1: \begin{abstract}

2: The Weak Chebyshev Greedy Algorithm (WCGA) is defined for any Banach space $X$ and 

3: a dictionary $\D$, and provides nonlinear $n$-term approximation 

4: for a given element $f \in X$ with respect to $\D$.

5: In this paper we study the generalized  Approximate Weak Chebyshev Greedy 

6: Algorithm (gAWCGA) --- a modification of the WCGA in which we are 

7: allowed to calculate $n$-term approximation with relative and 

8: absolute errors in computing a norming functional, 

9: an element of best approximation, and an approximant. 

10: Such permission is natural for the numerical 

11: applications and simplifies realization of the algorithm.

12: We obtain conditions that are sufficient for the convergence of the 

13: gAWCGA for any element of a uniformly smooth Banach space, 

14: and show that they are necessary in the class of uniformly smooth 

15: Banach spaces with the modulus of smoothness of non-trivial power 

16: type (e.g. $L_p$ spaces for $1 < p < \infty$).

17: In particular, we show that if all the errors are from $\ell_1$ 

18: space then the conditions for the convergence of the gAWCGA 

19: are the same as for the WCGA.

20: We also construct an example of a smooth Banach space

21: in which the algorithm diverges for a dictionary and an element,

22: thus showing that the smoothness of a space is not

23: sufficient for the convergence of the WCGA.

24: \end{abstract}

25: