1: \begin{abstract}

2: Fully implicit Runge-Kutta (IRK) methods have many desirable accuracy and

3: stability properties as time integration schemes, but are rarely used in

4: practice with large-scale numerical PDEs because of the difficulty of solving

5: the stage equations. This paper introduces a theoretical and algorithmic

6: framework for the fast, parallel solution of the nonlinear equations that arise

7: from IRK methods applied to nonlinear numerical PDEs, including PDEs with

8: algebraic constraints. This framework also naturally applies to discontinuous

9: Galerkin discretizations in time. Moreover, the new method is built using the

10: same preconditioners needed for backward Euler-type time stepping schemes.

11: Several new linearizations of the nonlinear IRK equations are developed,

12: offering faster and more robust convergence than the often-considered simplified

13: Newton, as well as an effective preconditioner for the true Jacobian if exact

14: Newton iterations are desired. Inverting these linearizations requires solving a

15: set of block $2\times 2$ systems. Under quite general assumptions on the spatial

16: discretization, it is proven that the preconditioned operator has a condition

17: number of $\sim\mathcal{O}(1)$, with only weak dependence on the number of

18: stages or integration accuracy. The new methods are applied to several

19: challenging fluid flow problems, including the compressible Euler and Navier

20: Stokes equations, and the vorticity-streamfunction formulation of the

21: incompressible Euler and Navier Stokes equations. Up to 10th-order accuracy is

22: demonstrated using Gauss integration, while in all cases 4th-order Gauss

23: integration requires roughly half the number of preconditioner applications as

24: required by standard SDIRK methods.

25: \end{abstract}

26: