1: \begin{abstract}

2: Kazhdan constants of discrete groups are hard to 

3: compute and the actual constants are known only for several 

4: classes of groups. 

5: 

6: By solving a semidefinite programming problem by a computer,  we obtain a lower bound of 

7: the Kazhdan constant of a discrete group. 

8: Positive lower bounds imply that the group has property (T).

9: 

10: We study lattices on $\tilde A_2$-buildings in detail.

11: For $\tilde A_2$-groups, our numerical bounds look identical 

12: to the known actual constants. That suggests that our approach is 

13: effective. 

14: 

15: For a family of groups, $G_1, \cdots, G_4$, that are studied

16: by Ronan, Tits and others, we conjecture the spectral gap

17: of the Laplacian is $(\sqrt 2-1)^2$ based on our experimental results. 

18: 

19: For $\mathrm{SL}(3,\Bbb Z)$ and $\mathrm{SL}(4,\Bbb Z)$ we obtain lower bounds of the Kazhdan constants,

20: 0.2155 and 0.3285, respectively, which

21: are better than any other known bounds. 

22: We also obtain 0.1710 as a lower bound 

23: of the Kazhdan constant of the Steinberg group

24: $\St_3(\Bbb Z)$. 

25: 

26: \end{abstract}

27: