| Abstract [eng] |
This work is based on an article by M. Juodis and A. Račkauskas concerning self-normalized sums. The object of this thesis is to further analyse a proven theorem about the convergance of self-normalized sums for dependant random variables using blocks. Firstly, we measure the rate of convergance as a function dependant on the total number of elements and the lenght of the blocks. Later, we find the extreme values of this function and give an estimate of the optimal block length for best convergance results. Finaly, we measure the rate of convergance using the method described in the article. |
