---
title: Hermitian Construction Methods
url: https://www.emergentmind.com/topics/hermitian-construction
type: topic
---

# Hermitian Construction Methods

In the literature covered here, **Hermitian Construction** does not denote a single universal procedure. It denotes a family of constructions in which a Hermitian inner product, Hermitian duality, Hermitian metric, or Hermitian projection is built into the defining mechanism. In coding theory, it most often means constructing classical Hermitian self-orthogonal or dual-containing codes over finite fields and then passing to quantum stabilizer codes, or enlarging such codes by matrix-product, unitary-matrix, quasi-cyclic, and LCD-preserving methods [1710.04999] [2012.11998]. In complex geometry, it denotes constructive schemes for Hermitian–Einstein metrics and Hermitian Yang–Mills instantons [2606.30121] [2408.09653]. In higher-categorical $K$-theory, it denotes the hermitian $Q$-construction and its comparison with the real $S$-construction [2410.07846]. In representation theory and matrix analysis, it denotes the construction of Hermitian Young projectors and Hermitian almost-companion matrices [1610.10088] [2301.07435].

| Domain | Hermitian datum | Output |
|---|---|---|
| Coding theory | Hermitian inner product, Hermitian duality | Classical self-orthogonal, self-dual, LCD, and quantum stabilizer codes |
| Complex geometry | Hermitian metric, mean curvature, HYM projector | Hermitian–Einstein metrics and Hermitian Yang–Mills instantons |
| Higher-categorical $K$-theory | Genuine duality, hermitian $Q$-construction | Genuine $C_2$-spaces and real $K$-theory models |
| Representation and matrix theory | Hermitian projectors, Hermitian matrices | Orthogonal SU$(N)$ projectors and ACM realizations of polynomials |

## 1. Coding-theoretic foundations

Over $\mathbb{F}_{q^{2m}}$, the basic Hermitian involution is $x \mapsto x^{q^m}$, and the Hermitian inner product on $\mathbb{F}_{q^{2m}}^n$ is
\[
\langle x, y\rangle_H=\sum_{i=1}^n x_i\, y_i^{q^m}.
\]
The Hermitian dual is
\[
C^{\perp_H}=\{x\in \mathbb{F}_{q^{2m}}^n:\langle x,y\rangle_H=0\ \forall y\in C\},
\]
and Hermitian self-orthogonality means $C\subseteq C^{\perp_H}$ [2012.11998]. In the standard case $m=1$, this is the usual Hermitian duality over $\mathbb{F}_{q^2}$.

A central classical mechanism is the **matrix-product construction**. If $C_1,\dots,C_s\subseteq \mathbb{F}_{q^2}^n$ are linear codes with generator matrices $G_i$, and $A\in\mathbb{F}_{q^2}^{s\times \ell}$, the matrix-product code
\[
C=[C_1,\dots,C_s]A
\]
has length $n\ell$, dimension at most $\sum_i k_i$, and satisfies
\[
d(C)\ge \min_{1\le i\le s}\{d(C_i)D_i(A)\},
\]
where $D_i(A)$ is the minimum Hamming weight of the length-$\ell$ code generated by the first $i$ rows of $A$; equality holds for nested inputs $C_1\supseteq\cdots\supseteq C_s$ [1710.04999]. When $s=\ell$ and $A$ is nonsingular,
\[
([C_1,\dots,C_s]A)^{\perp_H}=[C_1^{\perp_H},\dots,C_s^{\perp_H}](A^{-1})^{\dagger},
\]
with $A^{\dagger}=\overline{A}^{\,T}$.

The Hermitian self-orthogonality criteria in this setting are notably flexible. If $AA^{\dagger}$ is diagonal and every $C_i$ is Hermitian self-orthogonal, then
\[
[C_1,\dots,C_s]A\subseteq ([C_1,\dots,C_s]A)^{\perp_H}.
\]
If $AA^{\dagger}$ is anti-diagonal and $C_i\subseteq C_{s-i+1}^{\perp_H}$ for all $i$, then the same inclusion holds even when the inputs are not individually self-orthogonal [1710.04999]. The quasi-unitary and anti-quasi-unitary cases,
\[
AA^{\dagger}=\lambda I_s,\qquad AA^{\dagger}=\lambda J_s,
\]
appear as immediate corollaries, and the reversal condition $C_i=C_{s-i+1}^{\perp_H}$ yields Hermitian self-duality.

Explicit matrices are available. For instance, when $M\mid (r+1)$, the Fourier/Vandermonde-type matrix
\[
A_{i,j}=\omega^{(i-1)(j-1)}
\]
satisfies
\[
AA^{\dagger}=M I_M,\qquad D_i(A)=M-i+1.
\]
This strict decrease in the column-distance profile improves the lower bound on $d(C)$ [1710.04999]. The paper gives explicit examples such as Hermitian self-orthogonal codes with parameters $[18,6,6]_4$, $[8,3,4]_4$, $[8,3,4]_{16}$, and $[12,3,6]_4$.

A parallel finite-field line concerns **Hermitian LCD codes** over $\mathbb{F}_4$. For a generator matrix $G$, the basic criterion is
\[
C\text{ is Hermitian LCD}\iff GG^{\dagger}\text{ is nonsingular},
\]
and more generally
\[
\operatorname{rank}(GG^{\dagger})=k-\dim(\operatorname{Hull}_H(C)).
\]
Every quaternary Hermitian LCD code admits a generator matrix with orthonormal rows,
\[
G\overline{G}^{\,T}=I_k
\]
[2207.00801]. The same paper gives a hull-preserving transformation $A\mapsto A(x,y)$ for systematic generators $G=(I_k\mid A)$, where isotropic and pairwise Hermitian orthogonal vectors $x,y$ satisfy
\[
\langle x,x\rangle_H=\langle y,y\rangle_H=\langle x,y\rangle_H=0,
\]
and the transformed code preserves the Hermitian hull dimension because
\[
G'\overline{G'}^{\,T}=G\overline{G}^{\,T}.
\]
It also establishes a sharp puncture/shorten criterion: if $G\overline{G}^{\,T}=I_k$ and $l_i$ is the $i$-th column of $G$, then
\[
C^i\text{ is Hermitian LCD}\iff \mathrm{wt}(l_i)\text{ is even},\qquad
C_i\text{ is Hermitian LCD}\iff \mathrm{wt}(l_i)\text{ is odd}
\]
[2207.00801].

A related construction over $\mathbb{F}_4$ extends Hermitian LCD codes from $[n,k]$ to $[n+2,k+1]$ by adjoining two coordinates derived from Hermitian inner products with an even-weight vector $x\in \mathbb{F}_2^n$. In the Hermitian case, the new Gram matrix becomes
\[
G_4(x)\overline{G_4(x)}^{\,T}=\operatorname{diag}(1,G_4\overline{G_4}^{\,T}),
\]
so Hermitian LCD is preserved [2101.11821].

A further self-dual branch uses **unitary matrices over finite fields**. If $U^{\dagger}U=I_n$ and $\alpha$ is chosen so that $N(\alpha)=\alpha\overline{\alpha}=-1$ in odd characteristic, or $N(\alpha)=1$ in characteristic $2$, then $A=\alpha U$ satisfies $AA^{\dagger}=-I_n$ or $I_n$ respectively, and
\[
G=[\,I_n\mid A\,]
\]
generates a Hermitian self-dual $[2n,n]$ code because
\[
GG^{\dagger}=I_n+AA^{\dagger}=0
\]
[1911.10456]. This framework produces many MDS or almost MDS Hermitian self-dual codes.

## 2. Quantum stabilizer, rank-metric, and locality constructions

In coding theory proper, the phrase **Hermitian construction** most often refers to the passage from Hermitian self-orthogonal or Hermitian dual-containing classical codes to quantum stabilizer codes. In the standard CRSS-type statement, if $C$ is an $\mathbb{F}_{q^2}$-linear $[n,k]$ code with
\[
C\subseteq C^{\perp_H},
\]
then there exists a $q$-ary stabilizer quantum code with parameters
\[
[[n,\ n-2k,\ D]]_q,\qquad D\ge d(C^{\perp_H})
\]
[2012.11998]. The same paper generalizes this from $\mathbb{F}_{q^2}$ to $\mathbb{F}_{q^{2m}}$: if $C$ is an $\mathbb{F}_{q^{2m}}$-linear $[n,k]$ code with $C\subseteq C^{\perp_H}$, then there exists a $q$-ary stabilizer code with parameters
\[
[[mn,\ mn-2mk,\ D]]_q,\qquad D\ge d(C^{\perp_H})
\]
[2012.11998]. This is implemented by field restriction from $\mathbb{F}_{q^m}$ to $\mathbb{F}_q$.

The same Hermitian mechanism also appears in **binary quantum codes from quaternary quasi-cyclic codes**. For a 2-generator index-2 QC code over $\mathbb{F}_4$,
\[
\mathcal{C}_4(g_1,g_2,\nu),
\]
the paper establishes sufficient conditions ensuring Hermitian dual containment, including
\[
g_i(x)\mid g_i^{\perp_H}(x),\qquad
g_2(x)\mid (\nu(x)+\overline{\nu}(x))\,g_1^{\perp_H}(x),
\]
together with $\gcd(\nu(x)-1,x^n-1)=1$ for the dimension formula
\[
\dim \mathcal{C}_4(g_1,g_2,\nu)=2n-\deg(g_1)-\deg(g_2).
\]
The Hermitian construction then yields a binary stabilizer code
\[
[[2n,\ 2n-2\deg(g_1)-2\deg(g_2),\ d]]_2
\]
[2210.04487]. This produced 30 binary quantum codes improving Grassl’s tables, including $[[42,10,9]]_2$, $[[70,40,7]]_2$, $[[74,36,9]]_2$, and $[[78,40,9]]_2$.

A different specialization concerns **quantum rank-metric codes for stacked quantum memory**. Here the classical input is a Gabidulin code over $\mathbb{F}_{2^{2m}}$ with Hermitian self-orthogonality guaranteed by a self-dual basis and the condition $k\le m$. The resulting construction yields quantum rank-metric codes on an $L\times M$ stacked memory with
\[
L=2m,\qquad M=m,\qquad N_{\mathrm{phys}}=2m^2,
\]
and parameters
\[
[[2m\times m,\ 2m(m-k),\ k+1]]_2
\]
[2605.02571]. The paper emphasizes that this replaces CSS orthogonality with Hermitian orthogonality and removes the old odd-by-odd square-layout restriction.

Hermitian construction also transfers **locality**. If $C$ is an $[n,k,d]_{q^2}$ Hermitian dual-containing classical LRC with locality $r$ and $d(C^{\perp_H})\ge 2$, then there exists a qLRC
\[
[[n,\ \kappa,\ \delta]]_q,\qquad \kappa=2k-n,\qquad \delta=\mathrm{wt}(C\setminus C^{\perp_H})\ge d(C),
\]
with the same locality $r$ [2508.13553]. Using NMDS codes that support $t$-designs for $t\in\{2,3\}$, the paper constructs three explicit optimal families, including
\[
[[2^{m_1},\,2^{m_1}-2k,\,\ge k]]_{2^m},\qquad
[[2^{m_2}-1,\,2^{m_2}-2k-1,\,\ge k]]_{2^m},
\]
with locality determined explicitly by $m_1,m_2$, and proves optimality with respect to Singleton-like bounds [2508.13553].

A useful terminological point is that two equivalent-looking formulations coexist. Some papers start from self-orthogonal codes and write $C\subseteq C^{\perp_H}$, yielding parameters of the form $[[n,n-2k,d]]_q$ [2012.11998]. Others start from dual-containing codes and write $C^{\perp_H}\subseteq C$, yielding parameters of the form $[[n,2k-n,d]]_q$ [2210.04487] [2508.13553]. The underlying stabilizer mechanism is the same.

## 3. Algebraic-geometric and finite-geometric Hermitian constructions

A distinct Hermitian construction arises from **generalized Hermitian curves** and **Hermitian hypersurfaces** over finite fields. For odd $c\ge 3$ with $c=a+b$, $a=b+1$, and $\gcd(a,b)=1$, the generalized Hermitian curve over $\mathbb{F}_{q^c}$ is defined by
\[
\mathrm{Tr}_b\!\left(\frac{y^{q^a}}{x}\right)+\mathrm{Tr}_a\!\left(\frac{y}{x^{q^b}}\right)=1.
\]
The paper constructs explicit AG codes from the divisor
\[
G=vP_1+rP_0+sQ+tV
\]
and the evaluation divisor
\[
D=\sum_{\alpha,\beta} D_{\alpha,\beta},\qquad \deg(D)=(q^c-1)q^{c-1},
\]
where the affine points satisfy
\[
\mathrm{Tr}_c\!\left(\frac{\beta}{\alpha^{q^b}}\right)=1
\]
[1507.03418].

The key constructive step is an explicit basis of $\mathcal{L}(G)$. With
\[
z=\frac{y}{x^{q^b}},\qquad
u=a^{-1}-\left(\frac{y^{q^a}}{x}-\frac{y^q}{x^{q^a}}\right),\qquad
w=\frac{y^{q^a}}{x\,u},
\]
the monomials
\[
x^i z^j w^k,\qquad (i,j,k)\in \Omega_{v,r,s,t},
\]
form a basis of $\mathcal{L}(G)$, where $\Omega_{v,r,s,t}$ is defined by explicit linear inequalities coming from the divisors of $u,z,w$ [1507.03418]. This yields a generator matrix by evaluating basis elements at all rational points in $D$. The dual remains in the same family:
\[
C_{v,r,s,t}^{\perp}=C_{-1-v,\,-1-r,\,A-s,\,B-t},
\]
with explicit constants
\[
A=q^{c+a}+q^c-q^a-2,\qquad B=(q^{a-1}-1)N_c-1
\]
[1507.03418]. A concrete example over $\mathbb{F}_{32}$ has parameters
\[
[496,250,\ge 172],
\]
and the paper verifies that it oversteps the Gilbert–Varshamov bound [1507.03418].

The finite-geometric side studies **Hermitian hypersurfaces**
\[
X_0^{Q+1}+X_1^{Q+1}+\cdots+X_s^{Q+1}=0
\]
over $\mathbb{F}_q$ with $q=Q^2$ [1411.3706]. The number of $\mathbb{F}_q$-rational points is
\[
|H_s(\mathbb{F}_q)|=\frac{(Q^{s+1}-(-1)^{s+1})(Q^s-(-1)^s)}{Q^2-1}.
\]
For $s=2$ this gives $Q^3+1$, and for $s=3$ it gives $(Q^2+1)(Q^3+1)$.

The same paper gives a recursive construction of the full rational point set. Writing $h_s=|H_s|$ and $T_s=|\mathbb{P}^s(\mathbb{F}_q)|$, one has
\[
h_{s+1}=(Q+1)(T_s-h_s)+h_s.
\]
The mechanism is the norm map: if $[a_0:\dots:a_s]\notin H_s$, then
\[
b=a_0^{Q+1}+\cdots+a_s^{Q+1}\in \mathbb{F}_Q^\times,
\]
and the equation $X_{s+1}^{Q+1}=-b$ has exactly $Q+1$ solutions in $\mathbb{F}_q$ [1411.3706]. The paper also derives the zeta function of diagonal hypersurfaces from Wolfmann’s formulas and observes that the recursive point-set organization is relevant to evaluation codes on Hermitian surfaces. For linear forms, the resulting code $C_H(1,s,q)$ has exactly two nonzero weights,
\[
w_1=Q(2s-1)+(-1)^{s-1}Q^{s-1},\qquad w_2=Q(2s-1)
\]
[1411.3706].

## 4. Hermitian metrics, Hermitian–Einstein iteration, and HYM instantons

In complex differential geometry, Hermitian construction refers to the explicit production of canonical Hermitian metrics. For a stable holomorphic vector bundle $E\to M$ over a compact Kähler or Gauduchon manifold $(M,\omega_g)$, the recent iterative construction defines Hermitian metrics $\{h_m\}$ by
\[
\Lambda_{\omega_g}\!\left(\sqrt{-1}\,R^{h_{m+1}}\right)=(\lambda_E-\mu)\,h_{m+1}+\mu\,h_m,\qquad \mu>0.
\]
The paper proves that for any initial Hermitian metric $h_0$ there exists a unique sequence $\{h_m\}$ converging smoothly to a Hermitian–Einstein metric $h_\infty$ satisfying
\[
\Lambda_{\omega_g}\!\left(\sqrt{-1}\,R^{h_\infty}\right)=\lambda_E\,h_\infty
\]
[2606.30121].

The analysis uses the tensor $S^h=\Lambda_{\omega_g}(\sqrt{-1}R^h)$ and the endomorphism $K^h=S^h\cdot h^{-1}$, together with the comparison endomorphism
\[
Q_{m+1}=h_m h_{m+1}^{-1}.
\]
A central estimate is the discrete energy dissipation inequality for the Hermitian–Yang–Mills energy
\[
\Phi(h)=\int_M \operatorname{tr}_E\big((K^h-\lambda_E\operatorname{Id}_E)^2\big)\,\omega_g^n,
\]
namely
\[
\Phi(h_m)-\Phi(h_{m+1})\ge \|K^{h_m}-K^{h_{m+1}}\|^2_{L^2(M,\omega_g,h_m)}.
\]
The proof is independent of Donaldson’s variational framework and works on Gauduchon manifolds because the Gauduchon condition $\partial\bar\partial(\omega_g^{n-1})=0$ eliminates the torsion contribution appearing in the energy computation [2606.30121].

A related but distinct geometric construction concerns **Hermitian Yang–Mills instantons on six-dimensional coset manifolds**. Using the isomorphism $\mathrm{Spin}(6)\cong \mathrm{SU}(4)$ and generalized six-dimensional ’t Hooft symbols $\eta^a_{AB}$, the spin connection $\omega^{AB}$ produces an $\mathrm{SU}(4)$ gauge field
\[
A^a=\eta^a_{AB}\,\omega^{AB},\qquad F^a=\eta^a_{AB}\,R^{AB}.
\]
The key decomposition of two-forms is
\[
\Lambda^2=\Lambda^2_{\pm1}\oplus \Lambda^2_{\pm6}\oplus \Lambda^2_{\pm8},
\]
with respect to the operator
\[
*_\Omega(\alpha)=*(\alpha\wedge \Omega),
\]
where
\[
\Omega=e^1\wedge e^2+e^3\wedge e^4+e^5\wedge e^6.
\]
The HYM curvature lies in $\Lambda^2_{\pm8}$ and satisfies
\[
F=-*(F\wedge \Omega),\qquad F\wedge \Omega=0
\]
[2408.09653].

The paper works this out explicitly on $\mathrm{SU}(4)/\mathrm{U}(3)\cong \mathbb{CP}^3$ and on $\mathrm{Sp}(4)/[\mathrm{SU}(2)\times \mathrm{U}(1)]$ cosets. For $\mathbb{CP}^3$, the HYM part satisfies
\[
F^{\hat a}=2J^{\hat a}\in \Lambda^2_{+8},
\]
and the topological invariants include $\chi(\mathbb{CP}^3)=4$ and an HYM contribution with
\[
C_2(\mathrm{HYM})\cdot [\Omega]=\pi^3
\]
[2408.09653]. This use of Hermitian construction is thus not algebraic-coding-theoretic, but gauge-theoretic: the spin connection is reorganized into an HYM instanton by Hermitian projectors intrinsic to the $\mathrm{SU}(3)$-structure.

## 5. Hermitian projectors, almost-companion matrices, and constructive Hermitization

In representation theory, Hermitian construction refers to replacing non-Hermitian algebraic projectors by orthogonal ones. For a standard Young tableau $\Theta$, the ordinary Young operator $Y_\Theta$ is idempotent but generally not Hermitian. A recursive construction replaces it by
\[
P_\Theta=
\begin{cases}
Y_\Theta, & n\le 2,\\[4pt]
P_{\Theta_{(1)}}\,Y_\Theta\,P_{\Theta_{(1)}}, & n>2,
\end{cases}
\]
where $\Theta_{(1)}$ is obtained by removing the box labeled $n$ [1610.10088]. The resulting family satisfies
\[
P_\Theta^\dagger=P_\Theta,\qquad
P_\Theta^2=P_\Theta,\qquad
P_\Theta P_\Phi=\delta_{\Theta\Phi}P_\Theta,\qquad
\sum_{\Theta\in \mathcal{Y}_n}P_\Theta=\mathbb{1}_{V^{\otimes n}}.
\]
For lexically ordered tableaux there are compact closed forms,
\[
P_\Theta=\alpha_\Theta\,\bar Y_\Theta \bar Y_\Theta^\dagger
\quad\text{or}\quad
P_\Theta=\alpha_\Theta\,\bar Y_\Theta^\dagger \bar Y_\Theta,
\]
and the MOLD construction further compresses the recursion [1610.10088]. The earlier Hermitian Young-operator paper gives the same objective in operator form: to project orthogonally onto irreducible SU$(N)$ subspaces carried by $V^{\otimes n}$ [1307.6147].

In matrix analysis, the term appears in the construction of **Hermitian almost-companion matrices**. For a cubic polynomial
\[
P_3(z)=z^3+c_1 z^2+c_2 z+c_3,
\]
one first passes to the depressed cubic
\[
P_{3c}(\eta)=\eta^3+p\eta+q.
\]
An ACM is any matrix whose characteristic polynomial equals the prescribed polynomial. In the real-coefficient cubic case, the necessary and sufficient condition for a Hermitian ACM is
\[
\Delta(p,q)=\frac{p^3}{27}+\frac{q^2}{4}\le 0,
\]
and the Hermitian ACM can then be written as
\[
\widehat Q_{3c}(p,q)= -\sqrt{|p|/3}
\begin{bmatrix}
0 & 1 & e^{i\Phi_{13}(p,q)}\\
1 & 0 & 1\\
e^{-i\Phi_{13}(p,q)} & 1 & 0
\end{bmatrix},
\]
with
\[
\Phi_{13}(p,q)=\pi-\arccos\!\left(-\frac{3q}{2p}\sqrt{-\frac{3}{p}}\right)
\]
[2301.07435]. The same paper characterizes unitary ACMs of degree $3$ by an explicit coefficient parametrization and a concrete unitary matrix $W_3$.

A closely related constructive program is **Hermitization** of non-Hermitian Hamiltonians. On discretized quantum graphs, one solves
\[
H^\dagger \Theta=\Theta H
\]
for a positive metric $\Theta$ or an indefinite pseudometric $\mathcal P$. For the six-site toy model, the paper gives an explicit sparse $\mathcal P$ and a diagonal positive metric $\Theta^{(\mathrm{diag})}$ in the symmetric case $g=h$; more generally,
\[
\Theta(\alpha)=\Theta^{(\mathrm{diag})}+\alpha \mathcal P
\]
remains positive under an explicit inequality on $\alpha$ [1101.1015]. In finite-dimensional PT-symmetric quantum mechanics, Hermitian matrices arise as the special case of PT-symmetric matrices for which the dynamic metric reduces to the identity; the paper makes this explicit in the $2\times 2$ and $3\times 3$ parameterizations [1002.2676]. These constructions are not usually labeled “Hermitian construction” in the narrow coding-theoretic sense, but they are constructive Hermitian replacements of the same formal type.

## 6. The hermitian \(Q\)-construction and real \(K\)-theory

In higher algebra, Hermitian construction refers to the **hermitian $Q$-construction** for exact or Waldhausen $\infty$-categories with duality. For an exact $\infty$-category $(E,\mathrm{Cof},\mathrm{Fib})$, the simplicial object
\[
Q(E)\colon \Delta^{op}\to \mathrm{Cat}_\infty,\qquad [n]\mapsto Q(E)_{[n]}\subset \mathrm{Fun}(\mathrm{Tw}([n]),E),
\]
is defined by twisted-arrow diagrams satisfying exact-square conditions, fibrancy in one direction, and cofibrancy in the other [2410.07846]. Degree $1$ identifies canonically with the exact-square category:
\[
Q(E)_{[1]}\simeq \mathrm{Amb}(E).
\]

With genuine pro-duality, one obtains the **hermitian upgrade**
\[
Q^h(E)_{[n]}\subset E^{\mathrm{Tw}([n])},
\]
again subject to the same exactness constraints. This makes $Q^h(E)$ a Segal object in $\mathrm{Wald}_\infty^{gp}$ and then in $\mathrm{Wald}_\infty^{gd}$ [2410.07846]. In degree $1$ there is a canonical equivalence
\[
\lambda:\mathrm{Amb}(E)\xrightarrow{\simeq} Q^h(E)_{[1]}.
\]

The comparison target is the **real $S$-construction** $S(E)$, whose $n$-simplices are exact diagrams in $\mathrm{Ar}([n])$ with $A_{i,i}\simeq 0$. After edgewise subdivision, both $S(E)\circ e$ and $Q^h(E)$ are Segal objects. The main theorem states that for a Waldhausen $\infty$-category with genuine duality,
\[
\theta:\ S(E)\circ e \xrightarrow{\simeq} Q^h(E)
\]
is an equivalence of simplicial exact $\infty$-categories with genuine duality [2410.07846]. Consequently,
\[
KR(E)\simeq \Omega^{1,1}\big|Q^h(E)^{\simeq}\big|.
\]

The paper also identifies the Poincaré-category formulation. Using
\[
\mathrm{St}^{gd}\simeq Cat_\infty^p,
\]
the hermitian $Q$-construction of Calmès–Dotto–Harper–Horel agrees levelwise with $Q^h$ after translation of input data, yielding a natural equivalence of genuine $C_2$-spaces [2410.07846].

Fixed points recover classical Hermitian invariants. Hermitian objects are homotopy fixed points,
\[
\mathrm{Herm}(E)\simeq E^{hC_2},
\]
and under the standard exactness and idempotent-completeness hypotheses the $C_2$-fixed points satisfy
\[
(KR(E))^{C_2}\simeq \mathrm{GW}(E).
\]
This places Grothendieck–Witt theory, real $K$-theory, and the hermitian $Q$-construction inside a single genuine-equivariant framework [2410.07846].

This suggests a broad but precise unification: across coding theory, complex geometry, representation theory, matrix analysis, and higher $K$-theory, a Hermitian construction is a procedure that replaces a merely linear, exact, or symmetric structure by one controlled by Hermitian duality, a Hermitian metric, or a Hermitian projector. What changes from field to field is not the formal role of the word *Hermitian*, but the ambient category in which the construction is performed.

Source: https://www.emergentmind.com/topics/hermitian-construction