1: \begin{abstract}

2: The uniform electron gas (UEG) at finite temperature is of key relevance for many applications in the warm dense matter regime, e.g.\ dense plasmas and laser excited solids. Also, the quality of density functional theory calculations crucially relies on the availability of accurate data for the exchange-correlation energy. Recently, new benchmark results for the $N=33$ spin-polarized electrons at high density, $r_s={\bar r}/a_B \lesssim 4$ and low temperature, have been obtained with the configuration path integral Monte Carlo (CPIMC) method [T. Schoof \textit{et al.}, Phys. Rev. Lett. \textbf{115}, 130402 (2015)].

3: %which exhibited a significant deviation from existing restricted path integral Monte Carlo data [E. W. Brown \textit{et al.}, PRL \textbf{110}, 146405 (2013)].

4: To achieve these results, the original CPIMC algorithm [T. Schoof \textit{et al.}, Contrib. Plasma Phys. \textbf{51}, 687 (2011)] had to be further optimized to cope with the fermion sign problem (FSP). It is the purpose of this paper to give detailed information on the manifestation of the FSP in CPIMC simulations of the UEG and to demonstrate how it can be turned into a controllable convergence problem. 

5: In addition, we present new thermodynamic results for higher temperatures. Finally, to overcome the limitations of CPIMC towards strong coupling, we invoke an independent 

6: method---the recently developed permutation blocking path integral Monte Carlo approach [T. Dornheim \textit{et al.}, accepted for publication in J. Chem Phys., arXiv:1508.03221]. 

7: The combination of both approaches is able to yield {\em ab initio} data for the UEG over the entire density range, above a temperature of about one half of the Fermi temperature. Comparison with restricted path integral Monte Carlo data [E. W. Brown \textit{et al.}, Phys. Rev. Lett. \textbf{110}, 146405 (2013)] allows us to quantify the systematic error arising from the free particle nodes.  

8: \end{abstract}

9: