根据Kaveh的建议,我将我的评论作为(扩展的)答案。
关于Q1,请注意以下几点:即使距离甚远,即使是对数深度,也不是在讲多对数。因此,在非单调的世界中,真正的问题远没有那么雄心勃勃:
超越对数深度问题:证明电路的超线性(!)下界。
NC1
即使对于线性 电路,问题仍然存在(目前已超过30年)。这些是基于{ ⊕ ,1 }的 fanin- 2电路,并且它们在G F (2 )上计算线性变换f (x )= A x。简单的计数表明,几乎所有矩阵A在任何深度都需要
Ω (n 2 / log n )门。
NC12{⊕,1}f(x)=AxGF(2)AΩ(n2/logn)
关于Q2:是的,我们有
一些代数/组合测度,下界将超过对数深度电路。不幸的是,到目前为止,我们无法证明这些措施的范围足够大。例如,对于线性电路,这种度量是矩阵A的刚度R A(r )。这是为了将等级降低到r而需要更改的A条目的最小数目。这是很容易显示其ř 甲([R )≤ (ñ -NC1 RA(r)AAr对于每个布尔 n × n矩阵 A都成立,Valiant(1977)表明,该边界对于几乎所有矩阵都是严格的。为了击败对数深度电路,足以表现出布尔 n × n矩阵 A的序列,使得RA(r)≤(n−r)2n×nAn×nA
为常数 ε ,δ > 0。
RA(ϵn)≥n1+δϵ,δ>0
迄今为止我们所知道的最好的是矩阵与ř 甲([R )≥ (Ñ 2 / [R )日志(ñ / [R )。 对于Sylvester矩阵(即内积矩阵),很容易显示Ω (n 2 / r )的下限。
ARA(r)≥(n2/r)log(n/r)Ω(n2/r)
我们为一般(非线性)组合措施 -circuits,以及对于二分Ñ × Ñ
图表G ^,让吨(ģ )是最小的数吨,使得ģ可以写成的交点吨二分图,每个图最多是t个完整的二部图的并集。要击败一般的对数深度电路,只要找到具有NC1n×nGt(G)tGtt
为常数 ε > 0t(Gn)≥nϵϵ>0
(例如,请参见此处有关如何发生的信息)。同样,几乎所有的图都
。然而,最好的遗体的下界吨(ģ )≥ 登录3 Ñ为Sylvester矩阵,由于Lokam。
t(G)≥n1/2t(G)≥log3n
最后,让我提到,我们甚至有一个“简单的”组合度量(数量),一个弱(线性)下界,对于非单调电路,它甚至会产生指数(!)下界。对于二分图图G,令c (G )为从恒星开始产生G所需的fanin- 2联合(∪)和交点(∩)运算的最小数目。星形是将一个顶点与所有顶点连接在另一侧的一组边。几乎所有图形的c (G )= Ω (n 2n×nGc(G)2∪∩Gc(G)=Ω(n2/logn)
c(Gn)≥(4+ϵ)nϵ>0
Ω(2N/2)fGNGn×mm=o(n)c(Gn)≥(2+ϵ)n is enough (again, see, e.g. here on how this happens). Lower bounds c(G)≥(2−ϵ)n can be shown for relatively simple graphs. The problem, however, is to do this with "−ϵ" replaced by "+ϵ". More combinatorial measures lower-bounding circuit complexity (including the ACC-circuits)
can be found in the
book.
P.S. So,
are we by a constant factor of 2+ϵ from showing P≠NP?
Of course - not.
I mentioned this latter measure c(G) only to show that one should treat "amplification" (or "magnification") of lower bounds with a healthy portion of skepticism: even though the bounds we need look "innocently", are much smaller (linear) than almost all graphs require (quadratic), the inherent difficulty of proving a (weak) lower bound may be even bigger. Of course, having found a combinatorial measure, we can say something about what properties of functions make them computationally hard. This may be useful for proving an indirect lower bound: some complexity class contains a function requiring large circuits or formulas. But the ultimate goal is to come up with an explicit hard function, whose definition does not have an "algorithmic smell", does not have any hidden complexity aspects.