1: \begin{abstract}

2: Higher order schemes for stochastic partial differential 

3: equations that do not possess 

4: commutative noise require

5: the simulation of iterated stochastic integrals. 

6: In this work, we propose a derivative-free Milstein type scheme 

7: to approximate the mild solution of stochastic partial differential equations

8: that need not to fulfill a commutativity condition for the noise term and

9: which can flexibly be combined with some

10: approximation method for the involved iterated integrals.

11: Recently, the authors introduced two algorithms 

12: to simulate such iterated stochastic integrals; these clear the way for the 

13: implementation of the proposed higher order scheme.

14: We prove the mean-square convergence of the introduced derivative-free

15: Milstein type scheme which attains the same order as the original

16: Milstein scheme. The original scheme, however, is definitely outperformed

17: when the computational cost is taken into account

18: additionally, that is, in terms of the effective order of

19: convergence. We derive the effective order of convergence 

20: for the derivative-free Milstein type scheme analytically 

21: in the case that one of the recently proposed algorithms for the approximation of 

22: the iterated stochastic integrals is applied.

23: Compared to the exponential Euler scheme and the original Milstein scheme,

24: the proposed derivative-free Milstein type scheme

25: possesses at least the same and in most cases even a higher 

26: effective order of convergence

27: depending on the particular SPDE under consideration. 

28: These analytical results are

29: illustrated and confirmed with numerical simulations.

30: %

31: \end{abstract}