\bigr)\,\chi_f\,.$$9 (Collaboration et al., 2010). The dynamical-HISQ program likewise reports flatter $(2+1)$0 dependence in $(2+1)$1 and $(2+1)$2, with slopes reduced by roughly a factor of two relative to asqtad (Bazavov et al., 2010).

A common misconception is that HISQ only improves taste splittings in the light sector. The heavy-quark tuning of the Naik term is an equally central part of the construction. This is why later work could extend the heavy-HISQ method to valence heavy masses ranging from the charm quark up to very nearly the physical $(2+1)$3-quark on the finest lattices, while maintaining controlled discretization effects in joint continuum and heavy-mass fits (Miller et al., 10 Feb 2025).

4. Ensembles, scale setting, and lines of constant physics

HISQ simulations are typically organized along a line of constant physics. In the flux-tube study with $(2+1)$4 flavors, this means choosing $(2+1)$5 at each gauge coupling $(2+1)$6 so that hadron masses remain at their physical values, with $(2+1)$7 fixed to its physical value and $(2+1)$8, implying $(2+1)$9 MeV (Cea et al., 2017). The stated purpose is to guarantee comparability of flux-tube profiles measured at different lattice spacings and to ensure that quark masses track the physical point as $c_0=5/3$0 (Cea et al., 2017).

The early dynamical-HISQ scaling studies used four flavors of dynamical HISQ quarks at lattice spacings approximately 0.15, 0.12, and 0.09 fm, with $c_0=5/3$1, $c_0=5/3$2, and $c_0=5/3$3 (Collaboration et al., 2010). Later MILC/Fermilab HISQ ensembles extended to $c_0=5/3$4 flavors, lattice spacings from 0.15 fm down to 0.06 fm, and light sea-quark masses from $c_0=5/3$5 down to the physical point (Bazavov et al., 2012). Gradient-flow analyses on these ensembles explicitly covered four lattice spacings from $c_0=5/3$6 fm down to $c_0=5/3$7 fm and both physical and unphysical quark masses (Bazavov et al., 2014).

Several scale-setting prescriptions recur in the HISQ literature. One is the static-potential quantity $c_0=5/3$8, defined through

$c_0=5/3$9

with $c_1=-1/12$0 (Collaboration et al., 2010). Another is the decay constant of a fictitious “unmixed $c_1=-1/12$1” pseudoscalar, $c_1=-1/12$2, obtained by tuning the valence mass until the pseudoscalar has the physical ratio $c_1=-1/12$3 (Collaboration et al., 2010). A later review of dynamical-HISQ simulations gives representative lattice spacings from both $c_1=-1/12$4 and $c_1=-1/12$5, and emphasizes that differences $c_1=-1/12$6 vanish as $c_1=-1/12$7 (Bazavov et al., 2010).

Gradient-flow scales provide a third approach. Using Symanzik flow and the cloverleaf definition of $c_1=-1/12$8, the scales $c_1=-1/12$9 and $O(a^2)$0 are defined by

$O(a^2)$1

On MILC $O(a^2)$2 HISQ ensembles, the reported preliminary physical values are $O(a^2)$3 fm and $O(a^2)$4 fm (Bazavov et al., 2014). This suggests that HISQ calculations have supported a transition from purely potential-based scale setting to a mixed ecosystem of potential, pseudoscalar, and flow-based standards.

5. Phenomenology and nonperturbative applications

HISQ has been used extensively in precision flavor physics. In two-point correlator analyses on nineteen $O(a^2)$5-flavor HISQ ensembles, simultaneous fits with the full covariance matrix and Gaussian constraints were used to extract light, strange, and charmed pseudoscalar observables, supporting preliminary results

$O(a^2)$6

together with $O(a^2)$7 and quark-mass ratios $O(a^2)$8, $O(a^2)$9 (Bazavov et al., 2012).

In kaon semileptonic decays, HISQ valence quarks were first used on asqtad $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$0 ensembles and then on full HISQ $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$1 ensembles including physical light-quark masses. The partially quenched staggered-chiral analysis of $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$2 reported projected total uncertainties of $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$3–$$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$4 on the asqtad-sea data and anticipated $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$5–$$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$6 on full HISQ data because sea-strange tuning and dynamical charm are correct by construction and taste splittings are another factor of two smaller in HISQ sea (Gamiz et al., 2012). In semileptonic $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$7 and $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$8 work with NRQCD heavy quarks and HISQ light valence quarks, the HISQ component was specifically used to suppress taste-breaking and reduce renormalization uncertainties in form-factor calculations over a range of momentum transfer [(Bouchard et al., 2012); (Monahan et al., 2012)].

Spectroscopy and current normalization provide another major application. In the nonperturbative comparison of clover and HISQ strange-quark correlators, local vector and axial currents made from HISQ, clover, and mixed HISQ–clover fields were normalized nonperturbatively, and the physical $$(/D_{\rm HISQ})_{x,y} = \sum_{\mu}\bigl[\,\delta_{x+\hat\mu,y}\,X_\mu(x)\;-\;\delta_{x-\hat\mu,y}\,X_\mu^\dagger(x-\hat\mu)\bigr] \;+\;(1+\epsilon_N)\sum_{\mu}\bigl[\,\delta_{x+3\hat\mu,y}\,W_\mu(x)\,W_\mu(x+\hat\mu)\,W_\mu(x+2\hat\mu)\;-\;\mathrm{h.c.}\bigr]\,,$$9-meson quantities were reported as

$O(\alpha_S^2 a^2)$00

in good agreement with experiment, using HISQ valence quarks on $O(\alpha_S^2 a^2)$01 ensembles with light sea masses down to the physical point (Chakraborty et al., 2017). The same study emphasizes that the remnant $O(\alpha_S^2 a^2)$02 chiral symmetry of HISQ provides an absolutely normalized decay constant for the Goldstone-taste pseudoscalar through the PCAC relation (Chakraborty et al., 2017).

Beyond hadron structure and flavor physics, HISQ has been used in gauge-field diagnostics and vacuum-structure studies. In the flux-tube calculation with $O(\alpha_S^2 a^2)$03 HISQ fermions, Monte Carlo simulations using the HISQ/tree action in publicly available MILC code were used to investigate the transverse profile of the chromoelectric field generated by a quark-antiquark pair (Cea et al., 2017). In nonperturbative renormalization, HISQ bilinears have been analyzed in RI/MOM and RI/SMOM schemes on $O(\alpha_S^2 a^2)$04 HISQ ensembles, where the SMOM setup was reported to suppress infrared contamination strongly relative to exceptional MOM kinematics (Lytle, 2015).

6. Solvers, eigenmodes, and terminological scope

The computational cost of HISQ inversions has driven substantial algorithmic work. Standard propagator calculations typically use mixed-precision CG with even–odd preconditioning, but this suffers from critical slowing down as the light-quark mass approaches its physical value (Hostetler et al., 31 Jan 2025). On a $O(\alpha_S^2 a^2)$05, $O(\alpha_S^2 a^2)$06 fm ensemble at the physical light-quark mass, the baseline undeflated CG reported $O(\alpha_S^2 a^2)$07 iterations and $O(\alpha_S^2 a^2)$08 s per solve at residual $O(\alpha_S^2 a^2)$09; deflation with $O(\alpha_S^2 a^2)$10 eigenvectors reduced this to $O(\alpha_S^2 a^2)$11 iterations and $O(\alpha_S^2 a^2)$12 s, while $O(\alpha_S^2 a^2)$13 eigenvectors reduced it to $O(\alpha_S^2 a^2)$14 iterations and $O(\alpha_S^2 a^2)$15 s, a $O(\alpha_S^2 a^2)$16 speedup over CG (Hostetler et al., 31 Jan 2025).

Earlier numerical experiments already showed the same qualitative behavior. On a $O(\alpha_S^2 a^2)$17, $O(\alpha_S^2 a^2)$18 MeV HISQ ensemble, exact deflation reduced the CG iteration count from $O(\alpha_S^2 a^2)$19 to $O(\alpha_S^2 a^2)$20 with $O(\alpha_S^2 a^2)$21 deflated modes and to $O(\alpha_S^2 a^2)$22 with $O(\alpha_S^2 a^2)$23 modes, corresponding to speed-ups of $O(\alpha_S^2 a^2)$24 and $O(\alpha_S^2 a^2)$25 (Davies et al., 2017). The same study emphasized that eigenpairs must be converged to a residual at least an order of magnitude below the target solve residual; otherwise the CG residual levels off (Davies et al., 2017).

For eigenmode calculations, studies of the Hermitian positive-definite operator $O(\alpha_S^2 a^2)$26 with the HISQ Dirac operator compared Implicitly Restarted Lanczos, Thick-Restart Lanczos, and Block Lanczos as implemented in Grid and QUDA. Chebyshev polynomial filtering with $O(\alpha_S^2 a^2)$27 and degree $O(\alpha_S^2 a^2)$28 was reported as optimal for extracting $O(\alpha_S^2 a^2)$29 modes, QUDA’s TRL outperformed non-restarted Lanczos in all tested $O(\alpha_S^2 a^2)$30, and Block Lanczos with Split-Grid was nearly competitive with the unblocked code while reducing communication overhead (Jeong et al., 2022). This suggests that the practical history of HISQ is inseparable from the development of libraries such as MILC, QUDA, and Grid.

Finally, the acronym HISQ is not unique across arXiv. In lattice QCD it denotes Highly Improved Staggered Quark and refers to the fermion action discussed throughout this literature (Collaboration et al., 2010). In a 2025 quantum-control architecture paper, however, HISQ denotes a universal instruction set defined as an extension of RV32I and paired with a booking-based synchronization protocol in distributed control hardware (Zhao et al., 5 Sep 2025). The coexistence of these two usages is terminological rather than conceptual. In the lattice-QCD literature, HISQ retains its established meaning: a staggered-quark discretization built from Fat7 smearing, reunitarization, Lepage correction, and Naik improvement, paired in practice with improved gauge actions and used for high-precision calculations across a broad range of observables [(Bazavov et al., 2010); (Cea et al., 2017)].