1: \begin{abstract}

2: We extend to quenched disordered systems the variational scheme for real space renormalization group calculations that we recently introduced for homogeneous spin Hamiltonians. 

3: When disorder is present our approach gives access to the flow of the renormalized Hamiltonian distribution, from which one can compute the critical exponents if the correlations of the renormalized couplings retain finite range.

4: Key to the variational approach is the bias potential found by minimizing a convex functional in statistical mechanics. 

5: This potential reduces dramatically the Monte Carlo relaxation time in large disordered systems. 

6: We demonstrate the method with applications to dilute Ising, random field Ising and short-range spin glass models, on the two-dimensional square lattice.

7: % We extend a recently proposed real-space variational Monte Carlo Renormalizaiton Group approach to treat quench-disordered systems.   

8: % This method obtains the renormalized Hamiltonian of a disorder realization via minimization of a variational functional in statistical mechanics.        

9: % In addition, we propose a scheme to compute the critical exponents from the calculated distribution of coupling constants. 

10: % Upon  minimization of the variational functional a bias potential is constructed that dramatically reduces the Monte Carlo relaxation time of disordered systems, particularly for large lattice sizes. 

11: % We demonstrate the method with applications to dilute Ising, random field Ising and short range spin glass models on the two dimensional square lattice.   

12: \end{abstract}

13: