1: \begin{abstract}

2: Stochastic optimization algorithms have become indispensable in modern machine

3: learning. An unresolved foundational question in this area is the

4: difference between with-replacement sampling and without-replacement

5: sampling --- does the latter have superior convergence rate compared to the

6: former? A groundbreaking result of Recht and R\'e reduces the problem to a

7: noncommutative analogue of the arithmetic-geometric mean inequality where $n$

8: positive numbers are replaced by $n$ positive definite matrices. If

9: this inequality holds for all $n$, then without-replacement sampling indeed

10: outperforms with-replacement sampling. The conjectured Recht--R\'e

11: inequality has so far only been established for $n = 2$ and a special case

12: of $n = 3$. We will show that the Recht--R\'e conjecture is false for general

13: $n$. Our approach relies on the noncommutative Positivstellensatz, which

14: allows us to reduce the conjectured inequality to a semidefinite program

15: and the validity of the conjecture to certain bounds for the optimum

16: values, which we show are false as soon as $n = 5$.

17: \end{abstract}

18: