09.09.2014 Views

13th International Conference on Membrane Computing - MTA Sztaki

13th International Conference on Membrane Computing - MTA Sztaki

13th International Conference on Membrane Computing - MTA Sztaki

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

R. Pagliarini, O. Agrigoroaiei, G. Ciobanu, V. Manca<br />

The underlining idea for this definiti<strong>on</strong> is that when some (multiset of) objects<br />

v appear during the course of an evoluti<strong>on</strong> of a P system, we look for some<br />

(multiset of) rules G which have produced them. By producing we understand<br />

that we have an evoluti<strong>on</strong> step u =⇒ F<br />

u ′ in which u ′ ≥ v and F ≥ G such<br />

that exactly the rules in G are those resp<strong>on</strong>sible for the appariti<strong>on</strong> of v. Note<br />

that v can be written as the sum of v ∩ rhs(G) and v\rhs(G). The v ∩ rhs(G)<br />

part is the <strong>on</strong>e produced by G since it is included in rhs(G); for the remainder<br />

v\rhs(G) we require that it is composed <strong>on</strong>ly of objects which do not evolve in<br />

the c<strong>on</strong>sidered evoluti<strong>on</strong> step (this is what the first c<strong>on</strong>diti<strong>on</strong> amounts to). In<br />

other words, all the objects of v are either produced by rules of G or are not<br />

interacting with any rule. The sec<strong>on</strong>d c<strong>on</strong>diti<strong>on</strong> is equivalent to saying that no<br />

rule r can be subtracted from G such that the part of v produced by G remains<br />

the same - there are no “useless” rules in G with respect to producing elements<br />

of v.<br />

To view the noti<strong>on</strong>s above introduced, let us c<strong>on</strong>sider the following example<br />

of a transiti<strong>on</strong> P system with <strong>on</strong>ly <strong>on</strong>e membrane, with rules<br />

r 1 : x → a + b, r 2 : y → b, r 3 : a + b → y<br />

is<br />

and an initial multiset of objects u 0 = x+y +2a. The <strong>on</strong>ly possible evoluti<strong>on</strong><br />

x + y + 2a r1+r2<br />

=⇒ 3a + 2b =⇒ 2r3<br />

a + 2y =⇒ 2r2<br />

a + 2b =⇒ r3<br />

b + y =⇒ r2<br />

2b<br />

In [2], a general inductive procedure for finding the causes of a multiset has<br />

been introduced. Here we reas<strong>on</strong> directly over the example c<strong>on</strong>sidered, loosely<br />

following the inductive procedure.<br />

Let v = a + b be the multiset for which we intend to find its causes. We<br />

start by c<strong>on</strong>sidering the empty multiset 0 as a potential cause for v. The empty<br />

multiset is discarded because lhs(r 3 ) ≤ v\rhs(0) = v\0 (they are actually equal)<br />

which c<strong>on</strong>tradicts the first part of Definiti<strong>on</strong> 1. The next possible candidates for<br />

causes of v are either r 1 or r 2 or r 3 . Clearly r 3 suffers from the same problem as<br />

0, it does not fulfill the first c<strong>on</strong>diti<strong>on</strong> of Definiti<strong>on</strong> 1. However, both r 1 and r 2<br />

verify the c<strong>on</strong>diti<strong>on</strong>s to be causes of v. The next step is to add rule r 3 to either<br />

multiset r 1 or r 2 , i.e., we c<strong>on</strong>sider as potential causes r 1 +r 3 and r 2 +r 3 . We find<br />

that rule r 3 is actually “useless” as it does not produce any object of v. In other<br />

words, r 3 has the problem that rhs(G) ∩ v = rhs(G − r 3 ) ∩ v for G = r 1 + r 3<br />

or G = r 2 + r 3 . Moreover, this happens for any G ≥ r 1 + r 3 or G ≥ r 2 + r 3 .<br />

This means that no cause of v can c<strong>on</strong>tain r 3 . If we try G = r 1 + r 2 , then r 2<br />

becomes the “useless” rule: rhs(G) ∩ v = rhs(G − r 2 ) ∩ v. This also happens<br />

for G ≥ r 1 + r 2 . All we have left to check are either G = k · r 1 or G = k · r 2 ,<br />

for k ≥ 2. When G = k · r 1 we have that an instance of r 1 is a “useless” rule:<br />

rhs(G)∩v = rhs(G−r 1 )∩v; in other words, any additi<strong>on</strong>al r 1 besides a single r 1<br />

are “useless”. The case of G = k · r 2 for k ≥ 2 is similar. Thus the <strong>on</strong>ly possible<br />

causes for v = a + b are either r 1 or r 2 .<br />

354

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!