Qubit mapping and routing (QMR) is a critical step in quantum computing because it maps logical qubits to a physical layout while properly routing all gates. In QMR, achieving high qubit fidelity is essential as it ensures reliable execution of a quantum circuit on a quantum computer. However, existing QMR algorithms focus on optimizing the average fidelity (AVF) across all qubits, ignoring theworst case fidelity(WCF) of individual qubits. In practice, a particularly low-fidelity qubit can greatly diminish the reliability of a quantum circuit, as a single qubit failure can lead to the failure of the entire circuit. To address this issue, we introduce QDP, a parallel dynamic programming (DP)-based QMR algorithm that optimizes the WCF while balancing the overall AVF. QDP leverages a state graph as the optimal substructure for our DP formulation to capture all valid quantum circuit execution orders. To enhance performance, QDP introduces a parallel DP-based state traversal algorithm that maximizes WCF through remapping and reduces the size of the state graph via progressive state expansion. Compared with state-of-the-art QMR algorithms, QDP can enhance the WCF by up to 33.88% while achieving comparable AVF on real quantum computers with up to 53 qubits.
Random circuit models often describe local dynamics using generic two-qubit gates, which have proven successful in capturing entanglement growth and operator spreading in many contexts. This approach naturally leads to the expectation that detailed gate structure plays only a limited role in coarse-grained entanglement and scrambling diagnostics. We show that the internal structure of multipartite circuit primitives can significantly influence these dynamical rates, even within a fixed random-circuit architecture. To investigate this, we study an exactly simulable family of Clifford quantum circuits built from fixed
n-qubit graph-state preparation unitaries, which we treat as elementary building blocks. Specifically, we consider a one-dimensional chain of
N qubits initialized in a product state and evolved by layers in which nonoverlapping length-
n blocks are placed at uniformly random positions with sparsity
α. We find that different choices of graph-state building blocks lead to strongly varying dynamical rates. Graph states that are inequivalent under local Clifford (LC) transformations generate sharply different entanglement velocities
vE and butterfly velocities
vB, even though the circuits are drawn from the same ensemble with identical architecture and randomness parameters. We further show that this hierarchy is captured by two complementary block-level characteristics: the distribution of entanglement across internal bipartitions of the graph state, which correlates with
vE, and a graph-theoretic connectivity profile across bipartitions, which correlates with
vB. Neither descriptor alone fully determines the dynamics; rather, entanglement growth and operator spreading are controlled by distinct structural features of the local circuit blocks. Notably, AME states appear among the fastest scrambling building blocks within the ensembles studied here.
Testing quantum programs on NISQ (Noisy Intermediate-Scale Quantum) backends is challenging because the noise disturbs outcome distributions and can affect pass/fail decisions. We present Q-BRIDGE, a graph learning-based approach that converts noisy observations into denoised distributions suitable for oracle-based verification. Q-BRIDGE uses a graph transformer architecture to encode a transpiled quantum circuit, capturing the characteristics of its gates and their connectivity; the physical backend information is encoded together with the logical structure of the circuit. An additional conditioning layer, based on FiLM (Feature-Wise Linear Modulation), takes the encoding as input and integrates noisy observations to produce denoised outcomes. We evaluate Q-BRIDGE on 23 IBM noise backends and 6 circuit families representative of practical workloads. In the first setting, we train a separate Q-BRIDGE model for each backend; in the second setting, we train a single general model shared across all backends. Across both settings, Q-BRIDGE outperforms the state-of-the-art baseline in noise mitigation by a large margin. In testing scenarios with noisy executions, Q-BRIDGE achieves 93.97%-94.90% precision and 82.50%-83.51% recall in detecting bug-induced test failures, significantly outperforming the state-of-the-art baseline. These results indicate that considering the graph structure of the transpiled circuits and the physical characteristics of specific quantum backends is a practical route to more reliable noise-aware quantum program testing.