English

Almost-catalytic Computation

Computational Complexity 2024-11-25 v2

Abstract

Designing algorithms for space bounded models with restoration requirements on the space used by the algorithm is an important challenge posed about the catalytic computation model introduced by Buhrman et al. (2014). Motivated by the scenarios where we do not need to restore unless is useful, we define ACL(A)ACL(A) to be the class of languages that can be accepted by almost-catalytic Turing machines with respect to AA (which we call the catalytic set), that uses at most clognc\log n work space and ncn^c catalytic space. We show that if there are almost-catalytic algorithms for a problem with catalytic set as AΣA \subseteq \Sigma^* and its complement respectively, then the problem can be solved by a ZPP algorithm. Using this, we derive that to design catalytic algorithms, it suffices to design almost-catalytic algorithms where the catalytic set is the set of strings of odd weight (PARITYPARITY). Towards this, we consider two complexity measures of the set AA which are maximized for PARITYPARITY - random projection complexity (R(A){\cal R}(A)) and the subcube partition complexity (P(A){\cal P}(A)). By making use of error-correcting codes, we show that for all k1k \ge 1, there is a language AkΣA_k \subseteq \Sigma^* such that DSPACE(nk)ACL(Ak)DSPACE(n^k) \subseteq ACL(A_k) where for every m1m \ge 1, R(Ak{0,1}m)m4\mathcal{R}(A_k \cap \{0,1\}^m) \ge \frac{m}{4} and P(Ak{0,1}m)=2m/4\mathcal{P}(A_k \cap \{0,1\}^m)=2^{m/4}. This contrasts the catalytic machine model where it is unclear if it can accept all languages in DSPACE(log1+ϵn)DSPACE(\log^{1+\epsilon} n) for any ϵ>0\epsilon > 0. Improving the partition complexity of the catalytic set AA further, we show that for all k1k \ge 1, there is a Ak{0,1}A_k \subseteq \{0,1\}^* such that DSPACE(logkn)ACL(Ak)\mathsf{DSPACE}(\log^k n) \subseteq ACL(A_k) where for every m1m \ge 1, R(Ak{0,1}m)m4\mathcal{R}(A_k \cap \{0,1\}^m) \ge \frac{m}{4} and P(Ak{0,1}m)=2m/4+Ω(logm)\mathcal{P}(A_k \cap \{0,1\}^m)=2^{m/4+\Omega(\log m)}.

Keywords

Cite

@article{arxiv.2409.07208,
  title  = {Almost-catalytic Computation},
  author = {Sagar Bisoyi and Krishnamoorthy Dinesh and Bhabya Deep Rai and Jayalal Sarma},
  journal= {arXiv preprint arXiv:2409.07208},
  year   = {2024}
}

Comments

22 pages, A new lower bound on the subcube partition complexity of Hamming balls (Proposition 2.6 and Lemma 2.7), improving the bound and fixing an error in the previous version

R2 v1 2026-06-28T18:41:01.876Z