コメントに収まらない観察結果をいくつかご紹介します。
最初の回答が削除されたため、0)を添加:存在の解釈であり、Hnによって行と列を索引付け、すなわち、{0,1}n、対応するエントリ(x,y)である1かのアダマール積x⊙y=(x1y1,…,xnyn)は偶数パリティーを持ち、奇数パリティーを持つ場合は−1です。
1)一般に、ブロック行列のスペクトルは非常に複雑になる可能性があり、特性多項式がひどく見えるため、個々のブロックのスペクトルとは明らかに関係しません。しかし、対称ブロック行列のためのM=(ABTBC)のような再帰的な構造を介して発生する可能性のあるAn及びHn、各行列は正方形であり、上記だけ単純化のいずれかが発生BT及びC、通勤この場合、det(M)=det(AC−BBT)。その後の特性多項式Mなりdet((λI−A)(λI−C)−BBT)=det(λ2I−λ(A+C)+AC−BBT).
これが固有値の素敵な再帰式を導くためには、基本的に Cが必要ですC=−A線形λ項を削除する A。さらに場合はAとB対称と通勤している、我々は、GET
det(λI−M)=det(λ2I−(A2+B2)),
一方は容易対称通勤行列が認める事実を使用して固有値オフ読み出し、そこから一般的な固有基底。これは明らかかもしれませんが、これはすべて、固有値の適切な再帰式を得る限り、右下のブロックが−A and hope that lower left and upper right blocks are symmetric and commute with A, which is the case for the An (with B=I) and Hn matrices (with B=Hn−1=A).
2) On the random sign question: the signing of the adjacency matrix given in the paper is optimal in the sense of maximizing λ2n−1, which is needed for the lower bound via Cauchy interlacing, and can be seen from elementary means. For an arbitrary signing Mn of the adjacency matrix of the n-dimensional hypercube, one immediately gets
Tr(Mn)=∑i=12nλi(Mn)=0,Tr(M2n)=∑i=12nλi(Mn)2=∥Mn∥2F=n2n,
where λ1(Mn)≥λ2(Mn)≥…≥λ2n(Mn). If for some signing Mn one has λ2n−1(Mn)>n−−√, then
∑i=12n−1λi(Mn)>n−−√2n−1,∑i=12n−1λi(Mn)2>n2n−1.
One can then see it is not possible to satisfy the trace equalities above: the negative eigenvalues must sum to strictly more than n−−√2n−1 (in absolute value), and their squares must sum to strictly less than n2n−1. Minimizing the sum of squares while keeping the sum constant happens when they are all equal, but in this case will make the sum of squares too large anyway. So for any signing, one can see via elementary means that λ2n−1(Mn)≤n−−√ without knowing the magic signing in the paper, where equality holds iff the values are n−−√,…,n−−√,−n−−√,…,−n−−√. That there actually exists such a signing attaining it is pretty amazing. The eigenvalues of the normal adjacency matrix are −n,−n+2,…,n−2,n, where the ith eigenvalue has multiplicity (ni), so it's very interesting (to me, anyway) how the all-+1 signing maximizes λ1, while this signing maximizes λ2n−1.
As far as would a random signing work, it's harder to say because I think most non-asymptotic bounds on eigenvalues focus on spectral norm. One expects random signings to smooth out the extreme usual eigenvalues, and indeed, using the noncommutative Khintchine inequality and/or recent tighter bounds like in here, a uniformly random signing has E[∥Mn∥2]=Θ(n−−√). It's hard for me to imagine the middle eigenvalues would be on a similar polynomial order as the leading one in expectation (and asymptotic results like the semi-circular law for different matrix ensembles suggest similarly, I think), but maybe it's possible.