1: \begin{abstract}

2: 

3: We propose a general strategy for determining the minimal finite

4: amplitude disturbance to trigger transition to turbulence in shear

5: flows. This involves constructing a variational problem that searches

6: over all disturbances of fixed initial amplitude, which respect

7: the boundary conditions, incompressibility and the Navier--Stokes

8: equations, to maximise a chosen functional over an

9: asymptotically long time period.

10: The functional must be selected such that it

11: identifies turbulent velocity fields by taking significantly 

12: enhanced values compared to those for laminar fields. 

13: We illustrate this approach

14: using the ratio of the final to initial perturbation kinetic energies

15: (energy growth) as the functional and the energy norm to measure

16: amplitudes in the context of pipe flow. Our results indicate that the

17: variational problem yields a smooth converged solution providing the

18: amplitude is below the threshold amplitude for transition.  

19: This optimal is the nonlinear analogue of the well-studied (linear)

20: transient growth optimal. At and above this threshold, the optimising

21: search naturally seeks out disturbances that trigger turbulence by

22: the end of the period, and convergence is then practically impossible.

23: The first disturbance found to trigger turbulence as the amplitude is

24: increased identifies the `minimal seed' for the given geometry and

25: forcing (Reynolds number). We conjecture that it may be possible to

26: select a functional such that the converged optimal below threshold

27: smoothly converges to the minimal seed at threshold.  This seems at

28: least approximately true for our choice of energy growth functional

29: and the pipe flow geometry chosen here.

30: 

31: \end{abstract}