---
title: 'Symmetry Verification: Quantum Error Mitigation'
url: https://www.emergentmind.com/topics/symmetry-verification-sv
type: topic
---

# Symmetry Verification: Quantum Error Mitigation

Symmetry verification (SV) denotes a family of procedures that exploit known invariances to test, enforce, or certify membership in a symmetry sector. In quantum computing, the dominant usage is a quantum error mitigation technique that projects a noisy state onto the eigenspace of a symmetry operator and discards symmetry-violating runs; later work extends the idea from states to circuits and to channels. In few-particle physics, a distinct earlier usage denotes symmetrization verification of two-particle states by one-particle detection and tomography. Related verification literatures use symmetry to reduce proof obligations, infer quantified invariants, or certify symmetry-breaking constraints, but these are methodologically distinct from post-selection-based quantum SV [1902.11258] [2204.05852] [2503.13114] [1108.0646].

## 1. State-level SV as projection onto a symmetry sector

In the conventional, state-oriented formulation, one assumes that the ideal output state \(|\psi\rangle\) is stabilized by some unitary \(\hat S\in\mathcal P^N\), so that \(\hat S|\psi\rangle=|\psi\rangle\) and \([\hat S,H]=0\) for the relevant Hamiltonian. The Hilbert space then splits as \(\mathcal H=\mathcal H_+\oplus\mathcal H_-\), with projectors
\[
P_+ = \frac{I+\hat S}{2}, \qquad P_-=\frac{I-\hat S}{2}.
\]
If a noisy preparation yields \(\rho\), symmetry verification implements
\[
\rho \mapsto \rho_{\mathrm{SV}}=\frac{P_+\rho P_+}{\mathrm{Tr}[P_+\rho]},
\]
that is, post-selection onto the desired symmetry subspace [1902.11258].

A more general group-theoretic formulation replaces a single Pauli symmetry by a finite or compact group \(G\) with unitary representation \(\{U_g\}\). The Hilbert space decomposes into irreducible-representation sectors labeled by \(\mu\), with projectors
\[
P_\mu = \frac{d_\mu}{|G|}\sum_{g\in G}\chi_\mu^*(g)U_g.
\]
Conventional SV then prepares a noisy output \(\rho_{\mathrm{out}}=\mathcal U_\mathcal N(\rho_{\mathrm{in}})\), measures the POVM \(\{P_\mu\}\), and post-selects the desired sector \(\mu_0\), yielding
\[
\rho_{\mathrm{SV}}=\frac{P_{\mu_0}\rho_{\mathrm{out}}P_{\mu_0}}{\mathrm{Tr}[P_{\mu_0}\rho_{\mathrm{out}}]}.
\]
This formulation makes explicit that conventional SV is state-level: it requires the input to lie in a symmetry sector and uses symmetry only at the end [2503.13114].

The operational meaning of SV is therefore selective rejection of symmetry-breaking noise. If the noise channel can be written as \(\mathcal N(\cdot)=(1-p)\,\cdot + p\sum_j E_j \cdot E_j^\dagger\), and if each error Kraus \(E_j\) satisfies \(P_{\mu_0}E_jP_{\mu_0}=0\), then after SV the surviving state is proportional to the ideal output, with success rate \(1-p\). This identifies the central limitation of state-level SV: it filters only errors that move weight outside the target sector, while in-sector errors remain invisible [2503.13114].

## 2. Embedding SV in VQE and QAOA

Within VQE, SV can be implemented entirely in post-processing. For an \(N\)-qubit Hamiltonian \(H\) commuting with a Pauli-string symmetry \(\hat S\), the corrected expectation value of any Pauli observable \(\hat P\) in target sector \(s=\pm1\) is
\[
\langle \hat P\rangle^{\mathrm{(SV)}}=
\frac{\langle \hat P\rangle^{\mathrm{(raw)}}+s\,\langle \hat S\hat P\rangle^{\mathrm{(raw)}}}
{1+s\,\langle \hat S\rangle^{\mathrm{(raw)}}}.
\]
The symmetry-verified density matrix is then reconstructed as
\[
\rho^{\mathrm{(SV)}}=\frac{1}{2^N}\sum_{\hat P\in\mathcal P^N}\langle \hat P\rangle^{\mathrm{(SV)}}\hat P.
\]
For the \(\mathrm{H}_2\) molecule in a minimal STO-3G basis, Bravyi-Kitaev mapped and with two qubits projected out, the effective two-qubit Hamiltonian commutes with \(S=ZZ\), so total two-qubit parity is conserved [1902.11258].

The same work couples SV to positivity restoration. Direct linear-inversion tomography may yield \(\rho_{\mathrm{raw}}\) with small negative eigenvalues, so one solves the semidefinite program
\[
\min_{\rho_{\mathrm{phys}}}\ \|\rho_{\mathrm{phys}}-\rho_{\mathrm{raw}}\|_F^2
\quad\text{subject to}\quad
\rho_{\mathrm{phys}}\succeq0,\ \mathrm{Tr}[\rho_{\mathrm{phys}}]=1.
\]
SV is then applied to \(\rho_{\mathrm{phys}}\) or to its Pauli coefficients. In this setting, relaxation and residual excitation change total excitation number and therefore violate parity, whereas dephasing commutes with \(ZZ\) and is not removed by the projection [1902.11258].

For QAOA, the standard example is MaxCut on \(N\) qubits, where
\[
H=\frac12\sum_{(ij)\in E}(I-Z_iZ_j)
\]
is invariant under the global bit-flip \(\mathbb Z_2\) symmetry
\[
S=\prod_{i=1}^N X_i.
\]
If \(|\psi\rangle\) is the ideal QAOA state, then \(S|\psi\rangle=|\psi\rangle\), and measuring \(S\) while retaining only runs with outcome \(+1\) implements
\[
\rho \mapsto \rho_{\mathrm{SV}}=\frac{P_+\rho P_+}{\mathrm{Tr}[P_+\rho]}.
\]
Under local noise after each QAOA layer, with \(U_t=\exp(-i\gamma_tH)\exp(-i\beta_tB)\), the ideal fidelity is \(F_{\mathrm{noisy}}=\langle\psi|\rho_d|\psi\rangle\), while after SV
\[
F_{\mathrm{SV}}=\frac{F_{\mathrm{noisy}}}{\mathrm{Tr}[P_+\rho_d]}.
\]
Thus the fidelity improvement ratio is
\[
\frac{F_{\mathrm{SV}}}{F_{\mathrm{noisy}}}=\frac{1}{\mathrm{Tr}[P_+\rho_d]}.
\]
This makes explicit that SV improves fidelity by conditioning on the weight remaining in the target parity sector [2204.05852].

## 3. Noise models, acceptance probability, and finite-shot performance

Under depolarizing single-qubit noise,
\[
\mathcal E_i(\rho)=(1-p)\rho+\frac{p}{3}(X_i\rho X_i+Y_i\rho Y_i+Z_i\rho Z_i),
\]
the QAOA analysis derives for depth \(d=1\)
\[
\mathrm{Tr}[P_+\rho_1]=\sum_{m=0}^N(1-p)^{N-m}(p/3)^m f(m),
\]
with
\[
f(m)=\tfrac12\binom Nm(1+3^m)\ \text{for \(m\) even},\qquad
f(m)=\tfrac12\binom Nm(3^m-1)\ \text{for \(m\) odd}.
\]
Writing
\[
\mathcal F(N,p)\equiv \sum_{m=0}^N(1-p)^{N-m}(p/3)^m f(m),
\]
one obtains
\[
\frac{F_{\mathrm{SV}}}{F_{\mathrm{noisy}}}=\frac{1}{\mathcal F(N,p)},
\]
and in practice replaces \(N\mapsto Nd\) when errors accumulate over \(d\) layers. Under dephasing,
\[
\mathcal E_i(\rho)=(1-p)\rho+p\,Z_i\rho Z_i,
\]
only even-weight errors commute with \(S\), giving the closed form
\[
\frac{F_{\mathrm{SV}}}{F_{\mathrm{noisy}}}
=
\frac{2}{1+(1-2p)^{Nd}}.
\]
These formulas identify a basic trade-off: fidelity improves only by discarding runs, so acceptance probability controls sampling overhead [2204.05852].

A finite-shot theory makes this trade-off explicit. Let \(P\) be the projector onto the symmetry sector, let \(O\) be a symmetry-compatible observable, let \(a(p)=\mathrm{Tr}[P\rho_p]\) be the acceptance probability, let \(A_s\in\{0,1\}\) be the post-selection flag, let \(W_s\in[-K_W,K_W]\) be the measurement result of \(O\) when \(A_s=1\), and set \(Z_s=A_sW_s\). The clipped estimator is
\[
\widehat\mu_{\rm SV}^{\rm clip}
=
\frac{\frac1B\sum_{s=1}^B Z_s}
{\max\!\bigl\{\frac1B\sum_{s=1}^B A_s,\;a(p)/2\bigr\}}.
\]
For every finite \(B\ge1\),
\[
MSE_{\rm SV}(p,B)
=
\bigl[\mu_{\rm SV}(p)-\mu_\star\bigr]^2
+
\frac{\mathrm{Var}(Z_1-\mu_{\rm SV}(p)A_1)}{B}
+
\rho_{\rm SV}(p,B),
\]
with remainder bounded by
\[
\rho_{\rm SV}(p,B)\le \frac{C_{\rm SV}(p)}{B^2}+C'_{\rm SV}(p)\exp[-Ba(p)/12].
\]
Because Bernoulli postselection yields
\[
c_{\rm SV}(p)=\frac{\mu_{\rm SV}\,\mathrm{Var}(A_1)-\mathrm{Cov}(Z_1,A_1)}{a(p)^2}=0,
\]
there is no ratio-bias cross term \(\propto b_{\rm SV}c_{\rm SV}/B\) [2606.15464].

The same analysis isolates two irreducible effects. First, if \(\rho_p=\rho_\star+p\,\Delta+O(p^2)\) and \(\Delta_{\rm undet}=P\Delta P\), then
\[
\mu_{\rm SV}(p)=\mu_\star+p\,\delta_{\rm SV}+O(p^2),
\]
with
\[
\delta_{\rm SV}
=
\mathrm{Tr}[O\,\Delta_{\rm undet}]
-
\mu_\star\,\mathrm{Tr}[\Delta_{\rm undet}],
\]
so the bias floor scales as
\[
b_{\rm SV}(p)^2=\delta_{\rm SV}^2 p^2+O(p^3).
\]
Second, postselection inflates variance by \(1/a(p)\):
\[
v_{\rm SV}(p)=\frac{\mathrm{Var}(W_1\mid A_1=1)}{a(p)}.
\]
When comparing SV against an unmitigated estimator, the critical sample budget scales as \(B_0(p)=\Theta(p^{-2})\) or \(B_0(p)=\Theta(p^{-1})\), depending on whether the leading variance terms cancel. This result replaces an asymptotic “SV always helps” narrative by an operating-window picture in which benefit depends jointly on noise rate, acceptance probability, and finite budget [2606.15464].

## 4. From state symmetries to circuit and channel symmetries

State-based SV fails when the output state does not share a fixed stabilizer independent of the input, even though the circuit itself obeys useful commutation relations. Circuit-oriented SV addresses this by using spatio-temporal stabilizers (STS). For a circuit \(C=C_N\cdots C_1\), an STS specifies partial operators \(S_0,\ldots,S_N\) at times \(t=0,\ldots,N\) so that
\[
S\,C=C,
\]
or equivalently
\[
S = C\,S_0^\dagger C_1^\dagger S_1^\dagger\cdots C_N^\dagger S_N^\dagger.
\]
Hence,
\[
S_N C_N S_{N-1}\cdots C_1 S_0 = C.
\]
An ancilla-assisted check inserts controlled-\(S_n\) operations at the designated times and measures the ancilla in the \(X\)-basis; a \(-1\) outcome signals a symmetry violation. If a noisy Kraus component anti-commutes with one of the \(S_n\), it is filtered by the check. In a circuit with \(M\) faulty gates, any residual error must involve an even number of anti-commuting faults, so the error probability scales \(\sim O(\epsilon^2)\) rather than \(O(\epsilon)\). The sampling overhead factor is
\[
SOF(C,\{STS\})=1/p_{\rm pass}-1.
\]
A related commutativity test uses the quantum switch, whose controlled superposition of orders \(AB\) and \(BA\) filters anti-commutator-violating noise without full tomography [2112.13904].

Channel-level generalization is provided by symmetric channel verification (SCV). A channel \(\mathcal U\) is \(G\)-symmetric if
\[
U_g\,\mathcal U(\cdot)\,U_g^\dagger=\mathcal U(U_g\,\cdot\,U_g^\dagger)
\quad \forall g\in G.
\]
If the ideal channel commutes with an operator \(S\) having \(M\) distinct eigenvalues and projectors \(\{\Pi_i\}\), define
\[
V_S=\sum_{j=0}^{M-1}e^{2\pi ij/M}\Pi_j.
\]
A quantum-phase-estimation-like circuit applies controlled powers of \(V_S\), the noisy channel \(\mathcal U_\mathcal N\), controlled inverse powers of \(V_S\), an inverse QFT on the ancilla, and post-selection on ancilla outcome \(0^m\). The resulting trace-nonincreasing map is
\[
\Theta_S^{\det}[\mathcal U_\mathcal N](\cdot)
=
\sum_{i,j=0}^{M-1}\Pi_i\,\mathcal U_\mathcal N(\Pi_i\,\cdot\,\Pi_j)\,\Pi_j.
\]
If the noise satisfies \(\sum_i \Pi_i N_j \Pi_i \propto I\), then \(\Theta_S^{\det}[\mathcal U_\mathcal N](\cdot)\propto \mathcal U(\cdot)\); in particular, for \(\mathcal N=(1-p)I+p\sum_j N_j\cdot N_j^\dagger\) with \(\Pi_iN_j\Pi_i=0\), SCV yields pure \(\mathcal U\) with success rate \(1-p\) [2503.13114].

A hardware-efficient variant, virtual SCV (VSCV), reconstructs the same purification only at the level of expectation-value estimation. It uses one ancilla and controlled-Pauli Clifford gadgets to realize virtual supermaps
\[
\Theta^{\mathrm{vir}}_{ijkl}[\mathcal U_\mathcal N](\cdot)
=
\frac12\bigl[
P_k\,\mathcal U_\mathcal N(P_i\cdot P_j)\,P_l
+
P_l\,\mathcal U_\mathcal N(P_j\cdot P_i)\,P_k
\bigr],
\]
from which \(\Theta_S^{\det}\) is reconstructed by Pauli expansion. A depolarizing error \(p\) on the ancilla introduces only a constant factor \((1-4p/3)\), which cancels in normalization. In the Clifford-only regime, SCV under Pauli symmetry is stated to be the optimal purification method [2503.13114].

## 5. Experimental realizations and observed regimes

On a two-qubit circuit QED processor for \(\mathrm{H}_2\), SV was implemented with an exchange-interaction ansatz gate
\[
U_{\rm exch}(\theta)=
\begin{pmatrix}
1&0&0&0\\
0&\cos\theta&i\sin\theta&0\\
0&i\sin\theta&\cos\theta&0\\
0&0&0&1
\end{pmatrix},
\]
which preserves total excitation number and hence parity \(ZZ\). Full tomography used 36 tensor-product single-qubit pre-rotations, with \(10^3\) shots per generation during optimization and \(10^5\) shots for final tomography. Over the \(\mathrm{H}_2\) dissociation curve, the reported averages were
\[
\langle \Delta E^{\rm(SV)}/\Delta E^{\rm(raw)}\rangle \approx 1/10,
\qquad
\langle (1-F)^{\rm(SV)}/(1-F)^{\rm(raw)}\rangle \approx 1/9,
\]
with raw VQE energy errors typically \(\sim10^{-2}\) Hartree and symmetry-verified errors \(\sim10^{-3}\) Hartree [1902.11258].

For QAOA on an 11-qubit IonQ trapped-ion device with all-to-all connectivity, experiments covered all non-isomorphic graphs of size \(N=3,4\) at depths \(d=1,2,3\), using a global parity-check circuit. Across 3- and 4-node graphs and \(d\le 3\), SV changed the approximation ratio from \(-4.5\%\) up to \(+19.2\%\), with mean \(+4.3\%\). The post-selection success probability \(\mathrm{Tr}[P_+\rho]\) was \(\approx75\%-90\%\), and the reported two-qubit error rates of \(1\%-3\%\) matched the simulation-predicted regime in which SV improves the QAOA objective [2204.05852].

Circuit-level STS checks have been demonstrated on QFT and QAOA instances. For QFT under asymmetric Pauli channels with \(X\!:\!Z\) error-rate ratio \(10\!:\!1\), purity increased by \(\sim10\%-20\%\) at \(SOF\approx0.5-1.5\); under depolarizing noise, net benefit appeared when two-qubit gate error was \(\gtrsim0.5\%\). For single-stage QAOA, single-ancilla STS produced a purity boost of \(\sim5\%-10\%\) at \(SOF\approx0.2-0.8\), while a cat-state two-ancilla STS added \(\sim3\%\) purity. On IBMQ_Lima, a four-gate XX-rotation circuit protected by a reduced STS on one qubit showed a total error probability drop from \(\approx14.2\%\) to \(\approx8.5\%\) [2112.13904].

SCV and VSCV have likewise been evaluated on Hamiltonian simulation and phase estimation workloads. For an \(n=8\) Heisenberg chain under local depolarizing noise, SCV on the entire circuit reduced the trace distance \(\|\rho_{\mathrm{noisy}}-\rho_{\mathrm{ideal}}\|_1\) from \(O(p_{\mathrm{err}})\) to \(O(p_{\mathrm{err}}^2)\). In a SELECT operation for 2D Fermi-Hubbard qubitization, applying virtual SCV on all idling system qubits reduced the system-noise contribution to \(O(1)\) per qubit, so the total noise became \(O(d\log n)\), yielding a quadratic improvement in \(n\) [2503.13114].

Finite-shot comparisons against virtual distillation further sharpen the empirical picture. Gate-level simulation and archived runs on ibm_marrakesh and ibm_kingston reported that, at \(B=8192\), SV had \(MSE\approx10^{-2}\), significantly below the unmitigated \(MSE\approx4\times10^{-2}\), while virtual distillation remained above SV and could suffer denominator instability. In the tested QAOA instances, calibrated SV was therefore the practical winner, but the same study explicitly rejects a universal-winner interpretation and instead emphasizes regime structure [2606.15464].

## 6. Distinct usages and related symmetry-based verification paradigms

A separate meaning of the phrase appears in two-particle quantum physics. The 2011 proposal for symmetrization verification considers two identical particles in a state
\[
|2;f,g\rangle
=
\iint d^3p\,d^3q\,[f(p)g(q)\pm g(p)f(q)]\,a_p^\dagger a_q^\dagger|0\rangle,
\]
with the upper sign for bosons and the lower sign for fermions. Detection of one particle at position \(R\) is modeled by
\[
\psi(R)=\int d^3p\,\phi_p(R)\,a_p,
\]
and the surviving one-particle state is
\[
|\chi(R)\rangle
=
\psi(R)|2;f,g\rangle
=
\int d^3p\,[f(R)g(p)\pm g(R)f(p)]\,a_p^\dagger|0\rangle.
\]
Equivalently,
\[
|\chi(R)\rangle = f(R)|1_g\rangle \pm g(R)|1_f\rangle.
\]
The proposal does not follow a Hong-Ou-Mandel-type approach; instead it uses one-particle destructive detection plus state tomography of the survivor, and it can also generate single-particle superposition states [1108.0646].

Outside quantum information, symmetry-based verification has additional meanings. In combinatorial optimisation, “Certified Symmetry and Dominance Breaking for Combinatorial Optimisation” realizes symmetry verification through machine-checkable cutting-planes certificates, lex-leader pseudo-Boolean constraints, and dominance-based inference rules, enabling certification of symmetry breaking in SAT, maximum clique, and constraint programming [2203.12275]. In parameterized distributed protocols, SymIC3 and IC3PO use symmetry groups, clause orbits, and quantifier inference from orbit structure to derive quantified inductive invariants without a priori quantifier-template search [2103.14831]. In symbolic model checking, dynamic symmetry reduction avoids constructing the orbit relation BDD and instead interleaves reachability with on-the-fly representative selection and state symmetries [1006.1416]. In Coq proofs for triangulation algorithms, explicit rotation, reflection, and orbit lemmas collapse “without-loss-of-generality” case splits into canonical representatives [1809.00559]. In hybrid-systems scenario verification, SceneChecker uses symmetry abstractions and refinement to build quotient automata over geometrically similar plan segments and reports substantial speedups while using existing reachability tools as black-box subroutines [2011.10713].

These literatures share a common structural idea—explicit exploitation of symmetry to reduce ambiguity, redundancy, or noise—but they do not instantiate a single formalism. In quantum error mitigation, SV is fundamentally a projection-and-postselection procedure; in symmetrization verification it is a measurement-induced test of exchange symmetry; in formal verification and optimisation it is a certification or reduction methodology. A plausible implication is that “symmetry verification” is best treated as a family resemblance term whose precise meaning is fixed by the ambient theory of states, channels, clauses, or transition systems.

Source: https://www.emergentmind.com/topics/symmetry-verification-sv