1: \begin{abstract}

2: 	We consider robust pricing and hedging for options written on multiple 

3: 	assets given market option prices for the individual assets. The resulting 

4: 	problem is called the multi-marginal martingale optimal transport problem. 

5: 	We propose two numerical methods to solve such problems: using 

6: 	discretisation and linear programming applied to the primal side and using 

7: 	penalisation and deep neural networks optimisation applied to the dual 

8: 	side. We prove convergence for our methods and compare their numerical 

9: 	performance. We show how adding further information about call option 

10: 	prices at additional maturities can be incorporated and narrows down the 

11: 	no-arbitrage pricing bounds. Finally, we obtain structural results for the 

12: 	case of the payoff given by a weighted sum of covariances between the 

13: 	assets. 

14: \end{abstract}

15: