[论文解读] Tight and Efficient Gradient Bounds for Parameterized Quantum Circuits

Alistair Letcher, Stefan Woerner+1 more|arXiv (Cornell University)|Sep 22, 2023

Quantum Computing Algorithms and Architecture被引用 5

一句话总结

该论文在不依赖t-设计假设的前提下，推导出参数化量子线路的紧致非渐近梯度界，证明了含有非零局部项的混合可观测量可避免 barren plateaus。应用于量子生成对抗网络（qGANs）时，表明浅层生成器配合适当设计的判别器，可在量子比特数量增加时保持恒定的1-局部梯度权重，从而实现无指数梯度抑制的可扩展训练——通过在高斯混合分布上成功训练16量子比特qGAN得到验证。

ABSTRACT

The training of a parameterized model largely depends on the landscape of the underlying loss function. In particular, vanishing gradients are a central bottleneck in the scalability of variational quantum algorithms (VQAs), and are known to arise in various ways. However, a caveat of most existing gradient bound results is the requirement of t-design circuit assumptions that are typically not satisfied in practice. In this work, we loosen these assumptions altogether and derive tight upper and lower bounds on loss and gradient concentration for a large class of parameterized quantum circuits and arbitrary observables, which are significantly stronger than prior work. Moreover, we show that these bounds, as well as the variance of the loss itself, can be estimated efficiently and classically-providing practical tools to study the loss landscapes of VQA models, including verifying whether or not a circuit/observable induces barren plateaus. In particular, our results can readily be leveraged to rule out barren plateaus for a realistic class of ansätze and mixed observables, namely, observables containing a non-vanishing local term. This insight has direct implications for hybrid Quantum Generative Adversarial Networks (qGANs). We prove that designing the discriminator appropriately leads to 1-local weights that stay constant in the number of qubits, regardless of discriminator depth. This implies that qGANs with appropriately chosen generators do not suffer from barren plateaus even at scale-making them a promising candidate for applications in generative quantum machine learning. We demonstrate this result by training a qGAN to learn a 2D mixture of Gaussian distributions with up to 16 qubits, and provide numerical evidence that global contributions to the gradient, while initially exponentially small, may kick in substantially over the course of training.

研究动机与目标

• 解决由barren plateaus引起的变分量子算法可扩展性瓶颈。
• 在不依赖t-设计假设的前提下，推导出参数化量子线路的梯度界，而这类假设在实际中往往不现实。
• 分析由局部项和全局项组成的混合可观测量是否在实际设置中引发barren plateaus。
• 建立混合量子生成对抗网络（qGANs）在无指数梯度抑制条件下可训练的条件。
• 通过紧致梯度分析，证明在连续分布上实现可扩展qGAN训练的实际可行性。

提出的方法

• 仅基于可构造性验证的设计选择，为一大类参数化量子线路和任意可观测量，推导出梯度集中性的紧致上下界。
• 提出一种新颖的证明技术，避免使用t-设计假设，转而依赖基本对称性和期望值论证。
• 将梯度界应用于混合qGANs，证明即使在深度经典判别器下，1-局部梯度贡献的权重也保持与量子比特数量无关的恒定。
• 通过将可观测量分解为局部和全局项，证明局部项主导梯度方差，从而防止指数级抑制。
• 在数值实验中采用同时扰动随机逼近（SPSA）方法估计梯度，确保与硬件相关的评估。
• 通过在最多16量子比特的二维高斯混合分布上进行qGAN的数值训练，测量相对熵和概率密度函数保真度，验证结果。

Figure 1 : Illustration and examples of local vs. mixed vs. global observables. All observables act on $n$ qubits, but we omit the identity terms for convenience. For instance, $X$ denotes $X\otimes I^{\otimes\,n-1}$ .

实验结果

研究问题

• RQ1 能否在不依赖t-设计假设的前提下，为参数化量子线路推导出紧致梯度界？
• RQ2 含有非零局部项的混合可观测量是否在实际量子线路中引发barren plateaus？
• RQ3 即使判别器任意深，混合qGANs是否仍能避免barren plateaus？
• RQ4 在qGAN训练过程中，随着量子比特数量增加，全局与局部梯度贡献的行为如何？
• RQ5 根据理论界预测，具有浅层生成器的qGAN是否能在大规模下成功学习复杂连续分布？

主要发现

• 该论文在不依赖t-设计假设的前提下，为一大类参数化量子线路建立了紧致梯度界，提供了比以往工作更实用且更通用的框架。
• 含有非零局部项的混合可观测量不会引发barren plateaus，因为局部梯度贡献的方差始终保持远离零。
• 在qGAN中，即使判别器深度增加，1-局部梯度权重也保持与量子比特数量无关的恒定，这归因于可观测量和线路设计的结构。
• 数值实验表明，qGAN在最多16量子比特的二维高斯混合分布上成功训练，相对熵随时间减少，性能与经典GAN相当。
• 全局梯度贡献初始时虽呈指数级微小，但在训练过程中可显著增长，表明其在优化动力学中具有非平凡作用。
• 结果表明，用局部损失函数近似完整损失函数会增强梯度集中性，但可能引入不期望的局部极小点，与常见假设相反。

Figure 2 : EfficientSU2 circuit eff ( 2023 ) with pairwise entanglement and $(R_{Y},R_{Z})$ rotation layers, illustrating the circuit structure introduced in Theorem 1 . After the first layer $R_{0}$ of single-qubit orthogonal rotations, any Clifford gates $W_{k}$ and any multi-qubit rotations $R_{P