# What does “smooth solution” of Ricci flow mean?

A way to formalize smoothness of a flow is to think of the spacetime, see e.g. here. Let's say we are flowing a compact manifold $M$. Which of the following is true?

1. $g_{ij}$ is a $C^1$ in time and $C^\infty$ in the space coordinate.

2. $g_{ij}$ is a $C^\infty$ function on the spacetime?

3. $g_{ij}$ is a $C^1$ function on the spacetime?

I think 2 is correct but then I am confused about Shi's estimates for higher derivatives. If two nearby metrics are $C^\infty$ close then all derivatives of the curvature tensor are also close, but Shi's estimates blow up when $t\to 0$. What am I missing? Update: I see the source of my confusion about Shi's estimates: the constants there depend only on the dimension and bounds on curvature tensor, but are otherwise independent of the initial metric. So this does not contradict 2.

I suppose the answer is based on a general PDE principle. What is a good reference?

Finally, will anything change if we a flowing a non-compact manifold (on which Ricci flow exists and is smooth)?