1: \begin{abstract}\def\GF(#1){\mathbb F_{#1}}%

2: 	Polarization is an unprecedented coding technique in that it not only

3: 	achieves channel capacity, but also does so at a faster speed of convergence

4: 	than any other technique.  This speed is measured by the ``scaling

5: 	exponent'' and its importance is three-fold.  Firstly, estimating the

6: 	scaling exponent is challenging and demands a deeper understanding of the

7: 	dynamics of communication channels.  Secondly, scaling exponents serve as a

8: 	benchmark for different variants of polar codes that helps us select the

9: 	proper variant for real-life applications.  Thirdly, the need to optimize

10: 	for the scaling exponent sheds light on how to reinforce the design of polar

11: 	code.

12: 	

13: 	\smallskip

14: 	

15: 	In this paper, we generalize the binary erasure channel (BEC), the simplest

16: 	communication channel and the protagonist of many polar code studies, to the

17: 	``tetrahedral erasure channel'' (TEC).  We then invoke Mori--Tanaka's $2

18: 	\times 2$ matrix over $\GF(4)$ to construct polar codes over TEC.  Our main

19: 	contribution is showing that the dynamic of TECs converges to an

20: 	almost--one-parameter family of channels, which then leads to an upper bound

21: 	of $3.328$ on the scaling exponent.  This is the first non-binary matrix

22: 	whose scaling exponent is upper-bounded.  It also polarizes BEC faster than

23: 	all known binary matrices up to $23 \times 23$ in size.  Our result

24: 	indicates that expanding the alphabet is a more effective and practical

25: 	alternative to enlarging the matrix in order to achieve faster polarization.

26: \end{abstract}

27: