1: \begin{abstract} 

2: In this paper, we consider the restriction of finite dimensional

3: $GL_{mn} (\C)$-modules to the subgroup ${GL_m (\C )\times GL_n (\C)}$.

4: In particular, for a Weyl module $V_{\lambda } (\C^{mn})$ of $U_q

5: (gl_{mn})$ we ask to construct

6: a representation $W_{\lambda }$ of $U_q (gl_m )\otimes U_q (gl_n )$

7: such that at $q=1$, the restriction of $V_{\lambda } (\C^{mn})$ to 

8: $U_1 (gl_m )\otimes U_1 (gl_n )$ matches its action on $W_{\lambda }$ at

9: $q=1$. 

10: Thus $W_{\lambda }$ is a $q$-deformation of the module $V_{\lambda }$. 

11: We achieve this for (i) $\lambda $ consisting of upto two columns and 

12: $\lambda =(k)$ (i.e., the $Sym^k $ case) for general $m,n$, and 

13: (ii) all $\lambda $'s for $m=n=2$. This is 

14: achieved by first constructing a \eat{(commuting)} $U_q (gl_m )\otimes 

15: U_q (gl_n )$-module $\wedge^k $, a $q$-deformation 

16: of the simple $GL_{mn} (\C)$-module $\wedge^k (\C^{mn})$. 

17: We also construct the bi-crystal basis for (i) $\wedge^k $ and show that

18: it consists of signed subsets, and for (ii) $Sym^k$ and show that it consist

19: of unordered monomials. 

20: Next, we develop $U_q (gl_m ) \otimes U_q (gl_n )$-equivariant 

21: maps $\psi_{a,b} :\wedge^{a+1} 

22: \otimes \wedge^{b-1} \rightarrow \wedge^a \otimes

23: \wedge^b$. 

24: This is used as the building block to construct the general

25: $W_{\lambda }$ for the cases listed above. 

26: 

27: %\noindent{\em Keywords:} Computational complexity, quantum groups.

28: \end{abstract}