1: \begin{abstract}

2: 

3: Hydrodynamic simulations of \lya clouds based on {\it ab inito}

4: cosmological models have produced results that are in broad agreement

5: with observations.  However, further analyses have revealed that, with

6: progressively higher numerical resolution, the median or cutoff line

7: width of the simulated \lya clouds (i.e. the Doppler parameter)

8: appears to converge to a value significantly below what is observed at

9: $z \sim 3$ (by about a factor of $1.5$).  These convergence

10: test simulations do not

11: include feedback from star formation. 

12: Given the observed metallicity in the Lyman alpha clouds

13: we suggest that supernovae, 

14: which presumably polluted the IGM with metals, may have

15: deposited a sufficient amount of energy in the IGM to reconcile the

16: theory with observations.  Simple arguments immediately narrow the

17: redshift range of pollution down to $4<z_{dep}<9$.  

18: It seems quite

19: certain that dwarf and sub-dwarf galaxies with total 

20: masses in the range

21: $10^{6.5-9.0}\msun$ have to be largely responsible for the pollution.

22: Furthermore, it is implied that either star formation is very

23: efficient or metal yield is very high for these early dwarf

24: galaxies, if the mean metallicity in the universe at $z=3$ is

25: as high as $<Z>=10^{-2}\zdot$.

26: Finally, assuming 

27: the specific supernova heating energy is proportional to

28: the metallicity of a gas, we note that the picture proposed here

29: would be consistent with supernovae being

30: the apparently needed heating source for intra-cluster gas 

31: if the required 

32: heating of the intra-cluster gas is no greater

33: than $1$keV per particle.

34: 

35: 

36: 

37: \end{abstract}

38: