1: \begin{abstract}

2: 

3: By a new type of finite size scaling analysis on the square lattice, and by

4: renormalization group calculations on hierarchical lattices we investigate the

5: effects of dilution on optimal undirected

6: self-avoiding paths in a random environment. The behaviour of the

7: optimal paths

8: remains the same as for directed paths

9: in undiluted medium, as long as forbidden bonds

10: are not exceeding the percolation threshold.

11: Thus, overhanging configurations do not alter

12: the standard self-affine directed polymer scaling  regime, even above

13: the directed threshold, when they become unavoidable.

14: 

15: When dilution reaches the undirected threshold, the

16: optimal path becomes fractal, with fractal dimension

17: equal to $D_{\rm min}$, the dimension of the minimal

18: length path on percolation cluster backbone.

19: 

20: In this regime the optimal path energy fluctuation, $\overline{\Delta E}$,

21: can be ascribed entirely to minimal length

22: fluctuations, and satisfies $\overline{\Delta E} \propto L^{\omega}$,

23: with $\omega=1.02 \pm 0.06$ in $2d$, $L$ being the Euclidean

24: distance. Hierarchical lattice calculations confirm that $\omega$ is

25: also the exponent of the leading scaling correction to

26: $\overline E \propto L^{D_{\rm min}}$.

27:  Upon approaching threshold, the

28: probability, ${\cal R}$, that the optimal

29: path does not stick entirely on the

30: minimal length one, obeys ${\cal R} \sim \left( \Delta p\right) ^{\rho}$,

31: with $\rho \sim 1.0 \pm 0.05$ on hierarchical lattices. Such behaviour

32: could be characteristic of the crossover to fractal regime.

33: Transfer matrix results on square lattice show that a similar full

34: sticking does not

35: occur for directed paths at the directed percolation threshold.

36: 

37: \end{abstract}

38: