<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Category Methods for Analysis of Two Approaches to Modelling Logical Time Based on Concept of Clocks</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Grygoriy Zholtkevych</string-name>
          <email>g.zholtkevych@karazin.ua</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Hassan Khalil El Zein</string-name>
          <email>dr.hassanelzein@icloud.com</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Lyudmyla Polyakova</string-name>
          <email>l.yu.polyakova@karazin.ua</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Math and Comput. Sci. School, V.N. Karazin Kharkiv National University 4</institution>
          ,
          <addr-line>Svobody Sqr., Kharkiv, 61022</addr-line>
          ,
          <country country="UA">Ukraine</country>
        </aff>
      </contrib-group>
      <abstract>
        <p>This paper continues the study that has been begun in the series of authors' papers and has devoted to clarifying the interconnection between denotational and operational approaches for modelling logical time based on the concept of logical clocks. In the paper, a new category, namely the category of schedules, has introduced. The original definition of a morphism of schedules precedes to introducing this category. Refinement of a number of results of previous papers made it possible to establish the functorial nature of the method of associating the linear clock structure with a schedule and to prove that the corresponding functor determines the equivalence of the category of schedules and the category of linear clock structures.</p>
      </abstract>
      <kwd-group>
        <kwd>logical time</kwd>
        <kwd>clock</kwd>
        <kwd>denotational semantic model</kwd>
        <kwd>operational semantic model</kwd>
        <kwd>schedule</kwd>
        <kwd>clock structure</kwd>
        <kwd>clock morphism</kwd>
        <kwd>schedule morphism</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>-</title>
      <p>
        The trend of widespread use of distributed computing, observed in recent years,
is a technological answer to the practical achievement of the upper bound of
processor performance on the one side and the development of communication
tools on the other. In addition, there is a tendency to integrate cybernetic and
physical systems, which has been accelerated in the context developing
Internetof-Things. You can find the more detailed analysis of these trends in [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ].
      </p>
      <p>But this analysis allows us to state that the problems associated with parallel,
distributed and parallel computations turned out to be on the leading edge of</p>
    </sec>
    <sec id="sec-2">
      <title>Computer Science and Information Technology.</title>
      <p>
        This work is focused on the problem of modelling logical time in distributed
systems, in particular on the model based on the concept of logical time, and is
a continuation of authors’ results presented in [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ]1.
      </p>
      <p>
        1 Preliminary results of this article are available by http://ceur-ws.org/Vol-1844/
10000488.pdf in [
        <xref ref-type="bibr" rid="ref2">2</xref>
        ].
      </p>
      <p>The mentioned above paper is focused on the two approaches based on
logical clocks to the modelling of time. Authors had shown that the language of
the category theory is adequate for building models of Universum of events.
In other words, the denotational approach to the modelling can well be
formulated in the terms of the category theory. In contrast, for building operational
models, authors have selected the set-theoretic approach. Unfortunately, the
difference between mathematical backgrounds led to unclarity and incompleteness
of describing the relationship between two classes of models, denotational and
operational.</p>
      <p>Thus, we see that the relationship of two approaches for determining
semantic meaning of temporal constraint specifications based on the concept of logical
clocks did not clarify by the paper. Therefore, the following problem arises.
Problem. Establish the nature of the relationship between the two approaches
described above to modelling of logical time based on the concept of clocks.
The object of this paper to identify and ground the required relationship in the
terms of category theory.</p>
      <p>
        This paper has the following structure: Section 2 reminds the
categorytheoretic description of the denotational model of logical time given in [
        <xref ref-type="bibr" rid="ref1 ref2">2, 1</xref>
        ];
Section 3 contains the category-theoretic description the operational model of
logical time.
      </p>
      <p>Section 2 has an overview character and contains a small number of new
results, which are mostly of an auxiliary nature.</p>
      <p>In the contrast, section 3 is original. It contains the definitions of
schedule morphisms, the description of the category of schedules, and the proof of
Main Theorem that establishes the equivalence of the category of linear clock
structures and the category of schedules. This equivalence is precisely the
relationship that eluded authors in the set-theoretic formulation.
2</p>
      <sec id="sec-2-1">
        <title>Category of Clock Structures</title>
        <p>
          The main definitions and results obtained in [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ] are collected in this section. We
do not give those proofs that do not require any changes in comparison with the
corresponding proofs in [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ]. A proof is given only if this proof simplifies the
corresponding proof in [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ] or if this proof establishes a new fact.
        </p>
        <p>
          Consideration of causality relationship as a quasi-order (i.e. a reflexive and
transitive relation) on the set of event occurrences is a generally accepted
approach. Following this approach, we understand an Universum of events as a
set I of event occurrences with a quasi-order “4” . Usually, we consider along
with the relation “4” the relations , , # , and k . These relations are defined
by the causality relation as in Table 1 (see also [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ]). Now we give the
definition of clock structure as an Universum of events with additional structures and
properties. First of all we fix a finite set C whose each element is interpreted as
a reference to the source (it is called a clock) of occurrences of the same event.
        </p>
      </sec>
    </sec>
    <sec id="sec-3">
      <title>This leads us to the following definition in accordance with [1].</title>
      <p>Definition 1. A C-structure is a triplet S = I; ; 4 where</p>
      <sec id="sec-3-1">
        <title>I is the set of instants corresponding to the occurrences of events,</title>
        <p>: I ! C is a surjective mapping that associates the clock that is the source
of an instant with this instant,
“4” is a quasi-order on I that models the causality relation between
instants.</p>
        <p>The triplet meets also the following axioms
the axiom of unbounded liveness: the set I is infinite;
the axiom of finite causality: for any i 2 I , the corresponding principal ideal
( i ] is finite;
the axiom of total ordering for clock timelines: for each c 2 C , the c-timeline
Ic = 1(c) is linearly ordered by “ ”.</p>
        <p>
          This definition is not a novelty introduced in [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ], it was used in accordance with
[
          <xref ref-type="bibr" rid="ref3">3</xref>
          ]. But the following definition is such a novelty.
        </p>
        <p>Definition 2. Let S0 = (I0; 0; 4) and S00 = (I00; 00; 4) be C-structures then
a mapping f : I0 ! I00 is called a morphism of C-structures if the following
holds
for any i 2 I0 , the equation 00 ( f i) = 0 i is fulfilled,
for any i 2 I0 and j 2 I0 , f i 4 f j whenever i 4 j ,
for any i 2 I0 and j 2 I0 , f i # f j whenever i # j .</p>
      </sec>
    </sec>
    <sec id="sec-4">
      <title>The following proposition is evident.</title>
      <p>Proposition 1. The class of all C-structures equipped with morphisms of
Cstructures forms a category denoted below by StructC .</p>
      <p>
        An important special class of clock structures is formed by so-called linear
clock structures, defined in [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ].
is false for all i; j 2 I .
of clock structures.
      </p>
      <p>Definition 3. A C-structure L = (I; ; 4) is called a linear structure if i k j</p>
      <p>
        Now let us recall some results obtained in [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ] concerning the arrangement
Definition 4. Let S = ( ; ; 4) be a clock structure, A
      </p>
      <p>I</p>
      <sec id="sec-4-1">
        <title>I and i 2 A then i is called a minimal instant in A if the statement j i is false for any j 2 A.</title>
        <p>To refer to the subset of minimal instants of A we use the denotation min A .</p>
      </sec>
    </sec>
    <sec id="sec-5">
      <title>We associate the sequence of slices</title>
      <p>structure S in the following manner</p>
      <p>
        I[0] , I[
        <xref ref-type="bibr" rid="ref1">1</xref>
        ] , . . . , I[n] , . . .
      </p>
      <p>with any
CI[0] = min I ;
I[n] = min I n S I[k]</p>
      <p>!
n 1
k=0
for n 2 N+
(1)
where I is the instant set of S .</p>
      <p>Proposition 2. For a C-structure S = (I; ; 4) the following properties hold
3. the sequence of slices is a covering of the set of instants;
4. if i 2 I[n] , j 2 I , and j</p>
      <p>i then j 2 I[n] for some n 2 N ;
5. if i 2 I[n + 1] then there exists j 2 I[n] such that j i for some n 2 N .
We will need some other properties of slices. From item 2 we directly obtain the
follows.
implies i</p>
      <p>Corollary 1. If L with an instant set I is a linear C-structure then i; j 2 I[n]
Corollary 2. The restriction</p>
      <p>on I[n] is injective for all n 2 N .</p>
      <p>Proof. If i =</p>
      <p>j for some i; j 2 I[n] then the axiom of total ordering for the</p>
    </sec>
    <sec id="sec-6">
      <title>Further i</title>
      <p>j together with the mentioned axiom implies i = j .
clock timelines of Def. 1 makes impossible the case i k j in item 2 of Prop. 2.
tu
i 2 I[m] and j 2 I[n] then the following properties hold
Proposition 3. Let S = ( ; ; 4) be a C-structure and for some m; n 2 N ,</p>
      <p>I
1. if i 4 j then m
n , while i</p>
      <p>j implies m &lt; n ;</p>
      <sec id="sec-6-1">
        <title>2. if S is linear and m</title>
        <p>n then i 4 j , while m &lt; n implies i
j .
existence of j0 2 I[n] such that j0
Prop. 2 the fact j0; j 2 I[n] .</p>
      </sec>
    </sec>
    <sec id="sec-7">
      <title>To prove item 2 note the following.</title>
      <p>i . But j0
i 4 j contradicts by item 2 of
Proof. To prove item 1 suppose m &gt; n . Then item 5 of Prop. 2 implies the
If m = n we directly obtain the statement by Cor 1.
i0
j then i0
i and from i 4 i0
j we conclude i
j .</p>
      <p>If m &lt; n then item 5 of Prop. 2 implies the existence of i0 2 I[m] such that</p>
      <p>The established properties of the clock structures make it possible to
establish the properties of their morphisms.
for some m; n 2 N the following properties hold</p>
      <sec id="sec-7-1">
        <title>Proposition 4. For any morphism f : S0 !</title>
        <p>S00 from the C-structure S0 =
(I0; 0; 4) into the C-structure S00 = (I00; 00; 4) and i 2 I0[m] , f i 2 I00[n]
3. n</p>
        <p>m ;
4. if f is an isomorphism then m = n .</p>
      </sec>
      <sec id="sec-7-2">
        <title>2. if for some j 2 I0, we have i j then f j 2 I00[n] ; 1. the mapping f : I0 ! I00 preserves relations “4”, “ ”, “ ”, and “#”;</title>
        <p>Proof. To prove item 1 note that f preserves relations “4” and “#” by Def. 2.</p>
      </sec>
    </sec>
    <sec id="sec-8">
      <title>The statement i</title>
      <p>j is equivalent to i 4 j and j 4 i and, therefore, f preserves
“ ” . Further taking into account that i
j if and only if i 4 j and i # j one can
immediately obtain that f preserves “ ” .</p>
    </sec>
    <sec id="sec-9">
      <title>To prove item 2 let us use proven item 1. Really, i j ensures, as item 1 claims,</title>
      <p>f i</p>
      <p>f j. Now use of item 4 of Prop. 2 leads to the required statement.
To prove item 3 we use induction in m . For m = 0 it is true that n
m . Suppose
n &lt; k + 1 . Then by item 5 of Prop. 2 there exists j 2 I0[n] such that j
the statement holds for m</p>
      <p>k and let m = k + 1 , i 2 I0[m] , f i 2 I00[n]. Suppose
above shown f j
into account j</p>
      <p>f i . If f j 2 I00[r] then by induction hypothesis r
i and item 1 of Prop. 3 one can derive that r &lt; n . Thus, we
i. As
n. Taking
have obtained two mutually excluding inequality r
n and r &lt; n and conclude
that our supposition is incorrect i.e. n
m .</p>
    </sec>
    <sec id="sec-10">
      <title>Item 4 follows immediately from item 3.</title>
    </sec>
    <sec id="sec-11">
      <title>Morphisms of linear clock structures hold additional properties.</title>
      <p>following properties hold
1. if f i
f j then i
j ;</p>
      <sec id="sec-11-1">
        <title>Proposition 5. For any morphism f : L</title>
        <p>0 !</p>
      </sec>
      <sec id="sec-11-2">
        <title>L00 from the linear C-structure</title>
        <p>L
0 = (I0; 0; 4) into the linear C-structure L00 = (I00; 00; 4) and i; j 2 I0 the
2. if i 2 I0[m] , j 2 I0 , f i; f j 2 I00[n] for some m; n 2 N then j 2 I0[m] ;
tu
tu
3. f I0[m]</p>
      </sec>
      <sec id="sec-11-3">
        <title>I00[n] for some n such that m</title>
        <p>n .
structures. Item 1 of Prop. 4 ensures f i # f j but this contradicts to f i
Proof. To prove item 1 note that i 6 j is equivalent to i # j for linear clock
f j .</p>
        <p>To prove item 2 note f i; f j 2 I00[n] ensures f i</p>
      </sec>
    </sec>
    <sec id="sec-12">
      <title>Taking into account the previous item one can conclude that i j . Now we need f j for a linear clock structure. to use item 4 of Prop. 2 to obtain the required statement.</title>
    </sec>
    <sec id="sec-13">
      <title>Item 3 follows directly from the previous item.</title>
      <p>tu</p>
      <p>The operational approach to describe logical time dependences in distributed
systems (including cyber-physical systems) is based on observing streams of
system messages. Below this approach is described with the language of the
category theory.</p>
      <p>Definition 5. Let C be a finite set of logical clocks then any non-empty subset
of C is called a clock message.</p>
      <p>Informally, a clock message contains the information about which clocks ticked
at the same time-point.
the denotation MC .</p>
      <p>To refer to the set of clock messages associated with a clock set C we use below
Definition 6. A schedule (or more precisely a C-schedule) is an infinite
sequence</p>
      <p>
        = ( [0]; [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ]; : : : ; [n]; : : :) of clock messages.
      </p>
      <p>
        In [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ], authors have defined the relation of modelling between linear clock
structures and sets of schedules. Unfortunately, this relation does not associate
natural sets of schedules with linear clock structures.
      </p>
      <p>Thus, one can see that the use of the category theory language only for
the denotational semantic model does not give a tool for establishing the
required relationship with the operational semantic model being formulated in the
set-theoretic terms. In this context, an attempt to reformulate the operational
semantic model using the category theory language seems reasonable.
3</p>
      <sec id="sec-13-1">
        <title>Category of Schedules</title>
        <p>The operational approach to describe logical time dependences in distributed
systems (including cyber-physical systems) is based on observing streams of
system messages. Below this approach is described with the language of the
category theory.</p>
        <p>Definition 7. Let C be a finite set of logical clocks then any non-empty subset
of C is called a clock message.</p>
        <p>Informally, a clock message contains the information about which clocks ticked
at the same time-point.</p>
        <p>To refer to the set of clock messages associated with a clock set C we use below
the denotation MC .</p>
        <p>
          Definition 8. A schedule (or more precisely a C-schedule) is an infinite
sequence = ( [0]; [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ]; : : : ; [n]; : : :) of clock messages.
        </p>
        <p>For two C-schedules 0 and 00 , a C-morphism from 0 into 00 is a triple
h 0; k; 00i where k : N ! N is an injective mapping such that for any n 2 N
1. C-morphisms of the form h ; 1N; i are units of this composition;</p>
      </sec>
    </sec>
    <sec id="sec-14">
      <title>2. the associative law is fulfilled for this composition.</title>
    </sec>
    <sec id="sec-15">
      <title>Thus, the following statement is true.</title>
      <p>Proposition 6. The set of all C-schedules equipped with C-morphisms of
schedules forms a small category denoted below by SchedC .</p>
    </sec>
    <sec id="sec-16">
      <title>Now we can formulate the principal result of this paper.</title>
      <p>Main Theorem. The categories LinStructC and SchedC are equivalent i.e.
there exists a pair of functors</p>
      <p>F : SchedC ! LinStructC
and</p>
      <p>
        G : LinStructC ! SchedC
such that F G is naturally isomorphic to the identity endofunctor of LinStructC
and G F is naturally isomorphic to the identity endofunctor of SchedC .
Note 1. All necessary definitions and facts about natural transformations and
natural isomorphisms can be found in [
        <xref ref-type="bibr" rid="ref4">4</xref>
        ].
      </p>
      <p>We use the following theorem as the main tool to establish the validity of Main</p>
    </sec>
    <sec id="sec-17">
      <title>Theorem.</title>
      <p>Theorem 1. Categories C and D are equivalent if and only if there exists a
functor F : C ! D such that
in the following manner</p>
      <p>L = hI ; ; 4i
where
isomorphic;
is a bijection.
1. for any object d in D , there exists an object c in C such that F c and d are
2. for any objects c0 and c00 in C , the mapping F : C c0; c00 ! D F c0; F c00
The proof of this theorem one can find in [5, Theorem 7.1], the proof of the
more general statement is given in [4, IV.4, Theorem 1].</p>
      <p>
        In [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ], the linear C-structure L
was associated with any
2 MC schedule
I = fhc; ni 2 C
      </p>
      <p>N j c 2 [n]g ;
hc; ni = c
for hc; ni 2 I ;
hc0; n0i 4 hc00; n00i if and only if n0
n00
for hc0; n0i; hc00; n00i 2 I :
Here we generalize this association by its extension up to a functor F from the
category SchedC into the category LinStructC .</p>
    </sec>
    <sec id="sec-18">
      <title>To do this we assign the required correspondences as follows for</title>
      <p>2 SchedC
for h 0; k; 00i 2 SchedC 0; 00
and hc; ni 2 L
0</p>
      <p>F</p>
      <p>= L ;
F h 0; k; 00i hc; ni = hc; k(n)i :
Now we need to check the validity of some statements. The corresponding
checks are gathered in the following proposition.</p>
      <p>Proposition 7. Let 0 , 00 , and 000 be objects of SchedC then
1. for any hc; ni 2 I
belongs to I</p>
      <sec id="sec-18-1">
        <title>2. for any C-morphisms h 0; k1; 00i and h 00; k2; 000i, the following is true</title>
        <p>0 and h 0; k; 00i 2 SchedC 0; 00 , the item hc; k(n)i
F h 0; k1; 00
i h 00; k2; 000i = F h 0; k1; i00
F h 00; k2; 000i ;</p>
      </sec>
      <sec id="sec-18-2">
        <title>3. Fh 0; 1; 0i is the identity mapping from L</title>
        <p>0 into itself .</p>
        <p>Proof. Item 1 follows immediately from the definition of a schedule morphism
(see Def. 8).</p>
      </sec>
    </sec>
    <sec id="sec-19">
      <title>Item 2 and item 3 are checked by direct calculation. (2a) (2b)</title>
      <p>Corollary 3. Formulae (2a) and (2b) determine a functor</p>
      <p>F : SchedC ! LinStructC :</p>
    </sec>
    <sec id="sec-20">
      <title>Below the following lemma are also needed for us.</title>
      <p>is isomorphic to L .</p>
      <p>Lemma. For any linear C-structure L , there exists C-schedule such that L
Proof. Let L = ( ; ; 4) then we assign [n] =</p>
      <p>
        I
quence
= [0]; [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ]; : : : ; [n]; : : : is a C-schedule.
i for some i 2 I[n] . Cor. 2 guarantees that such i is uniquely
      </p>
      <p>N and hc; ni 2 I if c 2
[n] . In other words,
determined. Thus, we can determine the mapping f : I
conditions</p>
      <p>f hc; ni = i 2 I[n]
ensures that f is a morphism from L into L .</p>
      <p>if and only if
f (g i) = i for any i 2 I .</p>
      <p>Thus, we have proven that L and L are isomorphic.</p>
      <p>Now consider the mapping g : I
! I determined as follows
g i = h i; ni
where i 2 I[n] . The correctness of g is ensured by the construction of L .
Item 1 of Prop. 3 ensures that g is a morphism from L into L .</p>
      <p>It is evident that by construction g f hc; ni = hc; ni for any hc; ni 2 I and
tu
! I by the following
i = c . Item 2 of Prop. 3</p>
    </sec>
    <sec id="sec-21">
      <title>Now we have all necessary to prove Main Theorem.</title>
      <p>Proof (of Main Theorem). Our proof is based on applying Theorem 1 with the
constructed above functor F : SchedC ! LinStructC . In accordance with the
mentioned theorem, it is su cient to prove that functor F holds two following
properties
1. for any L 2 LinStructC , there exists
2 SchedC such that F is
isomorphic to L ;
is bijective.
2. for any 0; 00 2 SchedC the mapping</p>
      <p>h 0; k; 00i 2 SchedC 0; 00 7! F h 0; k; 00i 2 LinStructC F 0; F 00
Lemma and the method of constructing the functor F guarantee the validity of
the first property.
count that for each n 2 N, there exists c 2 C such hc; ni 2 I
The mapping in the second property is injective. Indeed, if F h 0; k1; 00i =
F h 0; k2; 00i then hc; k1(n)i = hc; k2(n)i for all hc; ni 2 I 0 . Taking into
ac</p>
      <p>The mapping in the second property is surjective. Really, if f is a morphism
from L
0 into L</p>
      <p>00 then f hc; ni = hc; k f (n)i . The occurrence of the same c on
both sides of the equation is caused that f is a morphism. Reasoning as above
we obtain the function k f : N ! N . It is evident that</p>
      <p>
        I 0 [n] = nc 2 C j hc; ni 2 I
Summing up the above, one can conclude that the approach based on the
category theory is more expressive than the approach based on the set theory.
Systematic using this approach we have established the character of the
relationship between denotational and operational approaches to modelling logical time
based on the concept of logical clocks. This relationship, as it has been shown
(see Main Theorem), is an equivalence of the corresponding categories. This
equivalence explains the equivalence between the denotational and operational
semantics for some subset of Clock Constraint Specification Language (CCSL)
called RCCSL [
        <xref ref-type="bibr" rid="ref6">6</xref>
        ].
      </p>
      <p>
        At the same time, it should be mentioned that the results presented above
do not give an exhaustive description of the categories introduced in [
        <xref ref-type="bibr" rid="ref2">2</xref>
        ] and
this paper. Among the problems being posed by this and previous papers are the
following.
1. What does a morphism of schedules mean informally or, in other words, how
are relations between schedules established by morphisms understanding
informally?
2. Does the equivalence of categories established above ensure the equivalence
of the denotational and operational semantics of CCSL?
3. Does the category-theoretic approach provide methods to compose more
complex systems using less complex systems or, in other words, what
category-theoretic constructions are realised in the introduced categories?
4. Does the theoretic-category approach give a general model theory for logical
time modelling based on the concept of logical clocks?
      </p>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          1.
          <string-name>
            <surname>Zholtkevych</surname>
            ,
            <given-names>G.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>El Zein</surname>
            ,
            <given-names>H. K.</given-names>
          </string-name>
          :
          <article-title>Two approaches to modelling logical time in cyber-physical systems</article-title>
          .
          <source>In N. Bassiliades</source>
          , et al., eds.: Information and Communication Technologies in Education, Research, and
          <string-name>
            <given-names>Industrial</given-names>
            <surname>Applications</surname>
          </string-name>
          . Volume
          <volume>826</volume>
          of CCIS., Springer, Cham (
          <year>2018</year>
          )
          <fpage>21</fpage>
          -
          <lpage>40</lpage>
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          2.
          <string-name>
            <surname>Zholtkevych</surname>
            ,
            <given-names>G.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>El Zein</surname>
            ,
            <given-names>H. K.</given-names>
          </string-name>
          :
          <article-title>Logical time models to study cyber-physical systems</article-title>
          . In V.
          <string-name>
            <surname>Ermolayev</surname>
            , et al., eds.: Information and Communication Technologies in Education, Research, and
            <given-names>Industrial</given-names>
          </string-name>
          <string-name>
            <surname>Applications</surname>
            . Integration, Harmonization and
            <given-names>Knowledge</given-names>
          </string-name>
          <string-name>
            <surname>Transfer</surname>
          </string-name>
          . Volume 1844 of Workshop Proceedings.,
          <string-name>
            <surname>CEUR-WS</surname>
          </string-name>
          (May
          <year>2017</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          3. Andre´,
          <string-name>
            <given-names>C.</given-names>
            ,
            <surname>Mallet</surname>
          </string-name>
          <string-name>
            <surname>F.</surname>
          </string-name>
          :
          <article-title>Clock constraints in UML MARTE CCSL</article-title>
          .
          <source>Research Report RR-6540</source>
          , INRIA (
          <year>2008</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          4.
          <string-name>
            <given-names>Mac</given-names>
            <surname>Lane</surname>
          </string-name>
          ,
          <string-name>
            <surname>S.</surname>
          </string-name>
          :
          <article-title>Categories for the Working Mathematicians</article-title>
          .
          <source>2nd edn</source>
          . Springer-Verlag (
          <year>1998</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          5.
          <string-name>
            <surname>Novikov</surname>
            ,
            <given-names>B.</given-names>
          </string-name>
          :
          <source>Lecture Notes on Category Theory. Luhansk</source>
          National Pedagogical University (
          <year>2004</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          6.
          <string-name>
            <surname>Zholtkevych</surname>
          </string-name>
          , G.,
          <article-title>other: Two semantic models for clock relations in the clock constraint specification language</article-title>
          . In V.
          <string-name>
            <surname>Ermolayev</surname>
            , et al., eds.: Information and Communication Technologies in Education, Research, and
            <given-names>Industrial</given-names>
          </string-name>
          <string-name>
            <surname>Applications</surname>
          </string-name>
          . Volume
          <volume>412</volume>
          of CCIS., Springer, Cham (
          <year>2013</year>
          )
          <fpage>190</fpage>
          -
          <lpage>209</lpage>
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>