scieee Science in your language
[en] (orig)

Computing with Spiking Neural P Systems: Traces and Small Universal Systems

Abstract

Recently, the idea of spiking neurons and thus of computing by spiking was incorporated into membrane computing, and so-called spiking neural P systems (abbreviated SN P systems) were introduced. Very shortly, in these systems neurons linked by synapses communicate by exchanging identical signals (spikes), with the information encoded in the distance between consecutive spikes. Several ways of using such devices for computing were considered in a series of papers, with universality results obtained in the case of computing numbers, both in the generating and the accepting mode; generating, accepting, or processing strings or infinite sequences was also proved to be of interest. In the present paper, after a short survey of central notions and results related to spiking neural P systems (including the case when SN P systems are used as string generators), we contribute to this area with two (types of) results: (i) we produce small universal spiking neural P systems (84 neurons are sufficient in the basic definition, but this number is decreased to 49 neurons if a slight generalization of spiking rules is adopted), and (ii) we investigate the possibility of generating a language by following the trace of a designated spike in its way through the neurons.

Read accessible full text

Computing with Spiking Neural P Systems: Traces and Small Universal Systems

Author: Ionescu, Mihai; Paun, Andrei; Paun, Gheorghe; Pérez Jiménez, Mario de Jesús
Publisher: Springer
Year: 2006
DOI: 10.1007/11925903_1
Source: https://idus.us.es/bitstreams/d861ad59-0bf2-4b3d-bdfd-f9b3c96b4440/download
Compu ing wi h Spiking Neu al P Sys ems:
T aces and Small Uni e sal Sys ems
Mihai Ionescu1, And ei P˘aun2,
Gheo ghe P˘aun3,4,andMa ioJ.P´e ez-Jim´enez4
1Resea ch G oup on Ma hema ical Linguis ics
Uni e si a Ro i a i Vi gili
Pl. Impe ial T`a aco 1, 43005 Ta agona, Spain
[email p o ec ed]
2Depa men o Compu e Science, Louisiana Tech Uni e si y
Rus on, PO Box 10348, Louisiana, LA-71272 USA, and
Uni e sidad Poli ´ecnica de Mad id – UPM, Faculdad de In o ma ´ıca
Campus de Mon egancedo s/n, Boadilla del Mon e
28660 Mad id, Spain
[email p o ec ed]
3Ins i u e o Ma hema ics o he Romanian Academy
PO Box 1-764, 014700 Bucha es , Romania
[email p o ec ed]
4Depa men o Compu e Science and AI, Uni e si y o Se illa
A da Reina Me cedes s/n, 41012 Se illa, Spain
[email p o ec ed], [email p o ec ed]
Abs ac . Recen ly, he idea o spiking neu ons and hus o compu ing
by spiking was inco po a ed in o memb ane compu ing, and so-called
spiking neu al P sys ems (abb e ia ed SN P sys ems) we e in oduced.
Ve y sho ly, in hese sys ems neu ons linked by synapses communica e
by exchanging iden ical signals (spikes), wi h he in o ma ion encoded
in he dis ance be ween consecu i e spikes. Se e al ways o using such
de ices o compu ing we e conside ed in a se ies o pape s, wi h uni-
e sali y esul s ob ained in he case o compu ing numbe s, bo h in he
gene a ing and he accep ing mode; gene a ing, accep ing, o p ocessing
s ings o infini e sequences was also p o ed o be o in e es .
In he p esen pape , a e a sho su ey o cen al no ions and e-
sul s ela ed o spiking neu al P sys ems (including he case when SN P
sys ems a e used as s ing gene a o s), we con ibu e o his a ea wi h
wo ( ypes o ) esul s: (i) we p oduce small uni e sal spiking neu al P
sys ems (84 neu ons a e sufficien in he basic defini ion, bu his num-
be is dec eased o 49 neu ons i a sligh gene aliza ion o spiking ules
is adop ed), and (ii) we in es iga e he possibili y o gene a ing a lan-
guage by ollowing he ace o a designa ed spike in i s way h ough he
neu ons.
1 In oduc ion
Spiking neu al P sys ems (in sho , SN P sys ems) we e in oduced in [6], wi h
he mo i a ion coming om wo di ec ions: he a emp o memb ane compu ing
o pass om cell-like a chi ec u es o issue-like o neu al-like a chi ec u es (see
[15], [12]), and he in iguing possibili y o encoding in o ma ion in he du a ion
o e en s, o in he in e al o ime elapsed be ween e en s, as i idly in es iga ed
in ecen esea ch in neu al compu ing (o “ hi d gene a ion”) [8], [9].
This double challenge led o a class o P sys ems based on he ollowing simple
ideas: le us use only one objec , he symbol deno ing a spike, and one-memb ane
cells (called neu ons) which can hold any numbe o spikes; each neu on fi es in
specified condi ions (a e collec ing a specified numbe o spikes) and hen sends
one spike along i s axon; his spike passes o all neu ons connec ed by a synapse
o he spiking neu on (hence i is eplica ed in o as many copies as many a ge
neu ons exis ); be ween he momen when a neu on fi es and he momen when
i spikes, each neu on needs a ime in e al, and his ime in e al is he essen ial
ing edien o he sys em unc ioning ( he basic in o ma ion ca ie – wi h he
men ioning ha also he numbe o spikes accumula ed in each momen in he
neu ons p o ides an impo an in o ma ion o con olling he unc ioning o
he sys em); one o he neu ons is conside ed he ou pu one, and i s spikes
p o ide he ou pu o he compu a ion. The sequence o ime momen s when
spikes a e sen ou o he sys em is called a spike ain. The ules o spiking
ake in o accoun all spikes p esen in a neu on no only pa o hem, bu no
all spikes p esen in a neu on a e consumed in his way; a e ge ing fi ed and
be o e sending he spike o i s synapses, he neu on is idle (biology calls his
he e ac o y pe iod) and canno ecei e spikes. The e a e also ules used o
“ o ge ing” some spikes, ules which jus emo e a specified numbe o spikes
om a neu on.
In he spi i o spiking neu ons, as he esul o a compu a ion (no necessa ily
a hal ing one) in [6] one conside s he numbe o s eps elapsed be ween he fi s
wo spikes o he ou pu neu on. E en in his es ic i e amewo k, SN P sys-
ems u ned ou o be Tu ing comple e, able o compu e all Tu ing compu able se s
o na u al numbe s. This holds bo h in he gene a i e mode (as ske ched abo e,
a numbe is compu ed i i ep esen s he in e al be ween he wo consecu i e
spikes o he ou pu neu on) and in he accep ing mode (a numbe is in oduced
in he sys em in he o m o he in e al o ime be ween he fi s wo spikes en e -
ing a designa ed neu on, and his numbe is accep ed i he compu a ion hal s).
I a bound is imposed on he numbe o spikes p esen in any neu on du ing a
compu a ion, hen a cha ac e iza ion o semilinea se s o numbe s is ob ained.
These esul s we e ex ended in [13] o se e al o he ways o associa ing a se o
numbe s wi h an SN P sys em: aking in o accoun he in e al be ween he fi s
kspikes o each spike ain, o all spikes, aking only al e na ely he in e als,
o all o hem, conside ing hal ing compu a ions. Then, he spike ain i sel
( he sequences o symbols 0, 1 desc ibing he ac i i y o he ou pu neu on: we
w i e 0 i no spike exi s he sys em in a ime uni and 1 i a spike is emi ed) was
conside ed as he esul o a compu a ion; he infini e case is in es iga ed in [14],
he fini e one in [2]. A se ies o possibili ies o handling infini e sequences o bi s
a e discussed in [14], while mo phic ep esen a ions o egula and o ecu si ely
enume able languages a e ound in [2]. The esul s om [2] a e b iefly ecalled
in Sec ion 5 below.
In his pape we di ec ly con inue hese in es iga ions, con ibu ing in wo
na u al di ec ions. Fi s , he abo e men ioned uni e sali y esul s ( he possibili y
o compu e all Tu ing compu able se s o numbe s) do no gi e an es ima ion on
he numbe o neu ons sufficien o ob aining he uni e sali y. Which is he size
o he smalles uni e sal “b ain” (o he o m o an SN P sys em)? This is bo h
a na u al and impo an ( om compu e science and, also, om neu o-science
poin o iew) p oblem, eminding he ex ensi e effo s paid o finding small
uni e sal Tu ing machines – see, e.g., [16] and he e e ences he ein.
Ou answe is a he su p ising/encou aging: 84 neu ons ensu e he uni e -
sali y in he basic se up o SN P sys ems, as hey we e defined in [6], while his
numbe is dec eased o 49 i sligh ly mo e gene al spiking ules a e used ( ules
wi h he possibili y o p oduce no only one spike, bu also wo o mo e spikes
a he same ime – such ules a e called ex ended).Thep oo isbasedonsimu-
la ing a small uni e sal egis e machine om [7]. (The ull de ails o he p oo
o hese esul s abou small uni e sal SN P sys ems will be p o ided elsewhe e
– see [11].)
Ex ended ules a e also use ul when gene a ing s ings: we associa e a symbol
biwi h a s ep when he sys em ou pu s ispikes and in his way we ob ain a
s ing o e an a bi a y alphabe , no only on he bina y one, as in he case
o s anda d ules. Especially flexible is he case when we associa e he emp y
s ing wi h a s ep when no spike is sen ou o he sys em we associa e ( ha is,
b0is in e p e ed as λ). Resul s om [3], conce ning he powe o ex ended SN P
sys ems as language gene a o s, a e also ecalled in Sec ion 5.
Then, ano he na u al issue is o b ing o he SN P sys ems a ea a no ion
in oduced o sympo /an ipo P sys ems in [5]: ma k a spike and ollow i s
pa h h ough he sys em, eco ding he labels o he isi ed neu ons un il ei he
he ma king disappea s o he compu a ion hal s. Because o he e y es ic i e
way o gene a ing s ings in his way, he e a e simple languages which canno
be compu ed, bu , on he o he hand, he e a e a he complex languages which
can be ob ained in his amewo k.
Due o space es ic ions, we do no gi e ull o mal de ails in defini ions and
p oo s (we e e o he abo e men ioned pape s o ha ); such de ails a e o will
be a ailable in sepa a e pape s o be ci cula ed/announced h ough [19].
2 Fo mal Language Theo y P e equisi es
We assume he eade o be amilia wi h basic language and au oma a heo y,
e.g., om [17] and [18], so ha we in oduce he e only some no a ions and no ions
used la e in he pape .
Fo an alphabe V,V∗deno es he se o all fini e s ings o symbols om
V; he emp y s ing is deno ed by λ, and he se o all nonemp y s ings o e V
is deno ed by V+.WhenV={a}is a single on, hen we w i e simply a∗and
a+ins ead o {a}∗,{a}+.I x=a1a2...a
n,a
i∈V, 1≤i≤n, hen he mi o
image o xis mi(x)=an...a
2a1.
A mo phism h:V∗
1−→ V∗
1such ha h(a)∈{a, λ} o each a∈V1is called
a p ojec ion, and a mo phism h:V∗
1−→ V∗
2such ha h(a)∈V2∪{λ} o each
a∈V1is called a weak coding.
I L1,L
2⊆V∗a e wo languages, he le and igh quo ien s o L1wi h
espec o L2a e defined by L2 L1={w∈V∗|xw ∈L1 o some x∈L2},
and espec i ely L1/L2={w∈V∗|wx ∈L1 o some x∈L2}. When he
language L2is a single on, hese ope a ions a e called le and igh de i a i es,
and deno ed by ∂l
x(L)={x} Land ∂
x(L)=L/{x}, espec i ely.
A Chomsky g amma is gi en in he o m G=(N,T,S,P), whe e Nis he
non e minal alphabe , Tis he e minal alphabe , S∈Nis he axiom, and
Pis he fini e se o ules. Fo egula g amma s, he ules a e o he o m
A→aB, A →a, o someA, B ∈N,a ∈T.
We deno e by FIN, REG, CF, CS, RE he amilies o fini e, egula , con ex -
ee, con ex -sensi i e, and ecu si ely enume able languages; by MAT we de-
no e he amily o languages gene a ed by ma ix g amma s wi hou appea ance
checking. The amily o Tu ing compu able se s o numbe s is deno ed by NRE
( hese se s a e leng h se s o RE languages, hence he no a ion).
Le V={b1,b
2,...,b
m}, o somem≥1. Fo a s ing x∈V∗, le us deno e
by alm(x) he alueinbasem+1o x(we use base m+ 1 in o de o conside
he symbols b1,...,b
mas digi s 1,2,...,m, hus a oiding he digi 0 in he le
hand o he s ing). We ex end his no a ion in he na u al way o se s o s ings.
All uni e sali y esul s o he pape a e based on he no ion o a egis e
machine. Such a de ice – in he non-de e minis ic e sion – is a cons uc M=
(m, H, l0,l
h,I), whe e mis he numbe o egis e s, His he se o ins uc ion
labels, l0is he s a label (labeling an ADD ins uc ion), lhis he hal label
(assigned o ins uc ion HALT), and Iis he se o ins uc ions; each label om H
labels only one ins uc ion om I, hus p ecisely iden i ying i . The ins uc ions
a e o he ollowing o ms:
–li:(ADD( ),l
j,l
k) (add 1 o egis e and hen go o one o he ins uc ions
wi h labels lj,l
knon-de e minis ically chosen),
–li:(SUB( ),l
j,l
k)(i egis e is non-emp y, hen sub ac 1 om i and go
o he ins uc ion wi h label lj, o he wise go o he ins uc ion wi h label
lk),
–lh:HALT ( he hal ins uc ion).
A egis e machine Mgene a es a se N(M) o numbe s in he ollowing way:
we s a wi h all egis e s emp y (i.e., s o ing he numbe ze o), we apply he
ins uc ion wi h label l0and we con inue o apply ins uc ions as indica ed by
he labels (and made possible by he con en s o egis e s); i we each he hal
ins uc ion, hen he numbe np esen in egis e 1 a ha ime is said o be
gene a ed by M. (Wi hou loss o gene ali y we may assume ha in he hal ing
configu a ion all o he egis e s a e emp y; also, we may assume ha egis e 1
is ne e subjec o SUB ins uc ions, bu only o ADD ins uc ions.) I is known
(see, e.g., [10]) ha egis e machines gene a e all se s o numbe s which a e
Tu ing compu able.
A egis e machine can also be used as a numbe accep ing de ice: we in-
oduce a numbe nin some egis e 0, we s a wo king wi h he ins uc ion
wi h label l0, and i he machine e en ually hal s, hen nis accep ed (we may
also assume ha all egis e s a e emp y in he hal ing configu a ion). Again,
accep ing egis e machines cha ac e ize NRE.
Fu he mo e, egis e machines can compu e all Tu ing compu able unc ions:
we in oduce he numbe s n1,...,n
kin some specified egis e s 1,...,
k,we
s a wi h he ins uc ion wi h label l0, and when we s op (wi h he ins uc ion
wi h label lh) he alue o he unc ion is placed in ano he specified egis e ,
, wi h all egis e s diffe en om being emp y. Wi hou loss o gene ali y we
may assume ha 1,...,
ka e he fi s k egis e s o M, and hen he esul o
he compu a ion is deno ed by M(n1,...,n
k).
In bo h he accep ing and he compu ing case, he egis e machines can be
de e minis ic, i.e., wi h he ADD ins uc ions o he o m li:(ADD( ),l
j)(add1
o egis e and hen go o he ins uc ion wi h label lj).
In he ollowing sec ions, when compa ing he powe o wo language gene -
a ing/accep ing de ices he emp y s ing λis igno ed.
3 Spiking Neu al P Sys ems
We gi e he e he basic defini ion we wo k wi h, in oducing SN P sys ems in he
o m conside ed in he small uni e sal SN P sys ems, hence compu ing unc ions
(which, ac ually, co e s bo h he gene a i e and accep ing cases).
A compu ing spiking neu al memb ane sys em (abb e ia ed SN P sys em), o
deg ee m≥1, is a cons uc o he o m
Π=(O,σ1,...,σ
m,syn,in,ou ),
whe e:
1. O={a}is he single on alphabe (ais called spike);
2. σ1,...,σ
ma e neu ons,o he o m
σi=(ni,R
i),1≤i≤m,
whe e:
a) ni≥0is heini ial numbe o spikes con ained in σi;
b) Riis a fini e se o ules o he ollowing wo o ms:
(1) E/ac→a;d,whe eEis a egula exp ession1o e a,c≥1, and
d≥0;
(2) as→λ, o s≥1, wi h he es ic ion ha o each ule E/ac→a;d
o ype (1) om Ri,weha eas/∈L(E);
3. syn ⊆{1,2,...,m}×{1,2,...,m}wi h (i, i)/∈syn o 1 ≤i≤m(synapses
be ween neu ons);
4. in, ou ∈{1,2,...,m}indica e he inpu and he ou pu neu ons o Π.
1The egula language defined by Eis deno ed by L(E).

The ules o ype (1) a e i ing (we also say spiking) ules, and hey a e applied
as ollows. I he neu on σicon ains kspikes, and ak∈L(E),k ≥c, hen he
ule E/ac→a;d∈Rican be applied. This means consuming ( emo ing) c
spikes ( hus only k−c emain in σi), he neu on is fi ed, and i p oduces a spike
a e d ime uni s (as usual in memb ane compu ing, a global clock is assumed,
ma king he ime o he whole sys em, hence he unc ioning o he sys em is
synch onized). I d= 0, hen he spike is emi ed immedia ely, i d= 1, hen he
spike is emi ed in he nex s ep, e c. I he ule is used in s ep and d≥1, hen
in s eps , +1, +2,..., +d−1 he neu on is closed ( his co esponds o he
e ac o y pe iod om neu obiology), so ha i canno ecei e new spikes (i a
neu on has a synapse o a closed neu on and ies o send a spike along i , hen
ha pa icula spike is los ). In he s ep +d, he neu on spikes and becomes
again open, so ha i can ecei e spikes (which can be used s a ing wi h he
s ep +d+1).
The ules o ype (2) a e o ge ing ules; hey a e applied as ollows: i he
neu on σicon ains exac ly sspikes, hen he ule as→λ om Rican be used,
meaning ha all sspikes a e emo ed om σi.
I a ule E/ac→a;do ype (1) has E=ac, hen we will w i e i in he
ollowing simplified o m: ac→a;d. I all spiking ules a e o his o m, hen
he sys em is said o be ini e (i can handle only a bounded numbe o spikes
in each o i s neu ons).
In each ime uni , i a neu on σican use one o i s ules, hen a ule om Ri
mus be used. Since wo fi ing ules, E1/ac1→a;d1and E2/ac2→a;d2,can
ha e L(E1)∩L(E2)=∅, i is possible ha wo o mo e ules can be applied in a
neu on, and in ha case, only one o hem is chosen non-de e minis ically. No e
howe e ha , by defini ion, i a fi ing ule is applicable, hen no o ge ing ule
is applicable, and ice e sa.
Thus, he ules a e used in he sequen ial manne in each neu on, bu neu ons
unc ion in pa allel wi h each o he ( he sys em is synch onized).
The ini ial configu a ion o he sys em is desc ibed by he numbe s n1,n
2,...,
nm, o spikes p esen in each neu on, wi h all neu ons being open. Du ing he
compu a ion, a configu a ion is desc ibed by bo h he numbe o spikes p esen
in each neu on and by he s a e o he neu on, mo e p ecisely, by he numbe o
s eps o coun down un il i becomes open ( his numbe is ze o i he neu on is
al eady open).
A compu a ion in a sys em as abo e s a s in he ini ial configu a ion. In
o de o compu e a unc ion :Nk−→ N, we in oduce kna u al numbe s
n1,...,n
kin he sys em by “ eading” om he en i onmen a bina y sequence
z=0
b10n1−110n2−11...10nk−110 , o someb, ≥0; his means ha he inpu
neu on o Π ecei es a spike in each s ep co esponding o a digi 1 om he
s ing z. No e ha we inpu exac ly k+ 1 spikes. The esul o he compu a ion
is also encoded in he dis ance be ween wo spikes: we impose o he sys em o
ou pu exac ly wo spikes and hal (some imes a e he second spike), hence
p oducing a ain spike o he o m 0b10 −110 , o someb, ≥0andwi h
= (n1,...,n
k).
I we use an SN P sys em in he gene a i e mode, hen no inpu neu on is
conside ed, hence no inpu is aken om he en i onmen ; we s a om he
ini ial configu a ion and he dis ance be ween he fi s wo spikes o he ou pu
neu on (o o he numbe s, see he discussion in he In oduc ion) is he esul o
he compu a ion. Dually, we can igno e he ou pu neu on, we inpu a numbe
in he sys em as he dis ance be ween wo spikes en e ing he inpu neu on, and
i he compu a ion hal s, hen he numbe is accep ed.
We do no gi e he e examples, because in he nex sec ion we show he ou
basic modules o ou small uni e sal SN P sys em.
4 Two Small Uni e sal SN P Sys ems
In bo h he gene a ing and he accep ing case, SN P sys ems a e uni e sal,
hey compu e he Tu ing compu able se s o numbe s. The p oo s om [6], [13]
a e based on simula ing egis e machines, which a e known o be equi alen
o Tu ing machines when compu ing (gene a ing o accep ing) se s o numbe s,
[10]. In [7], he egis e machines a e used o compu ing unc ions, wi h he
uni e sali y defined as ollows. Le (ϕ0,ϕ
1,...) be a fixed admissible enume a ion
o he se o una y pa ial ecu si e unc ions. A egis e machine Muis said o
be uni e sal i he e is a ecu si e unc ion gsuch ha o all na u al numbe s
x, y we ha e ϕx(y)=Mu(g(x),y). In [7], he inpu is in oduced in egis e s 1
and 2, and he esul is ob ained in egis e 0 o he machine.
l0:(SUB(1),l
1,l
2),l
1:(ADD(7),l
0),
l2:(ADD(6),l
3),l
3:(SUB(5),l
2,l
4),
l4:(SUB(6),l
5,l
3),l
5:(ADD(5),l
6),
l6:(SUB(7),l
7,l
8),l
7:(ADD(1),l
4),
l8:(SUB(6),l
9,l
0),l
9:(ADD(6),l
10),
l10 :(SUB(4),l
0,l
11),l
11 :(SUB(5),l
12,l
13),
l12 :(SUB(5),l
14,l
15),l
13 :(SUB(2),l
18,l
19),
l14 :(SUB(5),l
16,l
17),l
15 :(SUB(3),l
18,l
20),
l16 :(ADD(4),l
11),l
17 :(ADD(2),l
21),
l18 :(SUB(4),l
0,l
h),l
19 :(SUB(0),l
0,l
18),
l20 :(ADD(0),l
0),l
21 :(ADD(3),l
18),
lh:HALT.
Fig. 1. The uni e sal egis e machine om [7]
The cons uc ions om [6] do no p o ide a bound on he numbe o neu ons,
bu such a bound can be ound i we s a om a specific uni e sal egis e
machine. We will use he e he one wi h 8 egis e s and 23 ins uc ions om [7] –
o he eade con enience, his machine is ecalled in Figu e 1, in he no a ion
and he se up in oduced in he p e ious sec ion.
Theo em 1. The e is a uni e sal SN P sys em wi h 84 neu ons.
P oo . (Ou line) We ollow he way used in [6] o simula e a egis e machine by
an SN P sys em. This is done as ollows: neu ons a e associa ed wi h each egis e
( )andwi heachlabel(li) o he machine; i a egis e con ains a numbe n,
hen he associa ed neu on will con ain 2nspikes; modules as in Figu es 2 and 3
a e associa ed wi h he ADD and he SUB ins uc ions (each o hese modules
con ains wo neu ons – wi h p imed labels – which do no co espond o egis e s
and labels o he simula ed machine).
#
"
!#
"
!

 


#
"
!



SSS
Sw



HHHHH
Hj







ZZZ
Z~
li
a2→a;0
a→λ
l
il
i
a→a;0 a→a;0
lj
a2→a;0
a→λ
Fig. 2. Module ADD (simula ing li:(ADD( ),l
j))
The wo k o he sys em is igge ed by in oducing wo spikes in he neu on
σl0(associa ed wi h he s a ing ins uc ion o he egis e machine). In gene al,
he simula ion o an ADD o SUB ins uc ion s a s by in oducing wo spikes
in he neu on wi h he ins uc ion label. We do no desc ibe he e in de ail he
(p e y anspa en ) way he modules om Figu es 2 and 3 wo k – he eade
can consul [6] in his espec .
S a ing wi h neu ons σ1and σ2al eady loaded wi h 2g(x)and2yspikes,
espec i ely, and in oducing wo spikes in neu on σl0, we can compu e in ou
sys em in he same way as Mu; i he compu a ion hal s, hen neu on σ0will
con ain 2ϕx(y) spikes. Wha emains o do is o cons uc inpu and ou pu
modules, o eading a sequence o bi s and in oducing he igh numbe o
spikes in he neu ons co esponding o egis e s 1 and 2, and, in he end o he
compu a ion, o ou pu he con en s o egis e 0. Modules o hese ypes a e
gi en in Figu es 4, 5, ha ing se en and wo addi ional neu ons, espec i ely.
A e his di ec cons uc ion, we ge a sys em wi h 91 neu ons (9 o he
egis e s o he s a ing egis e machine – one u he egis e is necessa y o
echnical easons, 25 o i s labels, 24 ×2 o he ADD and SUB ins uc ions, 7
in he inpu module, and 2 in he ou pu module). Howe e , some “code op i-
miza ion” is possible, based on ce ain p ope ies o he egis e machine om
&
$
%






&
&
& %




/
AAAAAA
AU
QQQQ
Qs
?



/
HHHHH
Hj
JJ
J^
lia2→a;0
a→λ
(a2)+a/a3→a;0
a→a;1
l
i
l
i
a→a;0
a→a;1
lj
a2→a;0
a→λ
a2→a;0
a→λ
lk
Fig. 3. Module SUB (simula ing li:(SUB( ),l
j,l
k))
[7] ( o ins ance, consecu i e ADD ins uc ions can be simula ed by a specific
module, smalle han wo sepa a e ADD modules); we skip he echnical de ails
and we only men ion ha he final SN P sys em will con ain 84 neu ons.
This is a small numbe (a small “b ain”, compa ed o he human one; i would be
nice o know whe e in he e olu ion scale he e a e animals wi h abou 84 neu ons
in hei b ain), bu we do no know whe he i is op imal o no . Anyway, we
belie e ha in he p e ious se up, we canno significan ly dec ease he numbe
o neu ons om a uni e sal SN P sys em.
Howe e , we can do be e s a ing om he ollowing obse a ion. In many
modules men ioned abo e we need pai s o in e media e neu ons o duplica ing
he spike o be ansmi ed u he ( his is he case o neu ons σl
i,σ
l
iin Figu e
2), and his sugges s o conside a sligh ex ension o he ules o SN P sys ems:
o allow spiking ules o he o m E/ac→ap;d, whe e all componen s a e as
usual, and p≥1. The meaning is ha cspikes a e consumed and pspikes a e
p oduced. To be “ ealis ic”, we impose he es ic ion c≥p( he numbe o
p oduced spikes is no la ge han he numbe o consumed spikes).
Theo em 2. The e is a uni e sal SN P sys em wi h 49 neu ons, using ules o
he o m E/ac→ap;0,wi hp≥1.
(No e ha he delay is ze o in he ules o he ex ended o m used in he he-
o em.) As abo e, we do no know whe he his esul is op imal, bu we again
belie e ha i canno be significan ly imp o ed (wi hou , maybe, changing he
defini ion o SN P sys ems in an essen ial way).
2. H. Chen, R. F eund, M. Ionescu, Gh. P˘aun, M.J. P´e ez-Jim´enez: On s ing lan-
guages gene a ed by spiking neu al P sys ems. In P oc. o Fou h. B ains o ming
Week on Memb ane Compu ing, Se illa, 2006, ol. I, 169–193 (also a ailable a
[19]).
3. H. Chen, T.-O. Ishdo j, Gh. P˘aun, M.J. P´e ez-Jim´enez: Spiking neu al P sys ems
wi h ex ended ules. In P oc. o Fou h. B ains o ming Week on Memb ane Com-
pu ing, Se illa, 2006, ol. I, 241–266 (also a ailable a [19]).
4. O.H. Iba a, A. P˘aun, Gh. P˘aun, A. Rod ´ıguez-Pa ´on, P. Sosik, S. Woodwo h:
No mal o ms o spiking neu al P sys ems. In Fou h B ains o ming Week on
Memb ane Compu ing, Feb . 2006, Fenix Edi o a, Se illa, 2006, ol. II, 105–136
(also a ailable a [19]).
5. M. Ionescu, C. Ma in-Vide, A. P˘aun, Gh. P˘aun: Memb ane sys ems wi h sym-
po /an ipo : (unexpec ed) uni e sali y esul s. In P oc. 8 h In e na ional Mee ing
o DNA Based Compu ing (M. Hagiya, A. Ohuchi, eds.), Japan, 2002, 151–160.
6. M. Ionescu, Gh. P˘aun, T. Yokomo i: Spiking neu al P sys ems. Fundamen a In-
o ma icae, 71, 2-3 (2006), 279–308.
7. I. Ko ec: Small uni e sal egis e machines. Theo e ical Compu e Science, 168
(1996), 267–301.
8. W. Maass: Compu ing wi h spikes. Special Issue on Founda ions o In o ma ion
P ocessing o TELEMATIK, 8, 1 (2002), 32–36.
9. W. Maass, C. Bishop, eds.: Pulsed Neu al Ne wo ks, MIT P ess, Camb idge, 1999.
10. M. Minsky: Compu a ion – Fini e and In ini e Machines. P en ice Hall, Englewood
Cliffs, NJ, 1967.
11. A. P˘aun, Gh. P˘aun: Small uni e sal spiking neu al P sys ems. BioSys ems, o
appea .
12. Gh. P˘aun: Memb ane Compu ing – An In oduc ion. Sp inge -Ve lag, Be lin, 2002.
13. Gh. P˘aun, M.J. P´e ez-Jim´enez, G. Rozenbe g: Spike ains in spiking neu al P
sys ems. In e n. J. Found. Compu e Sci., o appea (also a ailable a [19]).
14. Gh. P˘aun, M.J. P´e ez-Jim´enez, G. Rozenbe g: Infini e spike ains in spiking neu al
P sys ems. Submi ed, 2006.
15. Gh. P˘aun, Y. Sakakiba a, T. Yokomo i: P sys ems on g aphs o es ic ed o ms.
Publica iones Ma hema icae Deb ecen, 60 (2002), 635–660.
16. Y. Rogozhin: Small uni e sal Tu ing machines. Theo e ical Compu e Science, 168
(1996), 215–240.
17. G. Rozenbe g, A. Salomaa, eds.: Handbook o Fo mal Languages,3 olumes.
Sp inge -Ve lag, Be lin, 1997.
18. A. Salomaa: Fo mal Languages. Academic P ess, New Yo k, 1973.
19. The P Sys ems Web Page: h p://psys ems.disco.unimib.i .