菜单
多目标量子工作流优化方法论。系统化选择 NISQ 设备上的编译策略、噪声抑制和误差缓解方案。基于 QBalance 框架，涵盖加权目标函数、非支配选择规则、生存乘积误差代理、贝叶斯候选排序和分布诊断。Activation: qbalance, quantum workflow, quantum compilation optimization, NISQ error mitigation, quantum noise suppression, multi-objective quantum strategy.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/hiyenwong/ai_collection --skill qbalance-quantum-workflow-optimization命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Graph-native Python reimplementation of the Information Dynamics of Music (IDyOM) model that represents predictive memories as explicit graph objects for musical expectation modeling and network analysis.
Physics-aware end-to-end deep reinforcement learning methodology for quadcopter control with actuator dynamics modeling.
Reinforced Dreamer methodology for asymmetric reinforcement learning using latent guidance to improve world model representations and behaviors in model-based RL.
基于 SOC 职业分类
正在显示 SKILL.md
| name | qbalance-quantum-workflow-optimization |
| description | 多目标量子工作流优化方法论。系统化选择 NISQ 设备上的编译策略、噪声抑制和误差缓解方案。基于 QBalance 框架，涵盖加权目标函数、非支配选择规则、生存乘积误差代理、贝叶斯候选排序和分布诊断。Activation: qbalance, quantum workflow, quantum compilation optimization, NISQ error mitigation, quantum noise suppression, multi-objective quantum strategy. |
多目标量子工作流优化方法论 — 用于 NISQ 设备上系统化选择编译、噪声抑制和误差缓解策略。
基于论文: QBalance: A Reproducible Multi-Objective Workflow for Quantum Compilation, Noise Suppression, and Error-Mitigation Strategy Selection (arXiv: 2605.02966)
NISQ 工作流的耦合决策维度:
数学建模:
有限多目标策略选择问题:
min_{s ∈ S} f(s) = w₁·f_compile(s) + w₂·f_noise(s) + w₃·f_mitigation(s)
约束条件:
- 硬件拓扑约束 T
- 噪声模型 N
- 预算限制 B
其中 S = 策略空间（编译 × 噪声抑制 × 误差缓解）
def weighted_objective(strategy, weights):
"""
加权多目标函数
strategy: 候选策略
weights: 各目标权重 (w_compile, w_noise, w_mitigation)
返回: 综合评价值
"""
compile_cost = evaluate_compile_cost(strategy)
noise_impact = estimate_noise_impact(strategy)
mitigation_gain = predict_mitigation_effectiveness(strategy)
return (weights['compile'] * compile_cost +
weights['noise'] * noise_impact -
weights['mitigation'] * mitigation_gain)
def non_dominated_selection(candidates):
"""
Pareto 前沿选择
候选策略按多目标 Pareto 最优性筛选
返回非支配解集
"""
pareto_front = []
for c in candidates:
is_dominated = False
for other in candidates:
if other != c and dominates(other, c):
is_dominated = True
break
if not is_dominated:
pareto_front.append(c)
return pareto_front
def dominates(a, b):
"""a 支配 b 当且仅当 a 在所有目标上不劣于 b，且至少一个目标更优"""
all_not_worse = all(a[obj] <= b[obj] for obj in objectives)
at_least_one_better = any(a[obj] < b[obj] for obj in objectives)
return all_not_worse and at_least_one_better
def survival_product_error(circuit, backend_config):
"""
生存乘积误差代理
估计量子电路在给定后端配置下的预期误差
核心思想: 误差随门数量指数增长
error_proxy = ∏(1 - p_gate_i) 对每个门 i
其中 p_gate_i 是第 i 个门的误差概率
"""
survival_prob = 1.0
for gate in circuit.gates:
gate_error = backend_config.get_gate_error(gate.type)
survival_prob *= (1 - gate_error)
return 1 - survival_prob # 返回误差代理
def bayesian_candidate_ordering(candidates, historical_data):
"""
使用贝叶斯线性回归对候选策略排序
基于历史实验数据构建代理模型
预测未评估策略的性能
"""
# 构建特征矩阵
X = extract_features(candidates)
y = historical_data['observed_performance']
# 贝叶斯线性回归
posterior_mean, posterior_var = bayesian_linear_fit(X, y)
# 预测 + 不确定性
predictions = posterior_mean @ X_new
uncertainties = posterior_var @ X_new
# Thompson Sampling 式排序
scores = predictions + alpha * uncertainties
return rank_by(scores)
def distributional_diagnostics(results):
"""
策略性能分布诊断
分析候选策略性能分布的统计特性:
- 均值/方差
- 偏度/峰度
- 分布拟合
- 异常值检测
"""
diagnostics = {
'mean': np.mean(results),
'variance': np.var(results),
'skewness': scipy.stats.skew(results),
'kurtosis': scipy.stats.kurtosis(results),
'outliers': detect_outliers(results),
'distribution_fit': fit_distribution(results)
}
return diagnostics
class QBalanceWorkflow:
"""
QBalance 多目标量子编译工作流
流程: 策略枚举 → 特征提取 → 贝叶斯排序 → 非支配选择 → 执行验证
"""
def __init__(self, backend, circuits, weights=None):
self.backend = backend
self.circuits = circuits
self.weights = weights or {
'compile_cost': 0.3,
'noise_impact': 0.4,
'mitigation_gain': 0.3
}
def run(self):
# 1. 枚举候选策略
candidates = self.enumerate_strategies()
# 2. 提取特征
features = self.extract_features(candidates)
# 3. 贝叶斯排序
ranked = self.bayesian_rank(candidates, features)
# 4. 非支配选择
pareto = self.non_dominated_select(ranked)
# 5. 执行最优策略
results = self.execute_best(pareto)
return results
def enumerate_strategies(self):
"""生成编译 × 噪声抑制 × 误差缓解策略组合"""
compilation_strategies = [
'sabre_layout', 'trivial_layout', 'dense_layout'
]
routing_strategies = [
'sabre_routing', 'stochastic_routing', 'lookahead_routing'
]
mitigation_strategies = [
, ,
]
(product(
compilation_strategies,
routing_strategies,
mitigation_strategies
))
class ReproducibleQuantumExperiment:
"""
可复现量子实验编排器
确保实验结果的可复现性和可追溯性
"""
def __init__(self, experiment_config):
self.config = experiment_config
self.artifact_store = ArtifactStore()
def run(self):
# 记录所有配置
self.artifact_store.save_config(self.config)
# 固定随机种子
np.random.seed(self.config['seed'])
# 执行策略选择
strategy = self.select_strategy()
# 记录策略决策
self.artifact_store.save_strategy(strategy)
# 执行量子实验
results = self.execute(strategy)
# 保存所有中间产物
self.artifact_store.save_results(results)
return results
class AdaptiveErrorMitigation:
"""
自适应误差缓解策略选择
根据电路特征和后端噪声自动选择最优缓解方案
"""
MITIGATION_STRATEGIES = {
'shallow_circuit': 'readout_mitigation',
'deep_circuit': 'zne',
'high_connectivity': 'randomized_compiling',
'low_connectivity': 'dynamical_decoupling',
}
def select(self, circuit, backend):
features = self.analyze_circuit(circuit)
noise_profile = backend.get_noise_profile()
# 匹配策略
if features['depth'] > THRESHOLD:
return 'zne'
elif features['connectivity_requirement'] > backend.connectivity:
return 'sabre_routing'
else:
return 'readout_mitigation'
from qiskit.transpiler import PassManager
from qiskit.transpiler.passes import (
SabreLayout, SabreSwap,
Optimize1qGates, CommutativeCancellation
)
pm = PassManager([
SabreLayout(coupling_map),
SabreSwap(coupling_map),
Optimize1qGates(),
CommutativeCancellation()
])
def zero_noise_extrapolation(circuit, backend, scale_factors=[1, 3, 5]):
"""
零噪声外推
在多个噪声尺度下执行电路
外推到零噪声极限
"""
results = []
for scale in scale_factors:
scaled_circuit = scale_gates(circuit, scale)
result = execute(scaled_circuit, backend)
results.append(result)
# 多项式外推到 scale=0
return extrapolate_to_zero(scale_factors, results)
def apply_dynamical_decoupling(circuit, dd_sequence='XY4'):
"""
应用动态解耦序列抑制退相干
XY4: X - Y - X - Y 脉冲序列
"""
if dd_sequence == 'XY4':
sequence = [XGate(), YGate(), XGate(), YGate()]
elif dd_sequence == 'CPMG':
sequence = [YGate(), YGate()]
return insert_dd_sequence(circuit, sequence)
pip install qiskit qiskit-ibm-runtime qiskit-aer
pip install numpy scipy pandas
pip install qbalance # 如果已发布
QBalance: Multi-Objective Quantum Workflow Optimization - 系统化 NISQ 策略选择方法论