As existing approaches to get around the restriction of the no-go theorem generally exhibit high ancillary qubit overhead costs, we propose a scheme for universal fault-tolerant quantum computation by using the pieceable fault-tolerant protocol and code switching techniques. In additional, by utilizing the construction of the pieceable fault-tolerant circuit, we adopt a decoding strategy based on a deep-neural-network algorithm to improve the error threshold of the non-Clifford logical gate circuit. We describe our universal construction in detail with a two-level nonuniform concatenated 25-qubit code and perform numerical simulations to analyze the depolarizing noise threshold of a given universal gate set with this code. The resources required to implement this universal gate set are also estimated to further demonstrate the efficiency of our scheme. We compare these results with the existing universal concatenation methods and conclude that our method outperforms them in terms of the lower bound of the error threshold and qubit resource overhead.

中文翻译：

---

通用量子计算的级联分段容错方案

由于绕开定理的限制的现有方法通常显示出较高的辅助qubit开销成本，因此，我们提出了一种采用分段容错协议和代码转换技术的通用容错量子计算方案。另外，通过利用分段容错电路的结构，我们采用了基于深度神经网络算法的解码策略，以提高非Clifford逻辑门电路的错误阈值。我们使用两级非均匀级联的25量子位代码详细描述了通用结构，并进行了数值模拟，以分析使用该代码的给定通用门集的去极化噪声阈值。还估计了实现此通用门集所需的资源，以进一步证明我们的方案的效率。我们将这些结果与现有的通用级联方法进行了比较，并得出结论，我们的方法在错误阈值的下限和qubit资源开销方面优于它们。