Weizmann Logo
ECCC
Electronic Colloquium on Computational Complexity

Under the auspices of the Computational Complexity Foundation (CCF)

Login | Register | Classic Style



REPORTS > DETAIL:

Revision(s):

Revision #2 to TR20-116 | 19th May 2021 06:55

Toward better depth lower bounds: the XOR-KRW conjecture

RSS-Feed




Revision #2
Authors: Ivan Mihajlin, Alexander Smal
Accepted on: 19th May 2021 06:55
Downloads: 332
Keywords: 


Abstract:

In this paper, we propose a new conjecture, the XOR-KRW conjecture, which is a relaxation of the Karchmer-Raz-Wigderson conjecture [KRW95]. This relaxation is still strong enough to imply $\mathbf{P} \not\subseteq \mathbf{NC}^1$ if proven. We also present a weaker version of this conjecture that might be used for breaking $n^3$ lower bound for De~Morgan formulas. Our study of this conjecture allows us to partially answer an open question stated in [GMWW17] regarding the composition of the universal relation with a function. To be more precise, we prove that there exists a function $g$ such that the composition of the universal relation with $g$ is significantly harder than just a universal relation. The fact that we can only prove the existence of $g$ is an inherent feature of our approach.

The paper's main technical contribution is a method of converting lower bounds for multiplexer-type relations into lower bounds against functions. In order to do this, we develop techniques to lower bound communication complexity using reductions from non-deterministic communication complexity and non-classical models: half-duplex and partially half-duplex communication models.



Changes to previous version:

Fix in the proof of the main theorem + lots of minor improvements


Revision #1 to TR20-116 | 25th November 2020 12:46

Toward better depth lower bounds: the XOR-KRW conjecture





Revision #1
Authors: Ivan Mihajlin, Alexander Smal
Accepted on: 25th November 2020 12:46
Downloads: 328
Keywords: 


Abstract:

In this paper, we propose a new conjecture, the XOR-KRW conjecture, which is a relaxation of the Karchmer-Raz-Wigderson conjecture [KRW95]. This relaxation is still strong enough to imply $\mathbf{P} \not\subseteq \mathbf{NC}^1$ if proven. We also present a weaker version of this conjecture that might be used for breaking $n^3$ lower bound for De~Morgan formulas. Our study of this conjecture allows us to partially answer an open question stated in [GMWW17] regarding the composition of the universal relation with a function. To be more precise, we prove that there exists a function $g$ such that the composition of the universal relation with $g$ is significantly harder than just a universal relation. The fact that we can only prove the existence of $g$ is an inherent feature of our approach.

The paper's main technical contribution is a method of converting lower bounds for multiplexer-type relations into lower bounds against functions. In order to do this, we develop techniques to lower bound communication complexity using reductions from non-deterministic communication complexity and non-classical models: half-duplex and partially half-duplex communication models.



Changes to previous version:

Multiple typos fixed.


Paper:

TR20-116 | 1st August 2020 01:05

Toward better depth lower bounds: the XOR-KRW conjecture


Abstract:

In this paper, we propose a new conjecture, the XOR-KRW conjecture, which is a relaxation of the Karchmer-Raz-Wigderson conjecture [KRW95]. This relaxation is still strong enough to imply $\mathbf{P} \not\subseteq \mathbf{NC}^1$ if proven. We also present a weaker version of this conjecture that might be used for breaking $n^3$ lower bound for De~Morgan formulas. Our study of this conjecture allows us to partially answer an open question stated in [GMWW17] regarding the composition of the universal relation with a function. To be more precise, we prove that there exists a function $g$ such that the composition of the universal relation with $g$ is significantly harder than just a universal relation. The fact that we can only prove the existence of $g$ is an inherent feature of our approach.

The paper's main technical contribution is a method of converting lower bounds for multiplexer-type relations into lower bounds against functions. In order to do this, we develop techniques to lower bound communication complexity using reductions from non-deterministic communication complexity and non-classical models: half-duplex and partially half-duplex communication models.



ISSN 1433-8092 | Imprint