Membrane creation and symport/antiport rules solving QSAT

In Membrane Computing, different variants of devices can be found by changing both syntactical and semantic ingredients. These devices are usually called membrane systems or P systems, and they recall the structure and behavior of living cells in the nature. In this sense, rules are introduced as a way for objects to interact with membranes, giving P systems the ability to solve computational problems. Some of these rules, as division, separation and creation rules are inspired by the membrane division through the mitosis process or new membranes are created through gemmation. These rules seem to be crucial in the path to solve computationally hard problems. In this work, creation rules are used in classical P systems with symport/antiport rules, where objects travel through membranes without changing to achieve enough computational power to efficiently solve PSPACE-complete problems. More precisely, a solution to the QSAT problem is given by means of a uniform family of these systems. This paper was originally submitted to the International Conference on Membrane Computing 2021.


Introduction
Membrane Computing, since its beginnings [1], has covered a wide spectrum of applications, from computability theory [2] and computational complexity theory [3] to pandemics [4] and engineering [5]. The model is inspired in the structure and behavior of living cells and the chemical reactions occurring within them. From the beginning, solving computationally hard problems has been a hot topic in this area [6], being one of the most studied fields with membrane systems.
Cell-like P systems can be seen as a hierarchical structure of membranes that let chemical compounds pass by them. Making an abstraction of integral membrane proteins and their role in the transport of molecules, symport/antiport rules let objects move to one region to other one or interchange them between two regions [7]. Usually, P systems are found to be Turing complete [8], but from the point of view of the computational complexity theory, they can only solve problems from the class P [9]. To increase the efficiency of these systems, different types of rules have been introduced: division rules [6], separation rules [10] and creation rules [11] are three of the protocols that have been implemented in membrane computing with this purpose. The latter uses a single object and transforms it into a new membrane.
In the framework of cell-like membrane systems, creation rules have been used lately besides evolutional communication rules in [12] to solve the SAT problem by means of a family of recognizer P systems with evolutional communication rules and creation rules. In [13], an efficient solution to QSAT is given by means of a family of recognizer polarizationless P systems with active membranes and membrane creation, and in [14], an efficient solution to QSAT is given by means of a family of P systems with evolutional symport/antiport rules of length (1, 1) and creation rules. Creation rules have not been used in P systems with classical symport/antiport rules. In this work, we try to replicate the results of the latter paper using recognizer P systems with classical symport/antiport rules with a minimal amount of objects per communication rule. This is a really important 1 3 result, since evolutional communication rules are intrinsically changing the nature of the objects, and thus both in a software simulation and in real-life implementations, it would need to implement these changes in some sense, leading to a more complex implementation. That is why the use of classical symport/antiport rules is really interesting in order to reduce the "number of ingredients" (we could think that the evolution part of evolutional symport/antiport rules is an ingredient) of the P systems using these rules.
The rest of the work is organized as follows: Section 2 is devoted to introduce some formal language and set theory concepts used later through the paper. In Sect. 3, the definition of recognizer P systems with symport/antiport rules and membrane creation is given. In the following section, a polynomial-time and uniform solution to QSAT is given by means of a family of recognizer P systems with symport/ antiport rules of length at most 1 and membrane creation, and an overview of the computations and the formal verification of the design are specified. Finally, some remarks and open research lines are indicated.

Preliminaries
Some basic notions of formal languages, set theory and other terms used throughout the paper are recalled in this section. For a deeper explanation on formal languages and membrane computing, we refer the reader to [15,16].
An alphabet Γ is a non-empty set, and its elements are called symbols. A string u over Γ is a finite sequence of symbols from Γ . The number of appearances of a symbol a in u is denoted by | u | a . The length of u, denoted by | u | is A multiset over Γ is a pair (Γ, f ) where f ∶ Γ → ℕ is a mapping from Γ to the set of natural numbers ℕ . Let m 1 = (Γ 1 , f 1 ), m 2 = (Γ 2 , f 2 ) two multisets over Γ . The union of m 1 and m 2 , denoted by m 1 + m 2 or m 1 ∪ m 2 is defined as

is the defined as
The empty multiset is denoted by ∅ , and the set of all finite multisets over Γ is denoted by M f (Γ).
The size of the set u is given by the total number of objects in u, and it is denoted by | u |.

Recognizer P systems with symport/ antiport rules and creation rules
In this section, a definition of recognizer P systems with symport/antiport rules and creation rules is given, and both the syntax and semantics are recalled.
Definition 1 A recognizer P system with symport/antiport rules and membrane creation of degree q ≥ 1 is a tuple where 1. Γ , Σ and E are finite alphabets of objects, where Σ, E ⊆ Γ, Σ ∩ E = �; 2. H is a finite set of labels; 3.
is a membrane structure whose elements are injectively labeled by elements of H; R is a set of rules of the following forms: • Symport rules: • Antiport rules: where skin is the label of the skin membrane, a ∈ Γ, u ∈ M f (Γ); 6. i in ∈ H is the label of the input membrane; 7. i out = env is the label of the output zone, in this case, the environment.
A configuration C t of a P system with symport/antiport rules and creation rules is described by the membrane structure at the moment t and the multisets of objects over Γ of each membrane, and the multiset of objects over Γ ⧵ E of the environment. We use the term region i to refer to a membrane if i ∈ H and to the environment if i = env . We can suppose that in each moment, there is an arbitrary number of objects from E in the environment. Let m be the input multiset encoding the corresponding instance of a

3
problem. The initial configuration is of such a P system Π is A symport rule (u, in) ∈ R i , called send-in rule, can be applied to a configuration C t if there exists a membrane labeled by i, and the parent region contains a multiset of objects u. When applying such a rule, the multiset of objects u is consumed from the parent region and a multiset of objects u is produced in the membrane i in the next configuration.
A symport rule (u, out) ∈ R i , called send-out rule, can be applied to a configuration C t if there exists a membrane labeled by i that contains a multiset of objects u. When applying such a rule, the multiset of objects u is consumed from the membrane i and a multiset of objects u is produced in the parent region.
An antiport rule (u, out;v, in) ∈ R i can be applied to a configuration C t if there exists a membrane labeled by i that contains a multiset of objects u, and whose parent region contains a multiset of objects v. When applying such a rule, the multisets of objects u and v are consumed from the membrane i and its parent region, respectively, and multisets v and u are produced in the membrane i and its parent region, respectively.
A creation rule [ a → [ u ] i ] j can be applied to a configuration C t if there exists a membrane labeled by j that contains an object a. When applying such a rule, object a is consumed from membrane j and a new membrane labeled by i and containing the multiset of objects u appears as a child membrane of j. A recognizer P system with symport/antiport rules and creation rules that does not send objects from the environment to the system is said to be a P system with symport/ antiport rules and creation rules without environment. In this case, the set of objects of the environment E is usually not defined in the tuple.
A transition of a P system Π is defined as a computational step of Π , passing from one configuration to the next one, and denoted by C t ⇒ Π C t+1 . A computation of a P system is a sequence of configurations such that a configuration C t+1 is always obtained from C t by applying a computation step. C 0 is the initial configuration of Π.
In [17,18], the semantics applied are maximalist in the following sense: In each membrane, an arbitrary number of creation rules can be applied, and they do not interfere with the application of other types of rules. In [12,13], more restrictive semantics were introduced. When a creation rule is applied in a membrane h, no other rules can be applied in the same computational step. In this case, dealing with P systems with symport/antiport rules and membrane creation, either communication rules in a maximal parallel way or a single creation rule can be applied in a membrane in a transition, but not both at the same time. That is, if a creation rule [ a → [ b ] ] h is applied, then neither other creation rules in that membrane h nor other symport/antiport rules from R h . In this paper, the latter semantics, called minimalist semantics are going to be used. As a recognizer membrane system, all the computations of a recognizer P system with communication rules and creation rules halt and either an object or an object (but not both) is sent to the environment at the last step of the computation.
The length of a symport rule r ≡ (u, in) or r ≡ (u, out) is given by the number of objects in multiset u; that is, it is equal to | u | . The length of an antiport rule r ≡ (u, out;v, in) is given by the total number of objects in the rule; that is, it is equal to | u | + | v | . Let us denote the length of a rule r by l(r).
The class of all recognizer P systems with symport/antiport rules and membrane creation of degree q is denoted by CCC(k) with minimalistic semantics, where k represents the maximal number of objects in a communication rule; that is, k = max(l(r) | r ∈ R i , 1 ≤ i ≤ q) . The class of recognizer membrane systems of this type when environment plays a passive role; that is, when no objects can be sent from the environment to the P system itself, is denoted by Ĉ CC(k).
All the concepts of a decision problem and the class of decision problems that can be solved by means of a uniform family of membrane systems from CCC(k) can be extracted from [12,15,19]. The class of problems that can be solved efficiently (i.e., in polynomial time with respect to the input) by means of a uniform family of recognizer P systems with symport/antiport rules of length at most k and membrane creation with environment (respectively, without environment) is denoted by PMC CCC(k) (resp., PMCĈ CC(k) ).

An efficient solution to in Ĉ CC(1)
In this section, we give an efficient solution to the problem by means of a uniform family of P systems from Ĉ CC(1) . Let t = ⟨n, p⟩ . Each P system Π(t), t ∈ ℕ, from solves all instances from with n variables and p clauses.
For each pair n, p ∈ ℕ , we consider a recognizer P system with symport/antiport rules of length 1 and creation rules that will solve all instances with n variables and p clauses, where * = ∃x 1 ∀x 2 … Q n x n (x 1 , … , x n ) is an existential fully quantified formula associated with a Boolean formula (x 1 , … , x n ) ≡ C 1 ∧ … ∧ C p in CNF, where each clause C j = l j,1 ∨ … ∨ l j,r j , Var( ) = {x 1 , … , x n } a n d l j,k ∈ {x i , ¬x i | 1 ≤ i ≤ n} . Let us suppose that the number of variables, n, and the number of clauses, p, is at least 2. We consider a polynomial encoding (cod, s) from in as follows: for each formula associated with an existential fully quantified formula * with n variables and p clauses, s( ) = ⟨n, p⟩ and cod( ) = {x i,j | x i ∈ C j } ∪ {x i,j | ¬x i ∈ C j } and s( ) = ⟨n, p⟩ 1. The working alphabet is defined as follows: 6. The set of rules R : 6.1 Rules for the counter of the elements of membrane ; let k = n 2 p + 13n + 5p + 2

Rules to return a positive answer
, out) ∈ R skin 6.3 Rules to return a negative answer

An overview of the computation
The proposed solution follows a brute force scheme of recognizer P systems with symport/antiport rules and membrane creation without environment, and it consists of the following stages:

Generation and first checking stage
By applying rules from 6.4, a membrane structure is generated. In some sense, it reminds a binary tree, but having some "garbage" membranes, labeled by # , used to generate the objects z i+1,t and z i+1,f . Besides, using rules from 6.5, objects from cod( ) will be passed throughout the membrane structure in such a way that in the level i, the i-th variable will be checked and, if the corresponding truth assignment makes true a literal in a clause j, then objects c i,j,t and c i,j,f will appear, that will be passed by the membranes up to a membrane labeled by ⟨n, r⟩ . This stage takes 2n 2 ⋅ 2p steps.

Second checking stage
Rules from 6.6 are in charge of checking whether all the clauses are satisfied in a truth assignment. For that, if there exists an object c n,j,r in a membrane labeled by ⟨n, r⟩ , it means that the corresponding truth assignment makes true the clause j. Therefore, a membrane labeled by j is created within such a membrane ⟨n, r⟩ . Object d 0 will go through all membranes, creating a "garbage" membrane within them and passing to the next one, possibly arriving to membrane p. In that case, object d p creates a new garbage membrane with an object d n,r , that will be useful in the next stage. This stage takes 3p + 5 steps.

Quantifier checking stage
If an object d i,r ( r ∈ {t, f } ) appears in a membrane, then the quantifier in this level is checked. Depending on the parity of the level, either a universal or an existential quantifier should be checked. In the generation stage, objects d ′ and d ′′ were created for this purpose. On the one hand, when an existential quantifier is being checked, an object d ′ will exist in such a membrane, and will change into an object d i−1,r if and only if there is at least one object d i,r that has created a membrane ⟨i, ∃, r, r � ⟩ . On the other hand, when a universal quantifier is to be checked, an object d ′′ will be present in such a membrane, and will change into an object d i−1,r if and only if there are two objects d i,r that have created two membranes labeled by ⟨i, ∀, r, r � ⟩ . These objects will reach the skin membrane giving way to the last stage. This whole stage is computed by the application of rules from 6.6. This stage takes 8n − 6 steps if n is even and 8n steps if n is odd.

Output stage
Counters and ′ are used in this stage to know if an object d 1,r has reached the skin membrane. In such a case, a membrane labeled by yes will be created, and when object n 2 +5n+2p+3 reaches the skin membrane, it will go into membrane yes and will change into an object that will be sent to the environment. In the case that object d 1,r does not appear in the skin membrane, object � n 2 +5n+2p+4 will generate a membrane labeled by no, that will make the counter n 2 +5n+2p+3 change into an object , and will be sent to the environment. Rules from 6.2 and 6.3 are the responsible in this stage. It takes p + 8 steps if n is even and p + 9 steps if n is odd.

Results
Next, we prove that provides a polynomial time and uniform solution to .
The family of P systems is polynomially uniform by Turing machines, polynomially bounded, sound and complete with regard to ( , cod, s) and both cod and s are polynomial-time computable functions.