1: \begin{abstract}

2: 

3: The precise description of quantum nuclear

4: fluctuations in atomistic

5: modelling is possible by employing path integral techniques,

6: which involve a considerable computational overhead due to the 

7: need of simulating multiple replicas of the system. 

8: Many approaches have been suggested

9: to reduce the required number of replicas. 

10: Among these, high-order factorizations

11: of the Boltzmann operator are particularly attractive

12: for high-precision and low-temperature scenarios. 

13: Unfortunately, to date several technical challenges have prevented

14: a widespread use of these approaches to study 

15: nuclear quantum effects in condensed-phase systems. 

16: Here we introduce an inexpensive molecular dynamics scheme 

17: that overcomes these limitations, thus making it 

18: possible to exploit the improved convergence of 

19: high-order path integrals without having to sacrifice

20: the stability, convenience and flexibility of conventional

21: second-order techniques.

22: The capabilities of the method are demonstrated 

23: by simulations of liquid

24: water and ice, as described by a neural-network potential fitted

25: to dispersion-corrected hybrid density 

26: functional theory calculations.

27: 

28: \end{abstract}

29: