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.

S. Verlan, J. Quiros<br />

digital FPGA circuit synchr<strong>on</strong>ized by a global clock signal, in <strong>on</strong>e cycle of FPGA<br />

it is possible to compute any functi<strong>on</strong>s whose implementati<strong>on</strong> has a delay which<br />

does not exceed the period of the global clock signal. A pipeline using arithmetical<br />

operati<strong>on</strong>s and, in general, any combinatorial and sequential asynchr<strong>on</strong>ous<br />

subsystems, are usually included in this group.<br />

In order to simplify the problem we split it into two parts corresp<strong>on</strong>ding to<br />

the c<strong>on</strong>structi<strong>on</strong> of following recursive functi<strong>on</strong>s:<br />

NBV ariants(Π, C, δ):<br />

gives the cardinality of the set Appl(Π, C, δ)<br />

V ariant(n, Π, C, δ), where 1 ≤ n ≤ NBV ariants(Π, C, δ):<br />

gives the multiset of rules corresp<strong>on</strong>ding to the n-th element of some initially<br />

fixed enumerati<strong>on</strong> of Appl(Π, C, δ).<br />

It is clear that if each functi<strong>on</strong> is computed in c<strong>on</strong>stant time, then the multiset<br />

of rules to be applied can also be computed in a c<strong>on</strong>stant time. In what follows<br />

we will discuss methods for the c<strong>on</strong>structi<strong>on</strong> of these two functi<strong>on</strong>s for different<br />

classes of P systems.<br />

In the following we will need the noti<strong>on</strong> of the rules’ dependency graph. This<br />

is a weighted bipartite graph where the first partiti<strong>on</strong> U c<strong>on</strong>tains a node labeled<br />

by a for each object a of Π, while the sec<strong>on</strong>d partiti<strong>on</strong> V c<strong>on</strong>tains a node labeled<br />

by r for each rule r of Π. There is an edge between a node r ∈ V and a node<br />

a ∈ U labeled by a weight k if a k ∈ lhs(r) (and a k+1 ∉ lhs(r).<br />

Example 2. C<strong>on</strong>sider a P system Π 1 having two rules r 1 : ab → u and r 2 : bc →<br />

v. These rules have the following dependency graph:<br />

r 1<br />

❄<br />

r 2<br />

❄❄❄❄❄❄<br />

❄ 8 ❄❄❄❄❄❄❄<br />

7 777777 8 8888 a<br />

b<br />

c<br />

Let N a , N b and N c be the number of objects a, b and c in a c<strong>on</strong>figurati<strong>on</strong> C.<br />

We define<br />

N 1 = min(N a , N b )<br />

N 2 = min(N b , N c )<br />

N = min(N 1 , N 2 )<br />

Suppose that Π evolves in a maximally parallel derivati<strong>on</strong> mode. Then the<br />

set Appl(Π, C, max) can be computed as follows:<br />

Appl(Π, C, max) =<br />

⋃ { }<br />

r p+k1<br />

1 r q+k2<br />

2 ,<br />

p+q=N<br />

where k j = N j ⊖ N, 1 ≤ j ≤ 2, where ⊖ is the positive subtracti<strong>on</strong> operati<strong>on</strong>.<br />

From this representati<strong>on</strong> it is clear that NBV ariants(Π, C, max) = N + 1,<br />

which can be computed in c<strong>on</strong>stant time <strong>on</strong> FPGA.<br />

438

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

Saved successfully!

Ooh no, something went wrong!