1: \begin{abstract}

2:     Neural networks can be trained to solve partial differential equations (PDEs) by using the PDE residual as the loss function.

3:     This strategy is called ``physics-informed neural networks'' (PINNs), but it currently cannot produce high-accuracy solutions, typically attaining about $0.1\%$ relative error.

4:     We present an adversarial approach that overcomes this limitation, which we call competitive PINNs (CPINNs).

5:     CPINNs train a discriminator that is rewarded for predicting mistakes the PINN makes.

6:     The discriminator and PINN participate in a zero-sum game with the exact PDE solution as an optimal strategy.

7:     This approach avoids squaring the large condition numbers of PDE discretizations, which is the likely reason for failures of previous attempts to decrease PINN errors even on benign problems.

8:     Numerical experiments on a Poisson problem show that CPINNs achieve errors four orders of magnitude smaller than the best-performing PINN.

9:     We observe relative errors on the order of single-precision accuracy, consistently decreasing with each epoch. 

10:     To the authors' knowledge, this is the first time this level of accuracy and convergence behavior has been achieved. 

11:     Additional experiments on the nonlinear Schr{\"o}dinger, Burgers', and Allen--Cahn equation show that the benefits of CPINNs are not limited to linear problems.

12: \end{abstract}

13: