1: \begin{abstract}

2: Many numerical methods for recovering ODE solutions from data rely

3: on approximating the solutions using basis functions or kernel functions

4: under a least square criterion. The accuracy of this approach hinges

5: on the smoothness of the solutions. This paper provides a theoretical

6: foundation for these methods by establishing novel results on the

7: smoothness and covering numbers of ODE solution classes (as a measure

8: of their ``size''). Our results provide answers to ``how do the

9: degree of smoothness and the ``size'' of a class of ODEs affect

10: the ``size'' of the associated class of solutions?'' In particular,

11: we show that: (1) for $y^{'}=f\left(y\right)$ and $y^{'}=f\left(x,\,y\right)$,

12: if the absolute values of all $k$th ($k\leq\beta+1$) order derivatives

13: of $f$ are bounded by $1$, then the solution can end up with the

14: $(k+1)$th derivative whose magnitude grows factorially fast in $k$

15: -- ``a curse of smoothness''; (2) our upper bounds for the covering

16: numbers of the $(\beta+2)-$degree smooth solution classes are greater

17: than those of the ``standard'' $(\beta+2)-$degree smooth class

18: of univariate functions; (3) the mean squared error of least squares

19: fitting for noisy recovery has a convergence rate no larger than $\left(\frac{1}{n}\right)^{\frac{2\left(\beta+2\right)}{2\left(\beta+2\right)+1}}$

20: if $n=\Omega\left(\left(\beta\sqrt{\log\left(\beta\vee1\right)}\right)^{4\beta+10}\right)$,

21: and under this condition, the rate $\left(\frac{1}{n}\right)^{\frac{2\left(\beta+2\right)}{2\left(\beta+2\right)+1}}$

22: is minimax optimal in the case of $y^{'}=f\left(x,\,y\right)$; (4)

23: more generally, for the higher order Picard type ODEs, $y^{\left(m\right)}=f\left(x,\,y,\,y^{'},\,...,y^{\left(m-1\right)}\right)$,

24: the covering number of the solution class is bounded from above by

25: the product of the covering number of the class $\mathcal{F}$ that

26: $f$ ranges over and the covering number of the set where initial

27: values lie.

28: \end{abstract}