这并非小问题：当梯度估计变得极其微弱时，任何优化策略都会变得低效，变分量子算法的可扩展性受到严重限制。即使使用反向传播等技巧，也无法根本绕开由电路结构和初始化导致的统计性下界。

Emily Laurence is a journalist, freelance writer and certified health coach living in Raleigh, North Carolina. She specializes in writing about mental health, healthy aging and overall wellness. For ...