1: \begin{abstract}

2: The successive perturbative estimates of the

3: pressure of QCD at high temperature $T$  show no sign of 

4: convergence,  unless the coupling constant $g$ is unrealistically small.

5: Exploiting known results of an

6: effective field theory which separates hard (order $2\pi T$) and soft (order

7: $gT$) contributions, we explore the accuracy of simple resummations which

8: at a given loop order systematically treat hard contributions

9: strictly perturbatively, but soft contributions without truncations.

10: This turns out to improve significantly the two-loop and the three-loop results

11: in that both remain below the ideal-gas value, and the

12: degree of renormalization

13: scale dependence decreases as one goes from two to three loop order,

14: whereas it increases in the conventional perturbative results.

15: Including the four-loop logarithms recently obtained by Kajantie et al.,

16: we find that this trend continues and that with a particular 

17: sublogarithmic constant the untruncated four-loop

18: result is close to the three-loop result, which itself agrees

19: well with available lattice results down to

20: temperatures of about $2.5T_c$.

21: We also investigate the possibility of optimization

22: by using a variational (``screened'')

23: perturbation theory in the effective  theory.

24: At two loops, this gives a result below the

25: ideal gas value, and also closer to

26: lattice results than the recent two-loop hard-thermal-loop-screened result

27: of Andersen et al.

28: While at three-loop order the gap equation of

29: dimensionally reduced screened perturbation

30: theory does not have a solution in QCD, this is

31: remedied upon inclusion of the four-loop logarithms.

32: \end{abstract}

33: