1: \begin{abstract}

2: 	We consider the one-dimensional squared Bessel process given by the stochastic differential

3: 	equation (SDE)

4: 	\begin{align*}

5: 		dX_t = 1\,dt + 2\sqrt{X_t}\,dW_t, \quad X_0=x_0, \quad t\in[0,1],

6: 	\end{align*}

7: 	and study strong (pathwise) approximation of the solution $X$ at the final time point $t=1$.

8: 	This SDE is a particular instance of a Cox-Ingersoll-Ross (CIR) process where the boundary point

9: 	zero is accessible.

10: 	We consider numerical methods that have access to values of the driving Brownian motion $W$

11: 	at a finite number of time points. We show that the polynomial convergence rate

12: 	of the $n$-th minimal errors for the class of adaptive algorithms as well as

13: 	for the class of algorithms that rely on equidistant grids are equal to

14: 	infinity and $1/2$, respectively.

15: 	This shows that adaption results in a tremendously improved convergence rate.

16: 	As a by-product, we obtain that the parameters appearing in the CIR process affect

17: 	the convergence rate of strong approximation.

18: \end{abstract}

19: