1: \begin{abstract}

2: We consider the Gumbel or extreme value statistics 

3: describing the distribution function 

4: $p_{\rm G}(\nu_{\rm max})$ of the maximum values of a random 

5: field $\nu$ within patches of fixed

6: size. We present, for smooth Gaussian random fields in two and 

7: three dimensions, an analytical estimate of $p_{\rm G}$ 

8: which is expected to hold in a regime where local maxima 

9: of the field are moderately high and weakly clustered.

10: 

11: When the patch

12: size becomes sufficiently large, the negative of the logarithm of the cumulative 

13: extreme value distribution is simply equal to the average of the Euler Characteristic

14: of the field in the excursion $\nu \ge \nu_{\rm max}$ inside the

15: patches. The Gumbel statistics therefore represents an interesting alternative probe

16: of the genus as a test of non Gaussianity, e.g. in cosmic

17: microwave background temperature maps or in three-dimensional galaxy

18: catalogs. It can be approximated, except in the remote positive tail, by a negative

19: Weibull type form, converging slowly to the expected Gumbel type

20: form for infinitely large patch size. Convergence is facilitated

21: when large scale correlations are weaker.

22: 

23: We compare the analytic predictions to numerical experiments for the case

24: of a scale-free Gaussian field in two dimensions, achieving impressive

25: agreement between approximate theory and measurements. 

26: We also discuss the generalization of our formalism to non-Gaussian

27: fields.

28:  

29: \end{abstract}

30: