Lecture CIMPA School

Some Applications of the Theory of
Representation of Algebras;
CIMPA School;
Homological Methods in the Theory of
Representations and Its Applications
Agustı́n Moreno Cañadas and Isaı́as David Marı́n Gaviria

National University of Colombia
06/18/2018
Contents
INTRODUCTION 4
Preliminaries 5
1 Preliminaries 6
1.1 Posets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Representations of Posets . . . . . . . . . . . . . . . . . . . . 8
1.3 Algorithms of Differentiation . . . . . . . . . . . . . . . . . . . 10
1.3.1 Some indecomposable objects . . . . . . . . . . . . . . 16
1.3.2 Tiled Orders . . . . . . . . . . . . . . . . . . . . . . . . 22
1.4 Quiver Representations . . . . . . . . . . . . . . . . . . . . . . 28
1.4.1 Auslander-Reiten Quiver . . . . . . . . . . . . . . . . . 29
1.5 Sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2 Some Applications of the Theory of Representation of Posets

and Quivers 34
2.1 On The Number of Antichains in Pn . . . . . . . . . . . . . . 34
2.2 Energy, The Drozd’s Theorem and the Vietá’s Formula . . . . 37
2.2.1 Cluster Algebras . . . . . . . . . . . . . . . . . . . . . 41
2.2.2 Snake Graphs, Continued Fractions and Cluster Algebras 44
2.2.3 Palindromic Snake Graphs and Sums of Two Squares . 46
2.3 Partitions into Triangular Numbers . . . . . . . . . . . . . . . 47
2.4 The Cryptosystem Γ-Pentagonal . . . . . . . . . . . . . . . . . 54
2.5 Visual Cryptography . . . . . . . . . . . . . . . . . . . . . . . 56
2.5.1 Lattice-Based VSSS . . . . . . . . . . . . . . . . . . . . 58
2.5.2 A Matrix Problem Induced by a Lattice-Based VSSS . 59
2.6 Generation of Emerging images Via Tiled Orders . . . . . . . 63
2.7 A Mathematical Model for Emerging and Multistable Images . 67
2
2.8 Digital Watermarking . . . . . . . . . . . . . . . . . . . . . . . 69
2.9 Homological Persistence Theory . . . . . . . . . . . . . . . . . 73
3 Appendix 79
3.0.1 DVII in MATLAB . . . . . . . . . . . . . . . . . . . . 79
3.0.2 Wavelet Transformation and Singular Value Decompo-
sition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
REFERENCES 84
3
Introduction
The impact of the theory of representation of algebras and in particular of

the theory of representation of posets in different fields of the mathematics is
remarkable, applications of these theories can be found out in combinatorics,
number theory and differential equations.
Categorification of integer sequences is an example of the application of the

theory of representation of algebras in diophantine analysis, this term was
coined by Ringel and Fahr to the process of interpreting numbers in integer or
real sequences as invariant of objects in a determined category. Ringel and
Fahr used the Auslander-Reiten quiver of the 3-Kronecker quiver in order
to define some partitions for Fibonacci numbers, whereas, Fernandéz et al
[13–16] used the preprojective component of the Auslander-Reiten quiver
of the 2-Kronecker quiver to categorify the integer sequence A052558 in
the OEIS (On-Line Encyclopedia of Integer Sequences). Actually, recently
Ringel, Krausse, Ingall, Thomas et al have pay much attention to the cate-
gorification of non-crossing partitions and its relationship with cluster alge-
bras. We also recall works of Assem et al Fomin, Zelevinsky et al, and Schiffler
et al who have given nice relationships between the theory of representation
of algebras in particular cluster algebras with triangulations, integer friezes,
continued fractions and perfect matchings [2, 3, 11, 36–38, 41, 52, 55–57, 62].
The Auslander-Reiten quiver of quivers 2-Kronecker and 3-Kronecker can

also be used to solve linear systems of differential equations (i.e, systems of
the form A dx + Bx = f (t) where both A and B are m × n-matrices) and the
dt
2 2 2 √
Shcrödinger equation ∂∂xψ2 + ∂∂yψ2 + ∂∂zψ2 = ( 2mc)2 ψ [35].
Recently the theory of homological persistence has arose as a new source of

applications of the theory of representation of quivers in computer science
more specifically in topological data analysis and big data analytics. In
homological persistence theory the homology functor is applied to a filtration
of topological spaces generating almost in a natural way a representation of
a Dynkin quiver of type An and some of its indecomposable representations
are interpreted as persistence modules determining when a given data born
or die [10, 12, 28].
4
In this notes, we intend to describe some applications of the theory of repre-
sentation of posets and quivers in number theory, combinatorics and cryptog-
raphy, in number theory we describe how indecomposable projective modules
of some path algebras give information on the number of partitions of an inte-
ger number into three triangular numbers, in particular, the Vietá’s formula
arise from the spectrum of Dynkin diagrams of type An and Dn and the
Euler’s counterexample on the primality of the Fermat number F5 arises as
mutations of clusters of a particular cluster algebra [37].
On the other hand, the theory of representation of posets was used by Zavad-
skij and Kirichenko to classify tiled orders or semimaximal rings [77, 78] such
tiled orders are helpful to determine the number of two-point antichains in
the powerset Pn of the set {1, 2, . . . , n} as a particular case of the famous
Dedekind’s problem which consists of enumerating the number of antichains
in Pn . Besides, 01-tiled orders can be used to construct emerging images,
this kind of images have been used in different CAPTCHAs (Completely Au-
tomated Turing Test to Tell Humans and Machines Apart), recently it has
been constructed emerging videos which are the based of Nu-CAPTCHA
which is one of the more strongest CAPTCHA developed up to date. We
also describe a relationship between some aspects of the visual cryptography
with some matrix problems.
5
Chapter 1
Preliminaries
In this chapter we describe the theoretical setting necessary to describe ap-

plications of the theory of representation of algebras and posets.
1.1 Posets
An ordered set (or partially ordered set or poset) is an ordered pair of the
form (P, ≤) of a set P and a binary relation ≤ contained in P × P, called
the order (or the partial order ) on P, such that ≤ is reflexive, antisymmetric
and transitive [25]. The elements of P are called the points of the ordered
set. We will write x < y for x ≤ y and x 6= y, in this case we will say x is
strictly less than y. An ordered set will be called finite (infinite) if and only
if the underlying set is finite (infinite). Usually we shall be a little slovenly
and say simply P is an ordered set. Where it is necessary to specify the order
relation overtly we write (P, ≤).
Let P be an ordered set and let x, y ∈ P we say x is covered by y if x < y

and x ≤ z < y implies z = x.
Let P be a finite ordered set. We can represent P by a configuration of

circles (representing the elements of P) and interconnecting lines (indicating
the covering relation). The construction goes as follows.
(1) To each point x ∈ P, associate a point p(x) of the Euclidean plane R2 ,

depicted by a small circle with center at p(x).
6
(2) For each covering pair x < y in P, take a line segment l(x, y) joining the
circle at p(x) to the circle at p(y).
(3) Carry out (1) and (2) in such a way that
(a) if x < y, then p(x) is lower than p(y),

(b) the circle at p(z) does not intersect the line segment l(x, y) if z 6= x
and z 6= y.
A configuration satisfying (1)-(3) is called a Hasse diagram or diagram of P.

In the other direction, a diagram may be used to define a finite ordered set;
an example is given below, for the ordered set P = {a, b, c, d, e, f }, in which
a < b < c < d < e, and f < c.
d e
d d
dc
@
d @db
f @
@d a
Figure 1
We have only defined diagrams for finite ordered sets. It is not possible to
represent the whole of an infinite ordered set by a diagram, but if its structure
is sufficiently regular it can often be suggested diagrammatically. Of course,
the same ordered set may have different diagrams. Diagram-drawing is as
much an art as a science, and, as we will see, good diagrams can be a real
asset to understanding and to theorem-proving.
An ordered set C is called a chain (or a totally ordered set or a linearly

ordered set) if and only if for all p, q ∈ C we have p ≤ q or q ≤ p (i.e., p and
q are comparable). If any two of points of C are incomparable then C is said
to be an antichain [1, 25, 63]).
Perhaps one of the most interesting problems regarding antichains was pro-
posed by Dedekind in 1897 in his famous Über Zerlegungen von Zahlen durch
ihre größten gemeinsamen Teiler (About breakdowns of numbers by their
7
greatest common divisor). The nth Dedekind number D(n) is given by the
number of monotone boolean functions on n variables or by counting the
number of antichains in the powerset 2n ordered by inclusion [72].
A formula for D(n) for any n has not been found up to date. Actually,
only the following eight Dedekind numbers are known at the present time
(A000372 in OEIS), [79]:
D(0) = 2
D(1) = 3
D(2) = 6
D(3) = 20
D(4) = 168 (1.1.1)
D(5) = 7, 581
D(6) = 7, 828, 354
D(7) = 2, 414, 682, 040, 998
D(8) = 56, 130, 437, 228, 687, 557, 907, 788.
1.2 Representations of Posets

The theory of representations of posets was developed at the early 1970s by
Nazarova, Roiter and their students in Kiev. The main aim of their study
was to determine indecomposable objects of the additive category rep P of
representations of a given poset (P, ) [1, 31, 44, 66]. In this case, an object
U ∈ rep P has the form:
U = (U0 ; Ux | x ∈ P) (1.2.2)
where U0 is a finite dimensional k-vector space, Ux is a subspace of U0 for

x ∈ P, and Ux ⊆ Uy provided x y. The direct sum of U and V is defined
by formula
U ⊕ V = (U0 ⊕ V0 ; Ux ⊕ Vx | x ∈ P).
8
A representation U ∈ rep P is said to be indecomposable if U 6= 0 and is not
a direct sum of two non-zero representations.
A map f : U → V ∈ rep P between representations U, V ∈ rep P is a k-linear

map f : U0 → V0 such that f (Ux ) ⊆ Vx for all x ∈ P [66]. Often, the additive
category of representations of a given poset P is denoted rep P. If Ind P
denotes a set of representatives of the isomorphism classes of indecomposable
objects of rep P then the poset P is said to be of finite representation type if
the set Ind P is finite. Thus, the main problem regarding representation of
posets consists of giving a complete description of the isomorphic classes of
indecomposable representations of a category rep P for a given poset P.
Attached to each representation U there exists its matrix presentation M =

MU choosing some basis B0 in U0 and for each x ∈ P, some P system Bx of
linearly independent generators of Ux modulo rad Ux = U x = Uy . Then
yx
M = Mx1 Mx2 ... Mxn (2)
with entries in k, partitioned horizontally into n = |P| (vertical) blocks (also

called strips) [44, 66].
Two representations A and B of a poset (P, ) are isomorphic if and only

if their matrix presentations can be turned into each other with help of the
following admissible transformations:
1. elementary transformations of rows of the whole matrix A;
2. elementary transformations of columns within each vertical strip;
3. additions of columns of a strip Ai to columns of a strip Aj if i j in

P.
For example if
P= d d d
a1 a2 a3
9
is a poset consisting of three incomparable elements, then the following is a
complete list of indecomposable representations:
1 0 1
k(ai ), k(ai , aj ), k(a1 , a2 , a3 ), and U3a = , where i ∈ {1, 2, 3}, and
0 1 1
i < j.
1.3 Algorithms of Differentiation

Up to date the main tools to classify posets have been the algorithms of
differentiation. Such algorithms are functors giving a relationship between
the category of representations of a given poset P and the category of rep-
resentations of a corresponding derived poset P0 . Actually, the main criteria
to classify ordinary posets of finite, tame and finite growth type were found
by Kleiner in 1972, Nazarova in 1975, and Nazarova and Zavadskij in 1981
respectively, with the help of the algorithm of differentiation with respect
to a maximal point introduced by Nazarova and Roiter in 1972, and the
algorithm of differentiation I introduced by Zavadskij in 1977 [32, 51, 73, 76].
The following is the definition of the algorithm of differentiation with respect

to a maximal point given by Nazarova and Roiter in 1972:
Let (P, ≤) be an ordinary poset then a maximal point x ∈ max P is suitable

for this differentiation if the subset N ⊂ P of all points n ∈ P incomparable
with x has width w(N ) ≤ 2.
The algorithm of differentiation with respect to a suitable point b ∈ max P

is defined in such a way that if N = P \ bM then
P0b = (P \ b) ∪ N
b
with a partial order induced by P \ b and N

b = N ∪ {x + y | x, y ∈ N, x 6≤
y, y 6≤ x}. Figure 3 below shows the Hasse diagram of this differentiation.
10
2+5 d
d5 db
@
d @d
2+4 5
2 d
@
2 d 0 @d
@ d4 d7 −→ @
1+4
d7
1 d @ b @ d1+3 @@ d
4
1 d
@ @
@d d6 @d d6
3 3
P = N + {6 < 7 < b} P0b = N

b + {6 < 7}
Figure 2
The functor of differentiation F b with respect to a maximal point b is defined

in such a way that if U = (U0 ; Ux | x ∈ P) is a k-linear representation
of a poset P suitable for this differentiation then the object F b (U ) = U 0 =
(U00 ; Ux0 | x ∈ P0b ) of P0b is defined by the following formulae:
U00 = Ub ,
Ux0 = Uxb for x ∈ P\ b,
0 (1.3.4)
Ux+y = U(x+y)b for each diad {x, y} ⊂ N,
ϕ0 = ϕ | Ub for any linear map-morphism ϕ : U0 → V0 ∈ rep P.
The following theorem was proved by Gabriel in 1973 [32].
Theorem 1.1. The functor of differentiation with respect to a maximal point
F b : rep P → repP0b ,
induces an equivalence between quotient categories

∼
rep P/hInd N i → rep P0b
and mutually inverse bijections between classes of indecomposable objects
Ind P \ [N ] Ind P0b .
in particular
|Ind P| = |IndP0b | + |N
b | + 1.
11
Where |Ind P| denotes the number of classes of indecomposable objects in
rep P, |N
b | is the size of N
b and hInd N i denotes the ideal consisting of all
morphisms passing through sums of indecomposable objects defined by the
subset N , in this case rep P/hInd N i is a quotient category [31, 44, 66].
These results allowed to Kleiner to give the following finite type representa-
tion criterion:
Theorem 1.2. A finite poset P is of finite representation type if and only if
P 6⊃ K1 , . . . , K5 where
d
d d
d d d d
d d d d d d d d d d
d d d d d d d d d d d d d d d d
K1 K2 K3 K4 K5
Figure 3
Often, posets K1 , . . . , K5 are called the Kleiner’s critical.
Theorem 1.1 was also used by Fernandéz et al to categorify the sequence

A002662 in the OEIS.
The algorithm of differentiation with respect to a suitable pair of points

(DI) was introduced by Zavadskij in 1977. Such an algorithm allows him
and Nazarova to give a criterion to classify ordinary posets of finite growth
representation type [51, 73, 76].
A pair of points (a, b) in a poset P is suitable for differentiation DI if poset

P can be written as follows:
P = aO + C + b M
where C = c1 < c2 < · · · < cn is a chain and for each i, 1 ≤ i ≤ n, points

a, b and ci are mutually incomparable. In such a case the derived poset P0(a,b)
with respect to the pair (a, b) is such that:
12
P0(a,b) = P\{c1 , c2 , . . . , cn } + {a + c1 , a + c2 , . . . , a + cn } + {c1 b, c2 b, . . . , cn b},
being considered as a subposet of the free lattice generated by P, we write

−
a + ci = c+ i and ci b = ci . The following is the Hasse diagram for this
differentiation.
c+
n d bd
db @
@d

d d
cn c+ c−
@ n
d 2
d @d
0 c+
−→
@
d
1d @d
c2 (a,b)
c−
@ 2
d d d @d

a c1 a c−
1
Figure 4
U0 -subspaces in a representation of the derived poset P0(a,b) are given by the

following identities:
U00 = U0
Uc0+ = Ua+ci = Ua + Uci
i
(1.3.5)
Uc0− = Uci b = Uci ∩ Ub
i
Ux0 = Ux for the remaining points x ∈ P0
DI allowed to Zavadskij and Nazarova to give a finite growth representation

type criterion for ordinary posets. A similar criterion was found by Zavadskij
by using the algorithm of differentiation VII for equipped posets (with weak
⊗ and strong points ◦ and two kind of relations (weak or strong) between
its points) whose representations are systems of G-subspaces of a fixed F -
subspace U0 where G is a quadratic extension of F [74, 75]. In this case the
following condition must be satisfied:
x ≤ y =⇒ Ux ⊂ Uy
x y =⇒ F (Ux ) ⊂ Uy .
13
where x y denotes a strong relation between points x and y in P whereas
x y denotes a weak relation between them. And F (Ux ) is the hull of the
subspace Ux obtained from Ux via a natural complexification process.
As an example, we consider the equipped poset P whose diagram is illustrated

in Figure 5 where {a, b, d} ⊂ P⊗ , c ∈ P◦ , {a, b} c, and b ≺ d.
dc d⊗
@
@
@
P= @
@
@
@
⊗ @
⊗
a b
Figure 5
We define a representation U for P over the pair (R, C), in such a way
that the set of vectors {e1 , e2 , e3 , e4 , e5 } is a basis for the R-space, U0 = R5 .
Furthermore;
Ua = C{e2 + ie3 },
Ub = C{e4 + ie5 },
(1.3.6)
Uc = C5 = U
f0 ,
Ud = C{e1 , e4 + ie5 }.
Each equipped poset P naturally defines a matrix problem of mixed type over
the pair (F, G). Consider a rectangular matrix M separated into vertical
stripes Mx , x ∈ P, with Mx being over F (over G) if the point x is strong
(weak):
x -y
⊗ ⊗ e e
M= G G F F
Such partitioned matrices M are called matrix representations of P over

(F, G). Their admissible transformations are as follows:
14
(a) F -elementary row transformations of the whole matrix M ;
(b) F -elementary (G-elementary) column transformations of a stripe Mx if
the point x is strong (weak);
(c) In the case of a weak relation x ≺ y, additions of columns of the stripe
Mx to the columns of the stripe My with coefficients in G;
(d) In the case of a strong relation x y, independent additions both real
and imaginary parts of columns of the stripe Mx to real and imaginary
parts (in any combinations) of columns of the stripe My with coefficients
in F (assuming that, for y strong, there are no additions to the zero
imaginary part of My ).
Two representations are said to be equivalent or isomorphic if they can be
turned into each other with help of the admissible transformations. The
corresponding matrix problem of mixed type over the pair (F, G) consists of
giving of all its indecomposable representations up to equivalence.
Remark 1.3. The matrix problem for representations (a)-(d) occurs natu-
rally in the classification of the objects U ∈ rep P up to isomorphisms. In
this case, it is associated to the representation U its matrix presentation
MU = (Mx ; x ∈ P) defined as follows:
If a point x ∈ P◦ (P⊗ ) then the columns of the stripe Mx consist of coordi-
nates (with respect to a fixed ordered basis B of U0 ) of a system of generators
G of Ux+ (respectively, G-subspace Ux ) modulo its radical subspace Ux+ (re-
spectively, Ux ). Problem (a)-(d) may be obtained by changing basis B and
the system of generators G.
For example the representation (1.3.6) is a sincere representation of the poset

P described in Figure 5 with an associated matrix representation of the form:
a b c d
1 1 e1
1 1 e2
MU = i e3
1 1 e4
i e5
In this case, each empty cell indicates a null coordinate with respect to the
15
ordered basis {e1 , . . . , e5 } of the vector space U0 .
1.3.1 Some indecomposable objects

In this section, we give some examples of indecomposable objects in the
category rep P, where P is an equipped poset.
If P is an equipped poset and A ⊂ P then P (A) = P (min A) = (F ; Px |

x ∈ P), Px = G if x ∈ A∨ and Px = 0 otherwise. In particular, P (∅) =
(F ; 0, . . . , 0).
If a, b ∈ P⊗ then T (a) and T (a, b) denote indecomposable objects with matrix

representation of the following form:
a a b
T (a) = 1 , a ∈ P⊗ , T (a, b) = 1 0 , with a ≺ b.
u u 1
If we consider the notation (1.2.2) for objects in rep P then the object T (a)
may be described in such a way that, T (a) = (T0 ; Tx | x ∈ P), where
T0 = F 2 and

2
T0 = G ,
 e if x ∈ aO ,
Tx = G {(1, u)t } , if x ∈ ag ,

0, otherwise.

where (1, u)t is the column of coordinates with respect to an ordered basis
of T0 .
On the other hand, representation T (a, b) may be described in such a way

that T (a, b) = (T0 ; Tx | x ∈ P), where T0 = F 2 and

t
G{(1, u) }, if a x ≺ b,

Tx = Te0 = G2 , if x ∈ aO ∪ b∨ ,

0, otherwise.

If a ∈ P⊗ and B ⊂ P is a subset completely weak such that a ≺ B then

we let T (a, B) denote the representation of P which satisfies the following
conditions with T0 = F 2 :
16

t
G{(1, u) },
 if x ∈ ag \ B,
Tx = Te0 = G2 , if x ∈ aO + B ∨ ,

0, otherwise.

In particular, T (a, ∅) = T (a).
The definitions given above allow us to write the matrix representation (1.3.6)
of the equipped poset described in Figure 5 as the following direct sum of
indecomposable objects:
a b c d
1 1 e1
1 1 e2
T (a) ⊕ T (b) ⊕ P (c, d) =MU = i e3
1 1 e4
i e5
Remark 1.4. In [74] it is proved that P (∅), P (ci ), T (ci ) and T (ci , cj ),
for 1 ≤ i < j ≤ n are the only indecomposable representations (up to
isomorphisms) over the pair (R, C) of a completely weak chain C = {c1 ≺
· · · ≺ cn }. In fact, if U = (U0 ; Uci | 1 ≤ i ≤ n) is a representation of C
over (R, C) then in the corresponding matrix representation each block Uci ,
1 ≤ i ≤ n, may be reduced via admissible transformations to the following
standard form:
I
I
Uci =
iI
where the columns consist of generators of Uci modulo its radical subspace
Uci = Uci−1 with respect to a fixed basis of U0 (In this case, empty cells
indicate null coordinates). This result can be generalized in a natural way
to the case (F, G) by using a suitable scalar u ∈ G instead of the constant
i ∈ C in the matrix presentation of Uci showed above.
17
The following is the description of the algorithm of differentiation VII:
A pair of incomparable points (a, b) in an equipped poset P is said to be
VII-suitable, if a ∈ P⊗ , b ∈ P◦ and
P = aO + bM + {a ≺ c1 ≺ · · · ≺ cn },
where {a ≺ c1 ≺ · · · ≺ cn } is a completely weak chain incomparable with

b, we can assume n ≥ 0 by setting c0 = a. We let C denote the set
{c1 ≺ · · · ≺ cn }.
If P is an equipped poset with a pair of points (a, b), VII-suitable then

the derived poset, P0 is an equipped poset with the following shape:
P0(a,b) = P \ {a + C} + {a− < a+ } + C − + C + ,
where a− ∈ (P0 )⊗ , a+ ∈ (P0 )◦ , C − and C + are completely

− weak+ chains with
− − − +
+
+
C = c1 ≺ · · · ≺ cn and C = c1 ≺ · · · ≺ cn , ci ≺ ci for each i.
Furthermore, a− ≺ c− < c+ 1 , cn < b. Points in the derived poset P
+ − 0
1, a
satisfy the following conditions:
(1) Points a− , a+ , c− +
i y ci inherit the relations that points a and ci had in
the original poset P \ {a + C}.
(2) Order relations in P \ {a + C}, together with the relations described

above induce the order relations in P0(a,b) (in particular, a− ≺ c−
n ).
Figure 6 shows the diagrams of P and the corresponding P0(a,b) .

P P0(a,b) c+ db
⊗ db n
⊗
cn
⊗− B
@
A ⊗ B A ⊗
−VII
+
cn
⊗ −→ c1 @
⊗
(a,b) ⊗
c1
⊗−
@
⊗ a+ d
a c1
⊗−
@
@
Figure 6 a
In Diagram a double line denotes strong relations between all points in the
up-cone A and the weak point a, this notation is not necessary in case of
relations between a strong point (b for instance) and any point of an equipped
poset.
18
The functor of differentiation D(a,b) : rep P −→ rep P0(a,b) = 0 : rep P −→rep
P0 (a,b) , is defined in such a way that for U 0 = D(a,b) (U ) ∈ rep P0(a,b) , if 1 ≤ i ≤
n then it holds that
U00 = U0 ,
Ua0 − = Ua ∩ Ub ,
Ua0 + = F (Ua ),
Uc0− = Uci ∩ Ub , (1.3.7)
i
Uc0+ = Uci + F (Ua ),

i
Ux0 = Ux for the remaining points x ∈ P0(a,b) ,

0
ϕ = ϕ, for any linear-map morphism ϕ : U0 −→ V0 .
For instance, P 0 (a) = P (a+ ), T 0 (a) = T 0 (a, ci ) = P 2 (a+ ).
The process of integration aims to recover a representation of the original

poset from the derived one.
A representation of an equipped poset P without direct summands of type

T (a), T (a, ci ) or P (a) is named reduced. A representation of the derived
poset P0(a,b) without direct summands of the form P (a+ ) is called reduced as
well [74].
If (E0 , W0 ) is a (Ua+ , Ub+ )-cleaving pair of the space U0 then U ↓ is the repre-
sentation of the reduced derived poset which satisfy the condition
U 0 = U ↓ ⊕ P m (a+ ),
where m = dim(Ua+ + Ub+ )/Ub+ which has the form U ↓ = (W0 ; Wx | x ∈ P0 ),

donde Wx = Ux0 ∩ Wf0 . U ↓ is independent up to isomorphisms of spaces E0
and W0 due the inclusion Wa+ ⊂ Wb .
As an example, for the poset P = ag + bM = {a, b, c, d} with a ≺ c, d b,

d c, a, c ∈ P⊗ , b, d ∈ P◦ , we define a representation U 0 ∈ rep P0(a,b) over the
19
pair of fields (R, C) in such a way that:
U00 = R8 = R{et | 1 ≤ t ≤ 8},
Ua0 + = C{ei | 1 ≤ i ≤ 6},
Ub0 = C{ek | 5 ≤ k ≤ 8},
(1.3.8)
Ua0 − = 0,
Uc0+ =Uf0 ,
Uc0− = Ud0 = C{e8 },
with U0 = E0 ⊕ W0 , E0 = R{ei | 1 ≤ i ≤ 4}, W0 = R{ei | 5 ≤ i ≤ 8}.
Then the representation W = U ↓ is such that :
W0 = R{ei | 5 ≤ t ≤ 8},
Wa+ = C{e5 , e6 },
Wb = W
f0 ,
(1.3.9)
Wa− = 0,
Wc+ = Wb = Wc+ ⊕ C{e7 },
Wc− = Wd = C{e8 }.
Therefore, if U ∈ rep P has an associated matrix presentation MU with the
form:
a cxd b
i
i
E0
1 1
i
MU =
1 1
1 1
Ub = W0
1 1
1
+x
a+ c d b
then U 0 ' U ↓ ⊕ P 4 (a+ ). Note that, blank cells denote null coordinates of
vectors with respect to a fixed basis of the space U0 .
The following is the main result regarding differentiation VII.
20
Theorem 1.5. Let P be an equipped poset with a pair of points VII-suitable
(a, b). Then the differentiation functor
D(a,b) : rep P −→ rep P0(a,b) ,
induces a categorical equivalence between quotient categories
∼
R/I → R0 /I0
where R = rep P, R0 = rep P0(a,b) , and
I = hP (a), T (a), T (a, ci )iR , I0 = hP (a+ )iR0 .
Corollary 1.6. If Γ(R) y Γ(R0 ) are the Gabriel’s quivers of R and R0 then
Γ(R) \ [P (a), T (a), T (a, ci ) | 1 ≤ i ≤ n] ' Γ(R0 ) \ [P (a+ )].
The following is the matrix presentation of the differentiation VII and its
corresponding process of integration.
a A c1 cn B b
iE
E E0
iE E
E
iE E
∗ X1 Xn Q
W0
∗ G ∗ ∗ F1 Y1 ∗ Fn Yn ∗ ∗ Ub
a −
c−
1 c−
n
a+ c+
1 c+
n
Methods of poset representation theory were also successfully applied to clas-

sify different type of rings. For instance, critical posets obtained by Kleiner
were used by Zavadskij and Kirichenko in order to give a criterion for tiled
orders of finite representation type [77, 78], and Arnold described some re-
lationships between Butler groups and representations of finite posets over
discrete valuation rings [1]. According to Arnold computations of represen-
tation type of representations of posets over discrete valuation rings lead to
explicit procedures for constructing indecomposable almost completely de-
composable groups of arbitrarily large finite rank with types in a fixed finite
set of types. Such constructions are quite difficult without the assistance of
techniques from representation theory.
21
1.3.2 Tiled Orders
A field T is said to be of discrete norm or discrete valuation if it is endowed
with a surjective map [77, 78]
ν : T → Z ∪ {∞},
which satisfies the following conditions :
(a) ν(x) = ∞ if and only if x = 0,
(b) ν(xy) = ν(x) + ν(y),
(c) ν(x + y) ≥ min{ν(x), ν(y)}.
We let O denote, the normalization ring of the field T , such that
O = {x ∈ T | ν(x) ≥ 0}.
An element π ∈ O such that ν(π) = 1 is a prime element of O. For each

x ∈ O we have that x ∈ O if and only if x = επ m , for some m ≥ 0 and
ε ∈ O∗ . Moreover, x ∈ T if and only if x = επ m for some m ∈ Z and ε ∈ O∗ .
Ring O is such that O ⊃ πO, where πO is the unique maximal ideal, therefore
ideals of O generate a chain of the form
O ⊃ πO ⊃ π 2 O ⊃ · · · ⊃ π m O ⊃ . . .
A tiled order or semimaximal ring Λ is a subring of the matrix algebra T n×n

with the form
 
O π λ12 O . . . π λ1n O
n  π λ21 O O . . . π λ2n O 
λij
P
Λ= eij π O =  .
 
.. .. .. ..
i,j=1  . . . . 
λn1 λn2
π O π O ... O
Λ consists of all matrices whose entries ij belong to π λij O, in this case the
eij ∈ T n×n are unit matrices such that eij ekl = δjk eil (δjk = 1, if j = k,
δjk = 0 otherwise). Numbers λij are integers which satisfy the following
conditions:
22
1. λii = 0, for each i,
2. λij + λjk ≥ λik for all i, j, k.
An order Λ is said to be Morita reduced or reduced if it satisfies the additional

condition:
(3) λij + λji > 0, for each i 6= j. In such a case, projective modules are
pairwise non-isomorphic, that is, in the decomposition of Λ = P1 ⊕ P2 ⊕
· · · ⊕ Pn via projective modules (i.e., the rows of Λ) all indecomposable
projective summands are pairwise not isomorphic, i.e., Pi 6' Pj if i 6= j.
The main problem for tiled orders consists of describing all finitely generated
torsionless Λ-modules which are called admissible modules.
A Λ-admissible right module (not null) is said to be irreducible if it is a

submodule of the unique simple module (up to isomorphism). For instance,
any indecomposable projective module Pi is irreducible. Thus,
Pi = (π λi1 O, π λi2 O, . . . , π λin O),
is a finitely generated irreducible Λ-module without O-torsion.
We denote Λ = (λij )i,j=1...n , note that Λ ⊂ T n×n = Q = Λ ⊗O T , where Q is

the rational hull of Λ, Rad Q = 0 and Λ has an unique right simple T -module
Pn
(up to isomorphism) denoted SR = (T, T, . . . , T ) = ei T , {ei | 1 ≤ i ≤ n}
i=1
is the standard basis such that ei ejk = δij ek . We assume the notation SL =
(T, T, . . . , T )t for left modules.
Any irreducible right Λ-module A has the form
A = (π α1 O, π α2 O, . . . , π αn O),
where αi + λij ≥ αj , αi ∈ Z, 1 ≤ i ≤ n. If A is a left module then we have

that λij + αj ≥ αi .
Henceforth, we denote a right (left) module A in the form A = (α1 , α2 , . . . , αn )

((α1 , α2 , . . . , αn )t , respectively).
Note that, A ' A0 if and only if αi = αi0 + k, for some k ∈ Z and any
1 ≤ i ≤ n.
23
Irreducible right modules which are contained in a Q-simple module of a Λ
order constitute a lattice denoted LR (Λ).
We denote P(Λ) = PR (Λ) the subposet of LR (Λ) which consists of irreducible
projective modules the projective modules Pi are called principals where
Pi = (λi1 , λi2 , . . . , λin ) = Pi0 , Pi0 ∈ PR (Λ)
In this poset there are as many projective modules as infinite chains. In such
a case, modules of the form
Pik = (λi1 + k, . . . , λin + k), k ∈ Z
are projective modules isomorphic to Pi0 . Therefore:
P(Λ) = {Pik | 1 ≤ i ≤ n}
where
(
k − l ≥ λij , PL (Λ),
Pik ≤ Pjl if and only if
k − l ≥ λji , PR (Λ).
Thus, the poset P(Λ) is infinite, periodic and the sum of n chains with the
form {Pik | 1 ≤ i ≤ n, k ∈ Z}, with width w (P(Λ)) ≤ n. For instance, the
tiled order
   
0 2 2 O π2O π2O
Λ =  2 0 2  =  π 2 O O π 2 O ,
2 2 0 π2O π2O O
has associated the following poset of projective modules:
.. .. ..
. . .
P1−4 d dP3
−4
aS
a
Sa !!
S a " d!!
a ! %
−4
!
P1−3 d S a !" a
! a
Pb
b 2
%
% P −3
!d 3
aS
a
Sa! "
! a b !
" a " d!!
"
aS a ! %
b
−3
! %
P1−2 d! %a P −2
! a
"SS Sa "a!a Pb
b 2 %
ba
bd 3
aS
a
Sa! "
!S! %b
a %!!
" a S d%!
"
aS a !
!% a%
b
−2 ab
P1−1 d! %a P −1
!
"SS Sa ""a !
!abP
b 2 % bd 3
"
! !S a
% b %
"
"!S S d% −1 %
a b
aa
P10 d! adP30
!
" S P2 % b b
S %
S d% 0
P2
.. .. ..
. 24. .
Figure 7
The map σ : PR (Λ) → PL (Λ), given by σ(Pik ) = Pi−k is a natural poset
antiisomorphism, thus the pair {O, P(Λ)} defines the tiled order Λ up to
isomorphism, in the sense that
0 0 0
Λ ' Λ if and only if {O, P(Λ)} ' {O , P(Λ )}
0 0
That is, O ' O and P(Λ) ' P(Λ ) which means that there exists a poset iso-
0
morphism ϕ : PL (Λ) → PL (Λ ) such that A ' B if and only if ϕ(A) ' ϕ(B).
0
Λ and Λ are Morita-equivalents provided that ϕ preserves isomorphisms.
The following result characterizes isomorphic orders via matrix problems [44]:
0
Theorem 1.7. Two orders Λ and Λ are isomorphic if the corresponding
0
exponents matrices λij and λij can be turner into each other with the help of
the following admissible t-transformations:
1. To add an integer n to each entry of a given row i and simultaneously

subtract n to each entry of the column i.
2. To transpose simultaneously rows i and j and columns i and j.
The following is the finite type representation criterion for tiled orders intro-
duced by Zavadskij and Kirichenko [77, 78].
Theorem 1.8. A tiled order Λ is of finite representation type if and only if

P(Λ) 6⊃ K1 , . . . , K5 .
For m ≥ 1 a (0, 1, 2, . . . , m)-tiled order is a tiled order Λ = (λij ), 1 ≤ i, j ≤ n

where λij ∈ {0, 1, 2, . . . , m}. In particular, if Λ = (λij ) is a (0, m)-tiled order
then Λ has associated a finite poset (N, ≤) = N(Λ) = ({1, 2, . . . , n}, ≤) where
i ≤ j if and only if λij = 0
for instance the following (0, 1)-orders are isomorphic (Actually, for each
n ≥ 3 it is possible to build 01 tiled order of these types Λn and Γn ):
25
0 0 0 0 0 0 0 0 d 7
1 0 1 1 0 0 1 0 @
@
1 1 0 1 0 1 0 0 d4 d5 @ d6
1 1 1 0 1 0 0 0
Λ3 = @
@
@
@
1 1 1 1 0 1 1 0 d1 @ d2 @ d3
1 1 1 1 1 0 1 0 @
@
1 1 1 1 1 1 0 0 @d
1 1 1 1 1 1 1 0 0
N(Λ3 )
0 1 1 1 1 1 1 1
d
0
0 0 1 1 1 1 1 1
@
0 1 0 1 1 1 1 1 @
d1 d2 @ d4
0 0 0 0 1 1 1 1
Γ3 = @ @
0 1 1 1 0 1 1 1 @ @
d3 @ d5 @ d6
0 0 1 1 0 0 1 1
@
0 1 0 1 0 1 0 1 @
@d
0 0 0 0 0 0 0 0 7
N(Γ3 )
Figure 8
26
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 0 1 1 1 0 0 0 1 1 1 0 0 0 1 0
1 1 0 1 1 0 1 1 0 0 1 0 0 1 0 0
1 1 1 0 1 1 0 1 0 1 0 0 1 0 0 0
1 1 1 1 0 1 1 0 1 0 0 1 0 0 0 0
1 1 1 1 1 0 1 1 1 1 1 0 0 1 1 0
1 1 1 1 1 1 0 1 1 1 1 0 1 0 1 0
1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 0
Λ4 =
1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 0
1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0
1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 0
1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0
1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0
1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1
0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1
0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1
0 0 1 1 0 0 1 1 1 1 1 1 1 1 1 1
0 1 0 1 0 1 0 1 1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1
Γ4 =
0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1
0 0 1 1 1 1 1 1 0 0 1 1 1 1 1 1
0 1 0 1 1 1 1 1 0 1 0 1 1 1 1 1
0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1
0 1 1 1 0 1 1 1 0 1 1 1 0 1 1 1
0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1
0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
27
1.4 Quiver Representations
A quiver Q = (Q0 , Q1 , s, t) is a quadruple consisting of two sets Q0 whose
elements are called vertices and Q1 whose elements are called arrows, s and
t are maps s, t : Q1 → Q0 such that if α is an arrow then s(α) is called
the source of α whereas t(α) is called the target of α [2]. A path of length
l ≥ 1 with source a and target b is a sequence (a | α1 , α2 , . . . , αl | b) where
t(αi ) = s(αi+1 ) for any 1 ≤ i < l. Vertices are paths of length 0 [2, 4, 7, 8, 66].
If Q is a quiver and k is an algebraically closed field then the path algebra
kQ of Q is the k-algebra whose underlying k-vector space has its basis the
set of all paths of length l ≥ 0 in Q, the natural graph- concatenation is the
product of two paths [2, 56].
A k-algebra A is said to be basic if it has a complete set {e1 , e2 , . . . , el } of
primitive orthogonal idempotents such that ei A ej A for all i 6= j.
A relation for a quiver Q is a linear combination of paths of length ≥ 2 with
same starting point and same end point, not all coefficients being zero [2].
Let Q be a finite and connected quiver. The two sided ideal of the path
algebra kQ generated by the arrows of Q is called the arrow ideal of kQ and
l
is denoted by RQ , RQ is the ideal of kQ generated as a k-vector space, by
the set of all paths of length ≥ l. A two sided ideal I of the path algebra kQ
is said to be admissible if there exists m ≥ 2 such that RQ m
⊆ I ⊂ RQ2
.
If I is an admissible ideal of kQ, the pair (Q, I) is said to be a bound quiver.
The quotient algebra kQ/I is said to be a bound quiver algebra.
Gabriel proved the following result:
Theorem 1.9. Let A be a basic and connected finite dimensional k-algebra.
There exists an admissible ideal I of kQA such that A ∼
= kQA /I.
A k-linear representation or representation M of a quiver Q is a system of
the form:
M = ((Mx , ϕα ) | x ∈ Q0 , α ∈ Q1 )
where Mx is a k-vector space for each x ∈ Q0 and ϕ : Ma → Mb is a k-linear
map associated to each arrow α : a → b ∈ Q1 .
Gabriel also proved the following theorem [2]:
28
Theorem 1.10. Let Q be a finite, connected, and acyclic quiver; k be an
algebraically closed field; and A = kQ be the path k-algebra of Q.
1. The algebra A is representation-finite if and only if the underlying graph

Q of Q is one of the Dynkin diagrams An , Dn , E6 , E7 and E8 with n ≥ 4.
2. If Q is a Dynkin graph, then the mapping dim : M → dim M induces

a bijection between the set of isomorphisms classes of indecomposable
A-modules and the set of positive roots of the quadratic form qQ of Q.
3. The number of the isomorphism classes of indecomposable A-modules

equals n(n+1)
2
, n2 − n, 36, 63 and 120, if Q is the Dynkin graph An , Dn ,
E6 , E7 and E8 with n ≥ 4.
1.4.1 Auslander-Reiten Quiver

For a finite dimensional k-algebra A, let h : M → N and u : L → M be
homomorphisms of right A-modules then an A-homomorphism s : N → M is
a section of h if hs = 1N and an A-homomorphism r : M → L is a retraction
of u if ru = 1L [3, 4, 66].
If s is a section of h, then h is surjective, s is injective, there are direct sum

decompositions M = Im s ⊕ Ker h ∼ = N ⊕ Ker h and h is a retraction of s.
Similarly, if r is a retraction of u then r is surjective, u is injective, u is a
section of r, and there exist direct sum decompositions M = Im u ⊕ Ker r ∼ =
L ⊕ Ker r.
An homomorphism h : M → N is called a section (or a retraction) if h

admits a retraction (or a section, respectively).
hn−1 h
− Xn−1 −−−→ Xn −−n→ · · · (infinite or finite) of right A-
A sequence · · · →
modules connected by A-homomorphisms is called exact if Ker hn = Im hn−1
for any n. In particular:
u r
0→
− L→
− M→
− N →0
is called a short exact sequence if u is a monomorphism, r is an epimorphism

and Ker r = Im u.
29
The pair (K, S) is called a Krull-Schmidt category with short exact sequences
provided K is a Krull-Schmidt category, and S is a set of pairs (g, f ) of maps
in K such that f is a kernel of g in K, and g is a cokernel of f in K (this
means that for (g, f ) ∈ S, we have gf = 0, and if gf 0 = 0 for an f 0 in K then
f 0 = f ν for an unique ν in K, and if g 0 f = 0 for a g 0 in K then g 0 = γg for an
unique γ in K). The pair (K, S) is said to be an exact category provided K
is a Krull-Schmidt category, such that there exists a full embedding K ⊆ A
with A an abelian category such that S is the set of all short exact sequences
belonging to K, and moreover K is closed under extensions in A [56].
Given a Krull-Schmidt category (K, S) with short exact sequences, then

(g, f ) ∈ S with f : X 0 → X, g : X → X 00 is said to be split provided
f is a split monomorphism (i.e., f is section) or equivalently g is a split
epimorphism (i.e., g is a retraction).
Given a Krull-Schmidt category K and X an object in K, a source map for

X in K is a map f : X → Y satisfying the following properties:
1. f is not split mono
2. Given f 0 : X → Y 0 , not split mono, there exists ν : Y → Y 0 with

f 0 = νf and
3. If γ ∈ End Y satisfies γf = f then γ is an automorphism.
Dually, let Z be an indecomposable object in K. Then a sink map for Z in

K is a map g : Y → Z satisfying the following properties
1. g is not split mono
2. Given g 0 : Y 0 → Z not split epi, there exists ν : Y 0 → Y with g 0 = gν;

and
3. If γ ∈ End Y satisfies g = gγ then γ is an automorphism.
source and sink maps were called by Auslander and Reiten minimal left
almost split map and minimal right almost split map respectively.
Let (K, S) be a Krull-Schmidt category with short exact sequences. The pair
(g, f ) is said to be an almost split sequence or Auslander-Reiten sequence of
30
(K, S) provided (g, f ) belongs to S with f being a source map in K and g a
sink map in K.
The quiver or Gabriel’s quiver ∆(K) of a Krull-Schmidt category is defined

as follows: its vertices are the isomorphism classes [X] of the indecomposable
objects X in K and there is an arrow [X] → [Y ] in case Irr(X, Y ) 6= 0. A
component of K is by definition the object class generated by the indecom-
posable objects belonging to a connected component of ∆(K) [56].
Given an Auslander-Reiten sequence (g, f ) of (K, S), with f : X → Y ,

g : Y → Z, X, Z indecomposable objects such that the isomorphism class
[X] of X is uniquely determined by the isomorphism class [Z] of Z, and
[Z] is uniquely determined by [X]. Then [X] is denoted by τ(K,S) [Z] (or
−
X = τ(K,S) Z) and similarly τ(K,S) [X] = [Z]. τ(K,S) is called the Auslander-
Reiten translation of (K, S).
n
If an object Y is decomposed in the form Y = ⊕ Yidi with Yi indecompos-
i=1
able and pairwise non-isomorphic (and di ≥ 1) then dimk Irr(X, Yi ) = di =
dimk Irr(X, Yi ) for all i thus there are di arrows [X] → [Yi ] and di arrows
[Yi ] → [Z]. The quiver ∆(K) becomes a translation quiver by adding τ(K,S) .
The translation quiver (∆(K), τ(K,S) ) is called the Auslander-Reiten quiver
Γ(K, S) of (K, S). Note that in case (K, S) is an exact category, then in
order to know that (g, f ) ∈ S is an almost split sequence it is necessary to
prove only that f is a source map or g is a sink map [2].
−
→
Below we show the Auslander-Reiten quiver of the algebra A = k(A4 ), ob-
tained by associating a linear orientation to the Dynkin diagram A4 .
d -d -d -d
4 3 2 1
1111
@
@
R1110
@
0111
@ @
@ @
0011
R0110
@ R
@
1100
@ @ @
@ @
R
@ @
R
@
R
@
0001 0010 0100 1000
Figure 9
31
1.5 Sections
Let Σ = (Σ0 , Σ1 ) be a connected and acyclic quiver. An infinite translation
quiver (ZΣ, τ ) has the set (ZΣ)0 = Z × Σ0 = {(n, x) | n ∈ Z, x ∈ Σ0 } as its
set of vertices, and for each arrow α : x → y ∈ Σ1 there exist two arrows
(n, α) : (n, x) → (n, y) (n, α0 ) : (n + 1, y) → (n, x) in (ZΣ)1 .
And these are all the arrows in (ZΣ)1 . The translation τ on ZΣ is given by
the formula τ (n, x) = (n + 1, x), and for every (n, x) ∈ (ZΣ)0 it is defined a
bijection between the set of arrows of target (n, x) and the set of arrows of
source (n + 1, x) by the formulas:
σ(n, α) = (n, α0 ) and σ(n, α0 ) = (n + 1, α).
Let (Γ, τ ) be a connected translation quiver. A connected full subquiver Σ

of Γ is a section of Γ if the following conditions are satisfied:
S(1) Σ is acyclic.
S(2) For each x ∈ Γ0 , there exists a unique n ∈ Z such that τ n x ∈ Σ0 .
S(3) If x0 → x1 → · · · → xt is a path in Γ with x0 , xt ∈ Σ0 , then xi ∈ Σ0 for

all i such that 0 ≤ i ≤ t.
For a translation quiver (Γ, τ ), the τ -orbit of a point x ∈ Γ0 is defined to be

the set of all points of the form τ n x with n ∈ Z. Thus, any section Σ meets
each τ -orbit exactly once.
The following are other conditions that a subquiver Σ must satisfy to be a

section of a translation quiver (Γ, τ ):
1. If x → y is an arrow in Γ and x ∈ Σ0 , then y ∈ Σ0 or τ y ∈ Σ0 .
2. If x → y is an arrow in Γ and y ∈ Σ0 , then x ∈ Σ0 or τ −1 x ∈ Σ0 .
Sections are useful to characterize representation-finite tilted algebras. Re-

garding this subject, we recall the Happel and Ringel’s criterion which states
that a connected representation-finite algebra B is a tilted algebra if and only
if the Auslander-Reiten quiver of Bcontains a section.
32
Henceforth, we let Ox denote the orbit of a fixed element x ∈ Γ0 . In par-
ticular, if Γ(Mod A) = (Γ0 , Γ1 ) is the Auslander-Reiten quiver of an algebra
of Dynkin type ∆n then each element of the τ -orbit of an indecomposable
projective module will be denoted τin , i ∈ N. We also note that in the case
of representation-finite hereditary algebras A the vertices of the Auslander-
Reiten quiver ΓA corresponding to the indecomposable projective modules
form in ΓA a section of Dynkin class.
As an example in the following picture we show an oriented quiver Q of type

A3 and the corresponding Auslander-Reiten quiver of the algebra A = kQ.
d d -d
1 2 3
τ.1 -τ.2 .τ3

..6 ..6A ..6
.. .. A ..
.. . .
.. ... A ...
τ1−1 -τ2−1 AU τ3−1

A
Figure 10
−1
−1
In this case sections are S 1 = {τ1 , τ2 , τ3 }, S2 = τ 1 , τ2 , τ3 , τ 1 , τ2 , τ3 ,
S3 = τ1−1 , τ2 , τ3−1 and S4 = τ1−1 , τ2−1 , τ3−1 all of them of type A3 .

Recently, Bravo et al [9] found out the following results regarding categori-
fication of integer sequences. Actually, these results regard categorification
of the integer sequences A049611, A084851 in the OEIS. In this case, S(Ak )h−
denotes the number of sections in the Auslander-Reiten quiver of an algebra
of Dynkin type Ak whose hth vertex (numbering from the left to the right)
is a sink.
n
X
Theorem 1.11. S(An )h− = 2n−4 (n2 + 5n + 2), n ≥ 4.
h=1
bh
2
c
k+1
k−2
P
Theorem 1.12. S(Ak )h− = (h + 1)2 − j h−2j
.
j=1
33
Chapter 2
Some Applications of the

Theory of Representation of
Posets and Quivers
In this chapter, we explore the use of tiled orders and representation of some
quivers in order to establish connections between number theory, combina-
torics and spectral theory.
2.1 On The Number of Antichains in Pn

In this section, we define a partition formula for numbers in sequence A016269
as V. Cifuentes et al described in [18], numbers in this sequence counts the
number of two-point antichains A2n in the powerset of an n-element set. Some
properties of numbers in this sequence are established as well.
Tiled orders Γn allows to conclude the following recursion formula:

n−2
Supp(Λ% Bnr
P
n ) =
r=0
where SΛ%n
(SΓ.
n
) is the size of the support of the upper diagonal (lower di-
agonal). And {Bnr | 0 ≤ r ≤ n − 2} is a set-partition of Supp(Λ% n ) such
that
r
Tn
Bnr gnr −k.
P
1. For each n ≥ 0, | | is a sum of consecutive positive integers
k=0
34
2.



2n−1 − 1, if r = 0,
 n−r−1
gnr = hrn =
P
[c(s, n − s) − c(s, n − s − r − 1)], if r ≥ 1

 s=1
 (n−1)n
t
n−1 = 2
, if r = n − 1.
3. (
c(h + 1, n), if 0 ≤ h ≤ n − 3,
Tnh =
n − 1, if h = n − 2.
n+k

where c(k, n) = k
and tj denotes the jth triangular number.
Remark 2.1. For each r and n fixed, Bnr consists of subsets ρrk with | ρkr |=
gnr − k defining uniquely a projective module Pk0 (r) ⊂ Λn .
SΓ.
n
= A2n = 3A2n−1 + t2n−1 −1 .
where A2n denotes the number of two-point antichains in N(Γn ) (see Figure
8).
We prove the following result by induction:
Theorem 2.2. For n ≥ 1, N(Λn ) ' Pn = (2n , ⊆) and A2n = t2n − 3n .
Proof. First of all, we note that projective modules Pi0 = (λi1 , λi2 , . . . , λi2n ) ∈
Λn are uniquely determined by points xi ∈ Pn by definition. That is,
λij ∈ {0, 1} with
λij = 0 if and only if xi ⊆ xj in Pn .
Note that in this case, xi ⊆ xj if and only if xi = {i1 , i2 , . . . , ik }, xj =

{j1 , j2 , . . . , jm }, k ≤ m and ih ≤ js for all 1 ≤ h ≤ k and 1 ≤ s ≤ m where
≤ is the usual order of the natural numbers.
Matrix blocks Bnr are obtained by fixing subsets of the form P(1, 2, . . . , r) =
{i1 , i2 , . . . , ir } and determining the set of incomparable
T points of each point
xs ∈ H(1, 2, . . . , r) ⊂ Pn with P(1, 2, . . . , r) = xs and | xi |=| xj |
xs ∈H(1,2,...,r)
for all xi , xj ∈ H(1, 2, . . . , r).
35
Secondly, we note that, A21 = t2 − 3 = 0, and A22 = t4 − 9 = 1. Suppose now
that the second assertion is true for all j, 1 ≤ j ≤ k − 1, in particular,
k−3
A2k−1 = 3i t2k−i−2 = A2k−1 = t2k−1 − 3k−1 .
P
i=0
Therefore,
k−2
A2k = 3i t2k−i−1 = 3(t2k−1 − 3k−1 ) + t2k−1 −1
P
i=0
then
k−1 (2k−1 +1) k−1 −1)(2k−1 )
2k−1
A2k = 3( 2 2
) + ( (2 2
) − 3k = 2
(32k−1 + 3 + 2k−1 − 1) − 3k .
Thus
2k−1
A2k = 2
[42k−1 + 2] − 3k = 2k−1 [2k + 1] − 3k = t2k − 3k .
Isomorphisms between tiled orders Λn and Γn allows to obtain the following

formula partition of A2n for each n ≥ 3:
n−3
Lemma 2.3. A2n =
P
[tgnr − tgnr −Tnr ].
r=0
For instance:
g30 = 3, h03 = c(1, 2) = 3, thus:
A23 = tg30 + tg30 −h03 + t2 = 6 + 3 = 9.
g40 = 7, g41 = 7
h04 = c(1, 3) = 4, h14 = c(2, 2) = 6, thus
A24 = (tg40 − tg40 −4 ) + (tg41 − tg41 −6 ) + t3 = (28 − 6) + (28 − 1) + 6 = 55.
Sequence A2n (A016269 in OEIS) grows rapidly, as an example the following

are the first ten terms of this sequence:
{0, 1, 9, 55, 285, 1351, 6069, 26335, 111645, 465751}
36
A250 = 633824582216127571845716253751,
A2100 = 803469022128979760250249034840178665431449990319442661731375.
(2.1.1)
The following results are easy to see via Theorem 2.2:
Corollary 2.4. For n ≥ 1 sequence A2n satisfies the following conditions:
1. If n ≡ 0 mod 4 then 5 | An2 and A2n ≡ 1 mod 3.
2. If n ≡ 1 mod 4 then 3 | An2 and 5 | A2n .
3. If n ≡ 2 mod 4 then A2n ≡ 3 mod 4, A2n ≡ 1 mod 3 and A2n ≡ 1
mod 10.
4. If n ≡ 3 mod 4 then 3 | An2 and A2n ≡ 9 mod 10 .
Theorem 2.5. The sequence A2n has no prime numbers.
Proof. According to the last Corollary, it suffices to consider the case n ≡ 2
mod 4. In such a case if A2n is a prime number p then ( −3
p
) = 1 and ( −1
p
) = −1
which is a contradiction.
Remark 2.6. Formula proposed in Theorem 2.2 was also found out by Riv-
iere in [58] by using other methods.
2.2 Energy, The Drozd’s Theorem and the

Vietá’s Formula
In this section, we describe the use of the spectrum of a tree in order to
classify algebras. To do that, we recall here the Drozd’s theorem regarding
classification of algebras [24, 26, 30, 44, 53].
Let k[x] and k[x, y] be the free associative algebras generated by x and by
x and y, respectively (the algebras of the quivers consisting of 1 or 2 loops).
A finite dimensional algebra A is called tame if it is not finitely represented
and if for each d ∈ N there are finitely many k[x] − A-bimodules (or se-
ries) Mi which are free of rank d over k[x] and such that each indecom-
posable A-module of dimension d is isomorphic to k[x]/(x − λ) ⊗k[x] Mi for
37
some i and some λ in k. It is called wild if there is a k < x, y > −A-
module which is finitely free over k < x, y > and such that the functor
F : modfp k < x, y >→ mod A, N → N ⊕k<x,y> M preserves indecompos-
ability and heteromorphisms in the sense of Gabriel and Roiter.
In the theory of representation of posets, a poset P is said to be tame if

µd < ∞ where µd is the minimum number of series containing almost all
d
indecomposable representations of a dimension d. Besides a tame poset P is
said to be of finite growth representation type if sup µd < ∞.
d
The following is the Drozd’s theorem:
Theorem 2.7. Each finite-dimensional algebra A is either wild or tame or

finitely represented.
The following definitions are useful to enunciate the M.I. Platzeck and E.
Fernández’s theorem.
Let ∆ = (∆0 , ∆1 ) be a graph then the adjacency matrix is the n × n-integer
matrix A∆ :
(
the number of edges between i and j, if i 6= j,
(A∆ )i,j =
two times the number of loops at i, if i = j.
If Λ∆ is the set of characteristic values of A∆ then max |λ| is the spectral

P λ∈Λ∆
radius of ∆, denoted ρ(Q), whereas E(∆) = |λ| is the energy of the tree
λ∈Λ∆
∆ or Gutman index. If Sn denotes an n-vertex star (a tree whose all of its
edges share the same vertex) then the following result is very important in
spectral theory:
Theorem 2.8. E(Sn ) < E(∆) < E(An ), for any n-vertex tree ∆.
We recall that the main problem in the theory of indices consists of finding
the value of the energy for a significant family of trees.
The characteristic polynomial of an n-vertex tree ∆ is denoted Φ∆ (x) =

bn
2
c
(−1)k m(∆, k)xn−2k where m(T, k) is the number of k-matchings in ∆.
P
k=0
38
Thus the energy E(∆) of ∆ is given by the Coulson integral:
n
Z ∞ b2c
2 1X
2
(−1)k m(∆, k)xn−2k dx. (2.2.2)
π 0 x k=0
Theorem 2.11 (proved by M.I. Platzeck and E.Fernández) is a direct conse-

quence of the theorem of classification of algebras given by Donovan-Freislich
and Nazarova and the following results from the spectral theory [24, 30, 53]:
Theorem 2.9. The largest eigenvalue λ1 and√ the maximum vertex degree δ∆
of a graph ∆ are related by the inequality δ∆ ≤ λ1 ≤ δ∆ .
Theorem 2.10. 1. A finite connected graph ∆ is a Dynkin diagram if

and only if ρ(∆) ≤ 2.
2. A finite connected graph ∆ is an Extended Dynkin diagram if and only

if ρ(∆) = 2.
Theorem 2.11. If Q is a connected quiver without oriented cycles then:
1. kQ is of finite representation type if and only if ρ(Q) < 2,
2. kQ is of tame representation type if and only if ρ(Q) = 2,
3. kQ is of wild representation type if and only if ρ(Q) ≥ 3.
Since it is well known that any number λ in the spectrum of a Dynkin diagram
nπk
of type An has the form cos( n+1 ), k = 0, . . . , n. We will focus the study of
the spectrum of An and Dm with n = 2k − 1 and m = 2k + 1, k ≥ 2.
Firstly, we recall the following result given by L.D. Servi [64]:
Theorem 2.12. If b1 lies in the interval [−2, 2] and bi ∈ {−1, 0, 1}, for i 6= 1
then
cos(f (bk , bk−1 , . . . , b1 )) = R(bk , bk−1 , . . . , b1 ). (2.2.3)
If in addition bk 6= 0 then
sin(f (bk , bk−1 , . . . , b1 )) = R(1, −bk−1 , bk−2 , bk−3 , . . . , b1 ). (2.2.4)
39
where
v v
u u s
u r
bk t b1 π
u u
2 + bk−1 2 + bk−2 2 + · · · + b2 2 + 2 sin(
t
R(bk , bk−1 , . . . , b1 ) = ).
2 4
1 bk bk bk−1 bk bk−1 bk−2 bk bk−1 bk−2 . . . b1
f (bk , bk−1 , . . . , b1 ) = ( − − − ··· − )π.
2 4 8 16 2k+1
(2.2.5)
As a consequence of Theorem 2.12, we have the following results:

Theorem√ 2.13. If λ is an element of the spectrum of A2r −1 , r ≥ 2 then
λ ∈ {0, ± 2} or it has the form:
v v
u u s
u r
b1 π
u u
R(bk , bk−1 , . . . , b1 ) = bk 2 + bk−1 2 + bk−2 2 + · · · + b2 2 + 2 sin(
t
t ).
4
(2.2.6)
with 1 ≤ k ≤ r − 1, and for each k fixed it holds that bj = ±1, 1 ≤ j ≤ k.

Proof. By induction. Bearing in mind that the characteristic polynomial
r −2
2P
2r +j

p(λ) associated to A2r −1 has the form p(λ) = 2j+1
(−1)j+1 x2j+1 .
j=0
Corollary 2.14. The Gutman energy of A2r −1 , r ≥ 2 is given by the identity:
j−1
r−2 2X
X (2k + 1)π
E(A2r −1 ) = 4 cos( ). (2.2.7)
j=0 k=0
2j+2
Corollary 2.15. E(D2r +1 ) = E(A2r −1 ) − E(A2r−1 −1 ).

r +1
Proof. The characteristic polynomial p(x) associated to D2r +1 is x2 −
2r−1
P−1 r −j r −j−2
r 2r −1 r
2x + (−1)j+1 ( 2j+1 − 2 j−1 )x2 −2j−1 .
j=1
Corollary 2.16. (Vietá0 s formula)

If Sp(A2r−1 ) denotes the spectrum associated to the adjacency matrix of A2r−1
and λ 6= 0 then
40
∞
Y |λ| Y ρ(D2r +1 ) 2
Lim = = .
r→∞ 2 2 π (2.2.8)
λ∈Sp(A2r −1 ) r=2
bj =1,1≤j≤k
Since the characteristic polynomial pAe 2r −1 (x) = x1 (pD2r +1 (x))−2 and pDe 2r +2 (x) =
x2 (pD2r +1 (x)) then the following result holds:
Corollary 2.17.
e 2r −1 ) = Sp(A2r−1 −1 ) ∪ {−2, 2},
Sp(A
(2.2.9)
e 2r +2 ) = Sp(A2r −1 ) ∪ {−2, 2}.
Sp(D
We recall that a complete description of characteristic polynomial of Dynkin

diagrams (laced and extended) has been given by Dokuchaev, Futorny et al
in [26].
2.2.1 Cluster Algebras

Cluster algebras were introduced by Fomin and Zelevinsky in 2000. The
classification of these algebras is similar to the classification of path algebras.
In the sequel, we follow ideas of L. Williams to describe cluster algebras
associated to a quiver Q [68].
Let Q be a quiver without loops and 2-cycles and k a vertex of Q. The

mutated quiver µk (Q) is defined as follows:
1. For each subquiver i → j → k add a new arrow i → j.
2. Reverse all arrows with source or target k.
3. Remove the arrows in a maximal set of pairwise disjoint cycles.

Let Q be a quiver without oriented cycles or loops then Q may be encoded by
an m×m-skew symmetric exchange matrix B(Q) = (bij ) where bij = −bji = l
where there are l arrows from vertex i to vertex j. B(Q) is called the signed
adjacency matrix of the quiver Q.
The extended exchange matrix B(µk (Q)) = (b0ij ) of the mutated quiver µk (Q)
is given by the following identities:
41


 −bij , if i = k or j = k,

b + b b , if b > 0 and b > 0,
ij ik kj ik kj
b0ij =


 b ij − b b
ik kj , if b ik < 0 and b kj < 0,

bij , otherwise.
if Q is a quiver with set of vertices Q0 = {1, 2 . . . , n, n + 1, . . . , m} and x =
(x1 , x2 , . . . , xm ) forms a free-generating set for F (an ambient field of rational
functions in n independent variables over Q(xn+1 , . . . , xm )) then (x, Q) is a
labeled seed where vertices 1, 2, . . . , n are called mutable and vertices n +
1, n + 2, . . . , m are called frozen.
The seed mutation µk in direction k transforms (x, Q) into the labeled seed
µk (x, Q) = (x0 , µk (Q)) where the cluster (x1 , x2 , . . . , xm ) is defined as follows:
x0j = xj for j 6= k,
Y Y Y b Y −b
x0k xk = xt(α) + xs(α) = xi ik + xi ik . (2.2.10)
α∈Q1 α∈Q1 bik >0 bik <0
s(α)=k t(α)=k
P
A cluster pattern is an assignment of a labeled seed t = (xt , Qt ) to every
vertex in the n-regular tree t ∈ Πn such that the seed assigned to the end
points of any edge k : t → k 0 are obtained from each other by the seed mu-
tation in direction k, the components of xt are written as (x1,t , x2,t , . . . , xn,t ).
We denote:
[
χ= xt = {xi,t : t ∈ Πn , 1 ≤ i ≤ n}. (2.2.11)
t∈Πn
Elements xi,t are called cluster variables, the cluster algebra A associated
with a given pattern is the Z[c]-subalgebra of the ambient field F generated
by all cluster variables. We denote A = (A, Q), (x, Q) is any seed in the
underlying cluster pattern.
As an example, we show all clusters of the cluster algebra associated to the

quiver Q = 1 −→ 2.
µ1 1+x2 µ2 1+x2 1+x1 +x2 µ1 1+x1 1+x1 +x2 µ2
x1 −→ x2 =⇒ x1
←− x2 =⇒ x1
−→ x1 x2
=⇒ x2
←− x1 x2
=⇒
1+x1 µ1
x2
−→ x1 =⇒ x2 ←− x1 .
42
The following is the interpretation in cluster algebras of the divisibility of
the Fermat number F5 given by Fomin, Williams and Zelevinsky [37]:
Consider the rank 2 cluster algebra with the initial seed (x, B) where x =
(x1 , x2 , x3 ) and
 
0 4
B =  −1 0 
1 −3
Thus,
x2 + x3
x01 = (2.2.12)
x1
Therefore, the following sequence of mutations has place (writing on the top
a corresponding
 specialization):
     
3 −1 5 −1 5 −641 −128 −641
 0 4  µ1  0 −4  µ2  0
    4  µ1  0
  −4  µ2
 −1 0  =⇒  1 =⇒  =⇒ 
  =⇒
0  −1 0  1 0 
1 −3 −1 1 0 −1 −1 −3
 
−128 −F5 /641
 0 4 
 .
 −1 0 
0 1
Since any cluster variable specializes to an integer (see, Theorem 2.19) it

follows that 641 divides F5 as Euler discovered before.
Cluster algebras are associated to triangulations as follows: Let T be any

triangulation of a d-gon, let m = 2d−3 and n = d−3. Set x = (x1 , x2 , . . . , xm )
then (x, Q(T )) is a labeled seed and it determines a cluster algebra A(T ) =
A(Q, T ). In this case, vertices are given by diagonals (corresponding to
mutable vertices) and sides (corresponding to frozen vertices), arrows of Q(T )
are obtained by inscribing new triangles on the vertices Q0 (T ) whose edges
are oriented clockwise, the edges of these triangles are the arrows desired (see
Figure 12).
The following results regard cluster algebras:
43
Theorem 2.18. (S.Fomin, A.Zelevinsky, 2002) All elements in a cluster al-
gebra are Laurent polynomials.
Theorem 2.19. In a cluster algebra of geometric type, each cluster vari-

able can be expressed as a Laurent polynomial with integer coefficients in the
elements of any cluster.
Theorem 2.20. (S.Fomin, A.Zelevinsky, 2003) Finite type cluster algebras

(with finitely many clusters) are classified using finite type Dynkin diagrams.
Theorem 2.21. (Lee, Schiffler, 2013; Gross, Hacking, Keel, Kontsevich, 2014)
The Laurent polynomial of each cluster variable has only nonnegative coeffi-
cients.
2.2.2 Snake Graphs, Continued Fractions and Cluster

Algebras
In this section, we describe some interesting relationships between contin-
ued fractions, Markov triples, snake graphs and cluster algebras reported by
Schiffler and Çanakçi in [11].
According to Schiffler and Çanakçi a snake graph G is a connected planar

graph consisting of a finite sequence of tiles G1 , G2 , . . . Gd such that Gi and
Gi+1 share exactly one edge ei and this edge is either the north edge of Gi
and the south edge of Gi+1 or the east edge of Gi and the west edge of Gi+1 .
A sign function on a snake graph is a map from the set of edges of G to the
set {+, −} such that on every tile the north and the west edge have the same
sign, the south and the east edge have the same sign and the sign on the
north edge is opposite to the south edge.
The following is an example of a sign function associated to a snake graph:
44
-
-
-
+
+ - +
+
- +
-
Figure 11
In this case, the sequence of signs associated has the form S = {−, −},
{+, +, +}, {−}, {+, +}, {−, −, −}.
84
S corresponds to the convergent [2, 3, 1, 2, 3] of the continued fraction 37 =
2 + 3+ 1 . Actually, G has 84 perfect matchings and there is a bi-
1
1+ 1
2+ 1
3
jection between snake graphs G[a1 , a2 , . . . , an ] and the continued fraction

[a1 , a2 , . . . , an ]. In fact, the following result has place:
Theorem 2.22. If m(G) denotes the number of perfect matchings of the

snake graph G then
m(G[a1 , a2 , . . . , an ])
[a1 , a2 , . . . , an ] = (2.2.13)
m(G[a2 , a3 , . . . , an ])
and the fraction at the right hand side is reduced.
Corollary 2.23. The number of snake graphs that have precisely N perfect
matchings is precisely φ(N ) where φ is the totient function of Euler.
As an example the snake graph of the continued fraction [a1 , a2 , . . . , an ] =

[1, 1, . . . , 1] is the straight snake graph with (n − 1)-tiles and Fn+1 perfect
matchings where Fn+1 denotes the (n + 1)st Fibonacci number.
45
2.2.3 Palindromic Snake Graphs and Sums of Two Squares
A continued fraction [a1 , a2 , . . . , an ] is called even if n is even and palindromic
if an−i = ai for all i.
A snake graph is called palindromic if it is the snake graph of an even palin-
dromic continued fraction. A snake graph has rotational symmetry at a
center tile if the rotation about 1800 at the center of the central tile is an
automorphism [11].
Theorem 2.24. A snake graph is palindromic if and only if it has a rota-
tional symmetry at its center tile.
pn
Theorem 2.25. Let [a1 , a2 , . . . , am ] = qn
. Then [an , . . . , a2 , a1 , a2 , . . . , an ] =
p2n +qn
2
pn−1 pn +qn−1 qn
.
An integer N is called sum of two squares if there exist integers p > q > 1,
gcd(p, q) = 1 such that N = p2 + q 2 .
Corollary 2.26. Let N > 0, if N is a sum of two squares then there exists
a palindromic snake graph G such that m(G) = N . The number of ways one
can write N as a sum of two squares is equal to one half of the number of
palindromic snake graphs with N perfect matchings.
Markov Triples
A triple of positive integers (a, b, c) is called a Markov triple if a2 + b2 + c2 =
3abc, a, b and c are called Markov numbers. The following is a list of Markov
numbers:
(29, 2, 169)
%
(29, 2, 5)
%
(1, 1, 1) −→ (1, 2, 1) −→ (1, 2, 5) −→ (29, 433, 5)
&
& (194, 13, 5)
(1, 13, 34)
The following is the conjecture proposed by Frobenius in 1913 regarding
Markov triples.
Conjecture Uniqueness. The largest integer in a Markov triple determines
the other two.
46
Theorem 2.27. Every Markov number is the numerator of an even palin-
dromic continued fraction.
Corollary 2.28. Every Markov number except 1 is a sum of two squares.
Schiffler, Musiker and Williams proved the following result giving a relation-
ship between cluster algebras and snake graphs [11].
Theorem 2.29. If γ is an arc in a triangulated surface (S, M ) the cluster
variable xγ is given by the formula:
1 X
xγ = x(P ). (2.2.14)
cross Gγ P ∈match G
γ
The following figure shows a quiver Q(T ) associated to a triangulation T and

the cluster variable xγ associated to the diagonal γ via matchings x1 , x2 and
x3 . In this case, hollow circles denote frozen vertices, shaded circles 1, 2 and 3
denote mutable vertices. Actually, Q0 (T ) = {a, b, c, d, e, f, 1, 2, 3}. Inscribed
triangles comprise Q1 (T ).
b
*◦

a ◦
• H1
@◦ c
xγ = x1 +x 2 +x3
6H @ x1 x2
j •2 @
H
@
f@◦@- •3 I◦?
@
d
1 2
x1
◦
I
@@ ?
e 2
b 1 2
x3
a 1 γ @ c
@ 1
2 d
@
2
@ b 1 3 2 c 1 2
f@@
3
d x2
a 1
e
Figure 12
2.3 Partitions into Triangular Numbers

On partitions into squares and triangular numbers, we must recall that Ja-
cobi in his famous Fundamenta Nova of 1829, introduced elliptic and theta
functions, and used them as tools in the study of the equation:
47
x21 + x22 + · · · + x2s = n,
Jacobi obtained elegant explicit formulas for the number of representations of

a positive integer n as the sum of s squares with s = 2, 4, 6, 8. For example,
the number of representations of n as the sum of four squares is given by:
r4 (n) = 8[2 + (−1)n ]σ0 ,
where σ0 denotes the sum of the odd divisors of n.
We also quoted that finding arithmetic relations for partitions of numbers

into three triangular o square numbers is a tricky problem, we recall here
one of the few results regarding the subject reported by Hirschhorn and
Sellers in [39].
Theorem 2.30. Let p34 (n) be the number of partitions of an integer n into
three triangular numbers and pd34 (n) be the number of partitions of n into
three distinct triangular numbers. Then for all n ≥ 0,
p34 (27n + 12) = 3p34 (3n + 1)

(2.3.15)
pd34 (27n + 12) = 3pd34 (3n + 1).
P n(n+1)
To do so, they used the Ramanujan function ψ(q) = q 2 , in order to
n≥0
establish corresponding generating function as follows:
X 1
p34 (n)q n = (ψ 3 (q) + 3ψ 2 (q)ψ(q) + 2ψ(q 3 ))
n≥0
6
(2.3.16)
X 1
pd34 (n)q n = (ψ 3 (q) − 3ψ 2 (q)ψ(q) + 2ψ(q 3 )).
n≥0
6
We consider the following family of bounded quivers Q obtained by attaching

to each vertex of a linearly oriented diagram of type An a copy of its corre-
sponding Auslander-Reiten quiver and defining a numbering on it induced by
an integer frieze. Now we can see the way the indecomposable representation
P (0) allows to obtain the number of partitions of an integer n into at most
three triangular numbers (See Figure 13).
48
d
I
@@
@d
@I@
d @d
@
I
−
→ I
@@ @
Γ3 (A4 ) @d @d
@
I@
d d @d
I
@@ @
I
@
d @d @d d
I
@
@ @
I
@ α1,i @
d @ d @d d d d
@
R
@
I
@
@ I
@@ I
@@ 6 @ α1,i β @α2,i
0
1,i @
d @ d @ d @d d R d Rd
@
@ @
6 @ 0 @ @
−
→ β1,i β2,i @
d d d Rd
@ @
Γ2 (A4 ) R
@ R
@ @
6 −
→
Γ1 (A4 )
d
−
→ 1≤i≤3
A4 arrows in each copy are denoted
according to the corresponding copy
d
18
I
@@
@ d15
@ I@
d 12 @ d13
I
@@ @I@
@d @ d12
10
@
I@
d9 d8 @d
9
I
@@ I
@
@
d
10
@d
6 7
d7 @d
I
@@ @
I@ @
d12 @ d @d d6 d4 Rd
@
@
7 4 8
I
@@ I
@@ I
@@ 6 @ @
d @ d @ d @d d2 R d Rd
@
@ @
@
15 9 5 3 5 10
6 @ @ @
d R d R d Rd
@
@ @
@ @@
1 3 7 13
6
d0
Figure 13
49
Quiver Q is bounded by the following relations:
s s 0
αj,i βj,i = 0,
r r 0
βj,i αj+1,i = 0, (2.3.17)
0 00
βj,i βj,i βj,i . . . = 0.
To each labeled path starting in a point n(n+1)

2
, P = { n(n+1)
2
, l(x1 ), l(x2 ), . . . , l(xn )}
it is associated a multiset ∂P , whose elements are the differences |xi − xi−1 |.
Thus two of these labeled paths P, P 0 are equivalent denoted P ' P 0 if
e(P ) = e(P 0 ),
(2.3.18)
∂P = ∂P 0 .
Besides two vertices xi , xj ∈ Q0 are equivalent provided l(xi ) = l(xj ). And

these are all relations defined on Q.
d
18
I
@@
@ d15
@ I@
d 12 @ d13
I
@@ @I@
@d @ d12
10
@
I@
d9 d8 @ d9
I
@@ I
@@
ψ 2 (q)
d
10
@d
6
@ d7 d7
I
@@ I
@@ @
ψ 2 (q) 3
d12
@ d7
@d d6 d4 R ψd (q)
@
@
4 8
I
@@ I
@@ I
@@ 6 @ @
3
d @ d @ d @ d3
ψ(q 2 )
d2 ψ 3 (q) d
@ R
@ R dψ (q)
@
@
15 9 5 5 10
6 @ @ @
d R d R d Rd
@
@ @
@ @@
1 3 7 13
6 ψ(q 3 ) ψ 3 (q) ψ 3 (q)
d0
Figure 14
50
Note that the path P = {0, 1, 3, 6, 7} is equivalent to the path P 0 = {0, 1, 2, 4, 7}.
P n(n+1)
Figure 14 shows the action of the Ramanujan function ψ(q) = q 2 on
n≥0
Q. Thus, each coordinate n of the indecomposable projective representa-
tion P (0) gives the number of partitions of n into at most three triangular
numbers. The labeling of Q is induced by a specialization of the following
mutating quiver of type An linearly oriented:
..
.
x4 x04 x00
4 x000
4

A A A A
A A A A
AA A A AAU iv
x03 x00 x000

x3
U U
A U
A x3
3 3
. . .
A

A
A

A
A

A
A
A
. . .
AU 0 AU 00 AU 000 AU iv
x2 x2 x2 x2 x2
A A
A A A
A A A A A
AA A A A AAU v
x0 x00 x000 xiv
x
U U
A A
U U
A x
1 1 1 1 1 1
x1 = x2 = x3 = x4 = 1
9
7 8 9
5 6 7 8 9
3 4 5 6 7 8
1 2 3 4 5 6 7 ...
1 2 3 4 5 6 ...
1 2 3 4 5 ...
1 2 3 4
1 2 3
1 2
1
Figure 15
51
Figure 16 shows an interpretation of the bounded quiver Q in the plane,
where equivalent paths belong to one of three types:
1. Type I induced by coefficients of type ψ(q), representing one triangular

number
2. Type II induced by coefficients of type ψ(q 2 ) or ψ 2 (q), representing a

sum of two triangular numbers.
3. Type III induced by coefficients of type ψ(q 3 ) and ψ 3 (q), representing

a sum of three triangular numbers.
52
14 15

10 - 11 - 12 - 13

12

9 - 10

6
- 7 -
8 9

-
6 7

- -
3 4 5

0 -1 -2 -3
Figure 16
Figure 17
Paths of type I, II and III represent sums of one, two or three triangular num-
bers, thus the indecomposable projective module P (0) allows to enumerate
this kind of partitions. Actually, we have the following result.
53
Theorem 2.31. Up to equivalence, the jst coordinate of P (0) as indecom-
posable projective representation of kQ is given by the number of partitions
of j into at most three triangular numbers.
2.4 The Cryptosystem Γ-Pentagonal

Since found out a formula for the number of partitions of an integer n into
three triangular numbers is a cumbersome problem, it can be used to define
a cryptosystem.
A cryptosystem is a sextuple
(A, P, C, K, E, D).
where:
A is an alphabet of definition (e.g. {0, 1}),
P is a finite set of plaintexts,
C is a finite set of ciphertexts,
K is a finite set of keys,
Associated to each key K there are encryption-decryption maps denoted

eK : P → C ∈ E and dK : C → P ∈ D respectively such that
dK (eK (x)) = x, for each plaintext x.
The cryptosystem Γ-pentagonal is defined as follows:
P = Z26 , C = Z26 × Z26 , K = {((x0 , y0 ), f0 , π, Γ) | π ∈ S(Γ)0 }.
where Γ is a secret connected graph with a secret initial vertex (x0 , y0 ),

f0 : Γ → Γ is a graph-automorphism (generally a shift operator) and π is a
permutation of vertices of Γ.
For a fixed key K the encryption-decryption system is defined as follows:
54
eK (x) = π(x) + α = (x1 , y1 ) ∈ Z × Z
(2.4.19)
dk (y) = π −1 (y) − α.
In this case, to each character x ∈ Z26 it is associated a label l ∈ Z26 of a

point (x0 , y0 ) ∈ (Γ)0 and α is the number of paths whose ending vertex is
(x0 , y0 ). In our case the secret graph Γ is the quiver Q defined before for
partitions of numbers into three triangular numbers. Thus, α is given by
compositions of numbers into three triangular numbers.
The following figure shows an example of a plaintext and the corresponding

ciphertext.
Figure 18
55
2.5 Visual Cryptography
In this section we describe some matrix problems based on some secret shar-
ing schemes as R.J. Serna et al describe in [17].
Visual cryptography is a kind of cryptography where the decoding process

only require the use of the human visual system without any special com-
putation. This kind of cryptography was introduced by Naor and Shamir
in 1994 [50]. In this case a secret is shared between a group of n persons
by giving to each of them an image (called shadow) with no information
about the secret, such images are printed onto a transparency in such a way
that the only way to recover the secret is by stacking the transparencies all
together [17, 18].
Naor and Shamir analyzed the case of k out n or (k, n)-threshold visual
cryptography schemes (TVS), in which the secret image is visible if and only
if any k transparencies are stacked together. Soon afterwards, this set up
was generalized by Blundo et al, Droste, and Klein in [5, 27] and [45].
Tsai et al, and Feng et al proposed sharing methods for multiple secret images
in [67] and [43] via XOR computing for embedding and extracting images,
and Lagrange’s interpolation, respectively [22]. According to Shyu et al [65],
Wu and Chen might be the first researchers to consider the problem of sharing
two secret images in two shares in visual cryptography [14, 70]. Afterwards,
Shyu et al proposed a generalization of the work of Wu and Chen, actually,
they defined a visual secret sharing scheme to encodes more than two secrets.
Nakajima [49] introduced the use of natural images in schemes of visual

cryptography (see Figure 26), besides Ross et al applied gray-level extended
visual cryptography schemes to preserve the privacy of digital images stored
in a central database [49, 59].
We also recall that some visual secret sharing schemes (VSSS) have been in-
troduced by using some mathematical structures. For instance, Cañadas et
al introduced a visual cryptography scheme with a special share T0 contain-
ing sets of nested images, all secrets can be revealed by superimposing some
transparencies to this fixed share. These authors also have used some prop-
erties of k-linear maps to generate schemes of multiple secret sharing [18].
56

Example of a Nakajima’s (2, 2)-scheme. In this case, shadows S1 (Dog) and

S2 (Pyramid) are used to encrypt the original image (Cat).
Figure 19
57
In 1998 Koga and Yamamoto [46] proposed a lattice-based TVS for color
images, in this case pixels are treated as elements of a suitable lattice S and
the stacking process is defined as an operation between elements of the lattice,
according to them the commutative and associative laws of such operation
allow to (k, n) VSSS to decrypt k shares by stacking up of all them in an
arbitrary order. Permitting the existence of inverses for all s ∈ S leads
to pathological VSSS for example, stacking a black subpixel with another
subpixel yield to a white or transparent subpixel, finite lattices are one of the
simplest structures that meet these requirements. Under these circumstances
we generalize this kind of TVS by using k-linear representations in such a way
that the VSSS is completely defined by the orbits defined by some admissible
transformations between columns and rows of a matrix representation.
2.5.1 Lattice-Based VSSS

Now, we present the definition of a lattice-based TVS in accordance with
Koga and Yamamoto [46]:
Let m > 0 be given and L a finite lattice of a finite number of colors that
can be physically realized. Suppose that C = {c1 , c2 , . . . , cJ } is a subset of
elements in L, which is not necessarily a sublattice of L. For all q satis-
fying 1 ≤ q ≤ k and distinct i1 , i2 , . . . , iq ⊆ {1, 2, . . . , n} define a mapping
h(i1 ,i2 ,...,iq ) : (Lm )n → Lm by
h(i1 ,i2 ,...,iq ) (x) = xi1 ∨ xi2 · · · ∨ xiq (2.5.20)
where x = (x1 , x2 , . . . , xn ) ∈ (Lm )n . If there exists (Xcj , Ycj )1≤j≤J is called

the lattice-based (k, n) VSSS with colors C.
1. For all j = 1, 2, . . . J and distinct {i1 , i2 , . . . , ik } ⊆ {1, 2, . . . , n}, all

x ∈ Xcj satisfy
h(i1 ,i2 ,...,ik ) (x) ∈ Ycj .
2. For all q < k and {i1 , i2 , . . . , ik } ⊂ {1, 2, . . . , n} define
(i ,i2 ,...,iq )
Xcj1 = {(xi1 , xi2 , . . . , xiq ) : (x1 , x2 , . . . , xn ) ∈ Xcj }.
58
(i ,i ,...,i )
Then Xcj1 2 q , j = 1, 2, . . . , J are indistinguishable in the sense that
they contain the same elements with the same frequencies.
6 1 ∈ L all the elements in Ycj are composed

3. For all cj ∈ C satisfying cj =
by 1’s and at least one cj . In case that cj = 1, Ycj has only one element
composed by m 1’s.
2.5.2 A Matrix Problem Induced by a Lattice-Based

VSSS
In this section, we interpret the Koga-Yamamoto scheme as a matrix problem.
To do that, we consider matrix representations (d, A) of a lattice L induced
by a finite number of colors. In this case, for a (k, n) VSSS we have that
A ∈ Ld0 ×d with d0 = m is the pixel expansion, di = dj = n for any i, j ∈
L is the size of the set of participants and k is the number of qualified
participants. In other words, each vertical stripe consists of n generators
(of the corresponding module Ucj for each color cj ∈ L) with k < n linear
independent columns. Besides a lattice-color matrix representation of L is an
m×d rectangular matrix separated into vertical stripes with the same number
c of columns. In this case, columns in each stripe Mx are indistinguishable
(i.e., they have the same elements appearing with the same frequency) and
constitute a composition (i.e., a partition where the order matters) of a given
vector Fx ∈ Lm for any x ∈ C.
For {x1 , x2 , . . . , xj } = C, let M and M 0 be lattice-color matrix representations

with associated vectors Fx1 · · · Fxt and Fx0 1 · · · Fx0 t respectively then M and
M 0 are called equivalent if and only if there exists some permutation π ∈ Sm
such that
π(1) π(m)
Fx0 j = (xj , · · · xj ) if Fxj = (x1j , · · · xm
j )
for each chosen xj ∈ L.
The following result establishes the existence of matrix representations whose

columns within each vertical stripe Mx constitute compositions of a given set
of fixed vectors Fx .
Theorem 2.32. If x ∈ C and Fx consists of k1 10 s and kx x0 s with k1 +kx = m

n
then there exist n indistinguishable vectors gx1 , . . . , gxn such that F (x) = gxi .
P
i=1
59
Proof. Let F (x) be such that, Fx = (a1 , . . . , am ) ∈ Lm with k1 1’s, kx x’s and
k1 + kx = m, and a permutation π ∈ Sm such that F x = (aπ(1) · · · aπ(m) ) =
(x · · · , x, 1, · · · , 1). We fix points y ∈ L ∩ x4 , z ∈ L, and an m × n matrix Rx
such that the entries of the first k(x) rows are y’s and the entries of the remain
rows are z’s. Then there exist integers q1 and r1 such that k1 = nq1 + r1 with
0 ≤ r1 < n, kx = nqx + rx with 0 ≤ rx < n.
Let us consider the matrix block

yIn xIn
.. ..
. .
yIn xIn
yA _ A
M x = Rx −
zIn In
.. ..
. .
zIn In
zB B
where In denotes an n × n matrix, the number of matrix blocks xIn (In ) in

M x is given respectively by nqx (nq1 ), in this case A is an rx × n matrix with
the form. Besides, −A is a matrix such that A ∨ (−A) = 0.
A= Irx
x, . . . , x
B is an r1 × n matrix with the form
B= Ir1
1, . . . , 1
Worth noting that empty blocks in A and B denote matrices whose entries
are all zeroes.
60
By construction columns of matrix M x constitute an n-elements set of in-
distinguishable vectors associated to the fixed vector F x . If matrix Mx is
obtained from M x by applying permutation π to the rows then columns of
Mx correspond to n indistinguishable vectors which define a composition of
the fixed vector Fx .
The following result defines admissible transformations which guarantees the

existence of equivalent lattice-color matrix representations. Therefore, it
guarantees the construction of different types of (k, n) lattice-based VSSS.
Theorem 2.33. Let M and M 0 be two lattice matrix representations of a

given lattice L then M and M 0 are equivalent if M and M 0 can be turned
one into each other by applying the following trasformations:
(a.) row permutations of the whole matrix.
(b.) column permutations within a given vertical stripe.
(c.) multiplication of a given column j in the stripe Mx by some scalar

z ∈ (λxj )O , where λxj is the maximum of all entries in such a column.
(d.) addition of a given jth column in the stripe Mx to the jth column in
the stripe My with coeficients in (δjy )M , where δjy is the minimum of all
entries in the column of My . If x ≤ y in L.
Proof. Row permutations of M determine same indistinguishable vectors

up to permutations. That is, if Fx1 , · · · , Fxt are fixed vectors attached to the
representation M and Fx0 1 , · · · , Fx0 t are corresponding fixed vectors attached
to the matrix M 0 obtained from M by row permutations then there exists a
permutation π ∈ St such that if
0 π(1) π(m)
Fxi = (x1j , · · · xm
j ) then Fxj = (xj , · · · xj );
thus M is equivalent to M 0 . Besides, column permutations in a given vertical

stripe Mx keep invariant vectors Fx1 , · · · , Fxt . Therefore, if M 0 is obtained
from M by transformations of type (a) then M is equivalent to M 0 .
On the other hand, if λxj is the maximum of all entries in a column j ∈ Mx

and z ∈ (λxj )O then z ≥ λxj and λxj ≥ gkj
x
where gjx = (g1j
x x
, · · · , gmj ) is the jth
x x 0
column of the stripe Mx then zgj = gj . Therefore if M has attached fixed
61
vectors as defined above and M is obtained via transformations of type (b)
then Fx0 i = Fxi for any 1 ≤ i ≤ t therefore M is equivalent to M 0 .
Finally, let us suppose that δjy is the minimum of the set of entries of the
jth column in a vertical stripe My (gjy = (g1j y y
, · · · , gmj ), that is, δjy ≤ gkj
y
for
all 1 ≤ k ≤ m and if gix = (g1i x x
, · · · , gmi ) is the ith column in Mx and we
add z ∧ gix to the column gjy with z ∈ (δjy )M then z ≤ gkj y
for all 1 ≤ k ≤ m
y y
x
thus (z ∧ gki ) ∨ gkj = gkj which means that (z ∧ gix ) ∨ gjy . Therefore if M 0 is
obtained from M via transformations of type (c) then M = M 0 .
The following result establishes the structure of vectors Fx with x ∈ C.

Theorem 2.34. If x ∈ C with x 6= a ∨ b for any x 6= a and x 6= b and Fx
consists of k1 10 s and kx x0 s with k1 +kx = m then an indistinguishable vector
gx consists of at least d kn1 e x0 s and at least d kn1 e 10 s where n is the number of
generators in Mx . Moreover if mx is the number of x0 s in gx and m1 is the
number of 10 s in gx then there exist m − (m1 + mx ) elements in xN in gx .
Proof. Let us suppose that Fx = (x, · · · , x, 1, · · · , 1) without loss of general-
ity, where k1 is the number of 1’s and kx is the number of x’s with k1 +kx = m
and x ∈ C. Since x cannot be obtained as a supremum of two points y and z
with y 6= x and z 6= x then the number of x’s must be at least d knx e. Indeed,
for each occurrence of x each part of the partition of the vector Fx contains
at least an x if they are ordered in the last kx rows of the vertical stripe the
result is obtained by using as few x’s as possible. Similarly, the result can
be obtained for a minimal number of 1’s if it is considered that 1 ∈ C. That
there exist m − (m1 + mx ) elements in xN follows from arguments used in
Theorem 2.32.
We note that structure of the form (ΓiQual , ΓiForb ) can be interpreted from
this point of view as indecomposable lattice-color matrix representations (see
Figure 21).
Example of a (2, 2) lattice-based encryption. In this case two color-shadows

and an eight color-lattice are used to encrypt an original image of Bart.
Figure 20
62
0 Y C G 1
1 0 0 Y Y C C G G 1 1
t
1 0 G 1 Y 1 G 1 1 1
t G
@
0 1 1 G 1 Y 1 G 1 1
@ 0 0 Y Y C C G G 1 1
Ct @tY
@
@ Y C G
@d 1 1 1
0 1 1 1
Y C G
FY FC FG
Example of a matrix representation induced by a (2,2) lattice-based VSSS.

Figure 21
2.6 Generation of Emerging images Via Tiled

Orders
In 2009, N. J. Mitra, H. Kuo Chu, T. Yee Lee, L. Wolf, H. Yeshurun and D.
Cohen-Or proposed a synthesis technique to generate emerging images of 3D
objects. Such images are gestalts with the property of to be detectable by
humans but difficult to process by computer vision algorithms. According
to Mitra et al [33], although such phenomenon popularized by the Gestalt
school has been well studied. The exact process of how we perceive such
objects is not known up to date. This fact allows to conclude that, it is
extremely challenging, if not impossible, to automate the recognition pro-
cess. For this reason, emergence can be used to make CAPTCHAs. Actu-
ally, these ideas have been used by Baird et al in [34] to generate dynamic
text strings-based CAPTCHAs, in this work, they also report an attack to
the famous NuCAPTCHA which has been considered the most secure and
usable CAPTCHA. In fact, the test of Baird et al is an improvement of
NuCAPTCHA.
Remark 2.35. According with L. Wolf et al [69], who have used emerging
images to generate Captchas. The main limitation with emerging images
seems to be the difficulty to create a large amount of recognizable models. In
this notes, we describe an algorithm introduced by M.A. Osorio et al which
allows to construct emerging images by using tiled orders.
63
Example of moving-image object recognition from NuCaptcha.
Figure 22
In the Osorio’s algorithm an image repository is said to be associated to a

01-tiled order if all of its images can be obtained after application of some re-
stricted admissible transformations. Such a system is named READT (recon-
struction of emerging images by restricted admissible transformations) [54].
The Osorio’s algorithm has two main stages:

1. Region recognition.
2. Image extraction.
Region recognition is the process for which it is chosen a subset I of points
of the poset P associated to the 01-tiled. I contains the pattern-image T
which can be obtained after some restricted transformations. Points m ∈ I
→ ← ↑
are called markers, they are denoted as mi,j , mi,j , mi,j or mi,j , in this case,
↓
arrows point out which entries in columns or rows of the tiled order must be
changed to extract the image.
Markers constitute a sequence M = {mi1 ,j1 , mi2 ,j2 , . . . , mik ,jk } which detects
the boundary of the pattern-image. To do that, the following procedure is
carried out:
1. If ms = mis ,js and mt = mit ,jt are consecutive markers with is < it
then both ms and mt belong to the boundary of T and have pixel value
p(m) = 1.
64
2. Compute the l1 -distance d = d(ms , Riws +1 ) and let mis +1,jk the pixel
or block of pixels such that d = d(ms , Riws +1 ) then mis +1,jk belongs to
the boundary of T . This process continues until the marker mit ,jt is
reached.
3. If there are two pixels or block of pixels m, m0 such that d = d(ms , m) =

d(ms , m0 ) then the vertex in the direction determined by the marker
must be chosen.
In the boundary detection process there are some pixels or block of pixels
belonging to the boundary of T with pixel value p(m) = 0, such kind of pixels
are said to be bridges with the main role of preserve the visual connected-
ness of the boundary. However, formally the boundary of T is an union of
components determined by the markers.
In the Image extraction stage the pattern-image is obtained after application

of the following restricted admissible transformations:
1. Row and column permutations.

← →
2. Additions of entries in columns at the left (right) of xi ( xi ).
↑
3. Additions of entries in rows above (below) of xi , (xi ).
↓
4. Completion (changing the value of an entry).
5. Image rotation.
Completion is a way of reducing the width of the poset associated to the

tiled order I. It is applied to the pixels within the region determined by the
markers. Via rotation the detection boundary algorithm is defined similarly
for any class of markers. As an example, Figures 23 and 24 show examples
of images associated to an array induced by a 01 order.
65
Subset of points induced by a 01-tiled order and image obtained after an
image extraction process.
Figure 23
Davinci’s drawings are models of images associated to the subset described.
Figure 24
66
2.7 A Mathematical Model for Emerging and
Multistable Images
In this section we describe a mathematical model introduced by Osorio’s et
al which uses triangulations to generate emerging images.
If M is a k-module then a model ν of an image repository D is a map (not
necessarily injective) ν : D → M which applies to each element I ∈ D an
element ν(I) = v, in this case, the element v can be seen as the image I and
v is said to be a M -model of I. Henceforth, if a model ν has been defined on
a repository D then we shall assume that each M -model ν(I) ∈ M is again
an image giving at least the same visual information provided by I. Actually,
model ν(I) is the image obtained from I after applications of some image
operations as noise addition, rotation, magnification, translation, reflection,
contraction or deformation.
Under the above circumstances tessellation, emerging image and multistable
image can be defined as follows:
Given an ordered repository of binary images:
(B, ) = {b1 , b2 , . . . , bk }
where
bi bj if image bi is embedded in image bj .
It is defined a X-model of B with the form:
S = (X, bi , χ, {Xbji } | 1 ≤ ji ≤ ti , 1 ≤ i ≤ k)
i
where for m, n fixed, X ⊂ Mm×n (Z2 ) is a finite k-vector space, χ ∈ X fixed.

In fact, entries of matrix χ are pixels induced by B (i.e., χij ∈ Z2 for each
1 ≤ i ≤ m, 1 ≤ j ≤ n).
B = {Xbji } is a finite generator of χ such that:
i
Xbri ∩ Xbsi = ∅ if r 6= s
Xbmi ∩ Xbnj = ∅ for all i, j, m, n
ti (2.7.21)
X
bi = aji Xbjii for any i aji ∈ Z2 1≤i≤k
ji =1
67
Identities (2.7.21) allow to write χ in the form:
th
k P
ajh Xbjhh
P
χ=
h=1jh =1
therefore, B is a tessellation of χ induced by B.
Examples of nested images. At right hand, four ordered images, i.e.,

b1 b2 b3 b4 .
Figure 25
If bi bj then bj can be written in the form
bj = bi + W j
where Wj is a complementary subspace of bi in bj . In such a case a chain

j−1
P
C = bi = bi1 ≺ bi2 ≺ · · · ≺ bij−1 ≺ bij = bj , with bij = bih , defines bj as an
h=1
emerging image. Therefore, an emerging image can be defined by a k-linear
representation of a finite chain.
If B 0 = {b01 , b02 , . . . , b0s } is an ordered repository of binary images and for some
h, 1 ≤ h ≤ s there exists br ∈ B with:
tr
X
b0h = ajr Xbjrr (2.7.22)
jr =1
68
then the system of binary images (br , b0h ) is a multistable image. This defini-
tion can be generalized by considering a collection B = {B h : 1 ≤ h ≤ p} of
binary images then if for each fixed B h it is defined a model νh : B h → M ,
M a fixed k-module. Thus a system of images (νh , νi ) is a multistable image
if there exists v ∈ M such that νh (I) = νj (I 0 ) = v for some images I ∈ B h ,
I 0 ∈ Bj .
Remark 2.36. Arguments used above to define emerging and multistable

binary images can be adapted for gray-level and RGB images.
Algorithm 2.7.1 allows to obtain models of type ν(I) from a binary-image

repository. Actually, the procedure can be generalized to gray-level and RGB
images.
System of emerging images obtained with the Osorio’s et al algorithm.
Figure 26
2.8 Digital Watermarking

Digital watermarks complement the encryption work to enable security in
the transmission of data through the internet (images, texts). A digital
watermark is a piece of information hiding into the content of a message in
such a way that it will be imperceptible to the human eye but easy to detect
by a computer. The advantage here is that the digital watermark cannot be
separated from the content.
According to I.D.M. Gaviria et al, the theory of algorithms of differentiation

is a theoretical framework for digital watermarking procedures. In this case
69
Algorithm 2.7.1:
1. Choose a binary-image repository D.
2. Define M = Mn (Z2 ) a Z2 -vector space of matrices with binary entries

and a suitable basis B = {bi | i ∈ S}, where S is a fixed finite set of
indices. bi ∩ bj = ∅ if i 6= j. That is, distinct elements in B have no
common entries.
3. Define χ ∈ M as follows:
(a) Choose a suitable large positive integer n = t1 × t2 , a

corresponding t1 × t2 -rectangular array A for its digits and a
security parameter t.
(b) Choose a sequence S of consecutive pairs of blocks (Bi , Bi+1 ),
Bi = (ni1 , ni2 , . . . , nili ), Bi+1 = (n(i+1)(1) , n(i+1)(2) , . . . , n(i+1)l(i+1) ),
nj ∈ A, i ≤ j ≤ l(i+1) , ni1 , n(i+1)1 6= 0, li + l(i+1) = t. t is the
minimum positive integer such that:
li
P lP
i+1
nih = n(i+1)m .
h=1 m=1
(c) Construct χ as a t1 × t2 -bitmap via the function γ : S → Zt2

defined in such a way that
γ(Bi , Bi+1 ) = (ci 1, ci2 , . . . , cili , d(i+1)1 , d(i+1)2 , . . . , dl(i+1)l(i+1) ), where
cij = 1 and d(i+1)s = 0, 1 ≤ j ≤ li , 1 ≤ s ≤ l(i+1) .
P
4. Write χ = bi .
i∈S
5. For each image I ∈ D write ν(I) =

P
bh .
h∈T ⊂S
the algorithm DVII is applied to representation of equipped posets defined

over an image algebra [48].
Now, we use representations of equipped poset P = {a ≺ c1 ≺ c2 } + {b}

and its corresponding derived poset in order to embed digital watermarks in
a system of images. To do that, we define the following representation over
70
Original image, digital watermark and watermarked image.
Figure 27
the pair (R, C), we assume the notation k(ei , ej ) = kei + kiej , where ei and
ej are images and k ∈ R is chosen in such a way that an user only can see
just one of the two images. In image algebra a system of images k(ei , ej ) is
interpreted as a sum of the form kei + kej , for any number k ∈ R.
c+ ⊗ d
3
b
@
@
⊗ c3 c+
2 ⊗ ⊗
@
@ c−
3
@
⊗ c2 c+
1 ⊗ @
⊗
@ c−
2
e
@
⊗ c1 a+
@
⊗
@ c−
1
VII @
⊗a d −−→ ⊗
@
b
(a,b) c−
0
Figure 28
Consider the representation of P over the image algebra given by the following
71
formulae:
X X
Ua = αi1 (ei1 , ej1 ) + βs1 hs1 ,
i1 s1
X X
Uc1 = Ua + αi2 (ei2 , ej2 ) + βs2 hs2 ,
i2 s2
X X
Uc2 = Uc1 + αi3 (ei3 , ej3 ) + βs3 hs3 ,
i3 s3
(2.8.23)
X X
Uc3 = Uc2 + αi4 (ei4 , ej4 ) + βs4 hs4
i4 s4
X 4 X
X
Ub = γi1 ei1 + βst hst .
i1 t=1 st
X X X X X
α i1 e i1 = αia1 eia1 + αic11 eic11 + αic12 eic12 + αic13 eic13 .
(2.8.24)
i1 ia
1
c
i11 i12
c c
i13
In this case, subspace Ub contains digital watermarks of images ejr , 0 ≤

r ≤ 4 (c0 = a). Algorithm of differentiation VII allows to extracted such
watermarks, to do that each subspace Uc0+ is defined as follows:
i
3 X
X
Ua0 + = αic1r eic1r
r=0 ic1r
X X
Uc0+r = Ua0 + + αir (eir , ejr ) + βsr hsr , 1 ≤ r ≤ 3, (2.8.25)
ir sr
X
Uc0−r = βsr hsr , 0 ≤ r ≤ 4.
sr
Coefficients in Ua+ are chosen in such a way that the only image that an
user can see is one of thePtype eic1r , 0 ≤ r ≤ 3. On the other hand, digital
watermarks of the form βst hst are given by subspaces Uc0− . And images
r
P st
αic1r eic1r , 0 ≤ r ≤ 3 are the recovered digital watermarks from subspace
ic1r
Ucr . Since this procedure is independent of domain, it can be applied in
spatial and frequency domain. In the appendix some examples of MATLAB
routines are implemented to embed system of digital watermarks into nested
emerging images (see Figure 30).
72
2.9 Homological Persistence Theory
In this section we describe some applications of the homological persistence
in number theory [21, 28].
Perhaps, Betti numbers and the fundamental group are the main invariants
dealing with applications of the algebraic topology in statistical data, however
these invariants are sensible to discontinuous changes. According to De Silva
[21] the challenge in topological statistics is to construct invariants which
enjoy the discriminatory power of the classical invariants but which have
robustness for statistical use. persistent homology is nowadays the most
powerful tool to tackle this problem.
The workflow runs as follows:
1. The input is a point cloud, that is, a finite subset of some euclidean
space or more generally a finite metric space (actually, it is not manda-
tory that the input be a metric space).
2. After an initial filtering step (to remove undesirable points or to focus

on high density regions of the data).
3. A set of vertices is selected from the data, and a simplicial complex S

is built on that vertex set, according to some prearranged rule.
In practice the simplicial complex depends on a coarseness parameter and

we have is a nested family {S }∈[0,∞] . If we discretise the variable to a
finite set of values, the family of simplicial complexes can be thought as a
diagram of spaces:
S1 → S2 → · · · → Sn
where the arrows represent the inclusion maps.
If we apply the k-dimensional homology functor Hk (−; k) with coefficients

in a field k, this becomes a diagram of vector spaces and linear maps, i.e., a
linear representation of the Dynkin diagram An linearly oriented:
V1 → V2 → · · · → Vn
73
where Vi = Hk (Si ; k). Such a representation is said to be a persistence
module.
It holds that indecomposable persistence modules are intervals, besides, any

persistence module M can be expressed as sum of intervals of the form:
M
M= I[b, d]mbd . (2.9.26)
1≤b≤d≤n
Escolar, Hiraoka, Di Silva and Carlson generalized these results to the so-
called zig-zag modules and commutative ladders.
We recall that according to Escolar and Hiraoka a ladder of length n is defined

by two quivers of type An of the same orientation o1 o2 . . . on together with
an arrow i → i0 for every i = 1, 2 . . . , n. The commutative ladder CL(τn ) is
defined as the associative algebra CL(τn ) = CL(τn )/I where I is generated by
commutative relations αβ = γδ, γ 0 α0 = δ 0 β 0 , with β : i → i+1(β 0 : i+1 → i),
γ : i0 → (i + 1)0 (γ 0 : (i + 1)0 → i0 ), α : i0 → i, δ : (i + 1)0 → i + 1. The
following are results regarding commutative ladders.
Theorem 2.37. The commutative ladders CL(τn ) of length n ≤ 4 are representation-

finite for arbitrary orientations, for n ≥ 5, CL(τn ) are representation-infinite
for arbitrary orientations.
The following figure shows an example of a ladder and its corresponding

Auslander-Reiten quiver:
74
3d - d
1 4d
6 6 6
d - d d
5 2 6
011 100 000 001

> 011 Z
> 000 Z
> 110 Z
> 001 Z

ZZ ZZ ZZ ZZ
110 011 ~
111 ~
100 ~
001 ~
000
> 000 Z Z Z 010 Z 011 110 111 Z 001
ZZ
>
ZZ
>
ZZ
>
ZZ
>
ZZ >

~ ~ ~ ~ ~
010 - 010 - 121 - 111 - 111 - 000 - 111 - 111 - 101 000
000 010 010 000 010 010 121 111 111 111
Z >
Z >
Z
> Z >
Z
> Z
~ 011
ZZ ZZ
~ 110 ZZ
~ 000 ZZ
~ 001 ZZ
~ 100 Z~
Z 000
000 010 Z 000 Z 011 Z 111 Z 100

>
>
>
>
~ 110
ZZ ~ 001
ZZ ~ 000
ZZ ~ 100
ZZ
110 000 011 100
Figure 29
Indecomposable projective modules are: Indecomposable injective modules are:

P(1)= 0 1 0 I(1)= 1 1 1
000 111
P(2)= 0 1 0 I(2)= 0 0 0
010 111
P(3)= 1 1 0 I(3)= 1 0 0
000 100
P(4)= 0 1 1 I(4)= 0 0 1
000 001
P(5)= 1 1 0 I(5)= 0 0 0
110 100
P(6)= 0 1 1 I(6)= 0 0 0
011 001
75
Persistent modules have been designed to carry topological information about
a data set at many different scales this information can be extracted from
persistence diagram of a persistence module on an An which is a multiset of
the form {(b, d) ∈ N2 | 1 ≤ b ≤ d ≤ n} such that the multiplicity of (b, d) is
mbd from the indecomposable decomposition.
The Gauss’s theorem regarding sums of three triangular number and the
Legendre-Gauss theorem on sums of three square numbers allow to analyze
points in N2 as a set of data. These results are useful to describe the structure
of some persistent modules.
Algorithm 2.9.1:
1. (Clustering) Define on points p = (i, j) ∈ N2 an equivalence relation
by associating to each point p = (i, j) the nonnegative number
np = i + j. In such a case two points p = (i, j) and p0 = (i0 , j 0 ) are
equivalent (belong to the same class o cluster, p) if and only if
np = np 0 .
2. We consider the metric space induced by the quiver Q (with (0, 0) as

starting point) described before endowed with the L1 -metric, in order
to connect clusters, we select balls of radius 1 , 2 , . . . where k = k
and k is a positive integer.
3. A hole is a region R ⊂ N2 determined by Q such that at least one

cluster induced by a point p ∈ R does not belong to an admissible
path.
As a consequence of this procedure we have the following results:

Theorem 2.38. If I is a persistent interval associated to the homology H1
determined by Algorithm 2.9.1 then I = ∅.
Proof. For every integer n it holds that there exist triangular numbers t1 , t2
and t3 such that n = t1 + t2 + t3 .
Theorem 2.39. If I is a persistent interval associated to the homology H1
determined by Algorithm 2.9.1 with np = i + 2j then I = [bp , ∞) where bp is
induced by points (2α−1 )(72α − 3 + m(22(α+1) − 1), 32α + 2m) ∈ N2 , α > 0,
m ≥ 0.
76
Proof. Any integer number of the form 2α (8m + 7), α, m ≥ 0 cannot be
written as a sum of three square numbers.
77
Example of a digital watermarked system based on emerging images and
DVII.
Figure 30
78
Chapter 3
Appendix
3.0.1 DVII in MATLAB

In this section a simply MATLAB program illustrates the algorithm of dif-
ferentiation VII as a way to embed digital watermarks in a system of nested
images.
beta = 0.01;
gamma = 0.01;
delta = 1;
epsilon = 0.01;
tau = 0.01;
iota = 0.01;
psi = 0.01;
kappa = 1;
lambda = 1;
mu = 0.01;
eta = 0.01;
nu = 0.01;
xi = 1;
ji = 1;
sigma = 1;
zeta = 1;
xi b = 1;
gamma b = 1;
nu b = 1;
79
alpha = 1;
alpha a = 0.01;
A = imread(0 Imagen37scan.jpg 0 );
T = imread(0 DV II.png 0 );
R 1 = imread(0 Imagen5.jpg 0 );
R 4 = imread(0 Imagen7scan.jpg 0 );
B = imread(0 doodle6.jpg 0 );
C = imread(0 ty.jpg 0 );
D = imread(0 Imagen1.jpg 0 );
W = imread(0 Imagen4scanbl.jpg 0 );
V = imread(0 Imagen22.jpg 0 );
F 1 = imread(0 zvr.jpg 0 );
F 2 = imread(0 zk.jpg 0 );
E = imlincomb(zeta, A);
F = imlincomb(0.01, B);
G = imlincomb(1, C);
H = imlincomb(1, D);
X = imlincomb(sigma, W );
A = imlincomb(1, V );
F 3 = imlincomb(1, F 1);
T 1 = imresize(T, [740, 740]);
R 5 = imresize(R 1, [256, 256]);
R 6 = imresize(R 2, [256, 256]);
R 7 = imresize(R 3, [256, 256]);
R 8 = imresize(R 4, [256, 256]);
I = imresize(E, [256, 256]);
J = imresize(F, [256, 256]);
K = imresize(G, [256, 256]);
L = imresize(H, [256, 256]);
Y = imresize(X, [256, 256]);
B = imresize(A, [256, 256]);
F 5 = imresize(F 3, [256, 256]);
F 6 = imresize(F 4, [256, 256]);
a = imlincomb(xi, R8 , ji, J);
80
a 1 = imlincomb(alpha, a);
a 2 = imresize(a 1, [256, 256]);
c 1 = alpha a ∗ a 2 + beta ∗ K + gamma ∗ (I) + delta ∗ R 5;
c 2 = epsilon ∗ a 2 + tau ∗ (I) + iota ∗ L + psi ∗ K + nu ∗ Y + kappa ∗ R 6;
F 9 = imlincomb(1, c 2);
G = imresize(F 9, [256, 256]);
O 1 = eta ∗ F 9 + lambda ∗ R 7 + mu ∗ F 5;
O 2 = imlincomb(1, O 1);
c 3 = imresize(O 2, [256, 256]);
b = xi b ∗ J + gamma b ∗ I + nu b ∗ Y ; beta
f igure, imshow(Y, []), title(0 c+ 2b0 )
f igure, imshow(I, []), title(0 c+ 1b0 )
f igure, imshow(a 1, []), title(0 a0 )
f igure, imshow(c 1, []), title(0 c 10 )
f igure, imshow(c 2, []), title(0 c 20 )
f igure, imshow(c 3, []), title(0 c 3(system of watermarked f rames)0 )
f igure, imshow(T 1, []), title(0 DV II 0 )
3.0.2 Wavelet Transformation and Singular Value De-

composition
In this section a MATLAB routine interprets DVII as a system of digital
watermarks in frequency domain. To do that, a singular value decomposition
is applied after a wavelet procedure in order to embed a digital watermark
alpha = 0.55;
beta = 0.75;
R1 = imread(0 cameraman.tif 0 );
R2 = imread(0 rice1.jpg 0 );
R3 = imread(0 alberti.jpg 0 );
A = imread(0 kirichenko.jpg 0 );
T = imread(0 32dee.jpg 0 );
B = imread(0 zavadskij.jpg 0 );
C = imread(0 ty.jpg 0 );
D = imread(0 iue.jpg 0 );
81
W = imread(0 zavadskij and bautista.jpg 0 );
V = imread(0 Imath.jpg 0 );
F 1 = imread(0 zvr.jpg 0 );
F 2 = imread(0 zk.jpg 0 );
E = imlincomb(1, A);
F = imlincomb(0.01, B);
G = imlincomb(1, C);
H = imlincomb(1, D);
X = imlincomb(1, W );
A = imlincomb(1, V );
S1 = imresize(R1, [256, 256]);
S2 = imresize(R2, [256, 256]);
S3 = imresize(R3, [256, 256]);
I = imresize(E, [256, 256]);

L = imresize(H, [256, 256]);
Y = imresize(X, [256, 256]);
B = imresize(A, [256, 256]);
F 5 = imresize(F 3, [256, 256]);
F 6 = imresize(F 4, [256, 256]);
T 6 = imresize(T, [256, 256]);
[LL1, HL1, LH1, HH1] = dwt2(R1,0 haar0 );

[LL2, HL2, LH2, HH2] = dwt2(LL1,0 haar0 );
[LL3, HL3, LH3, HH3] = dwt2(D,0 haar0 );
p = size(LL2);
t = size(LL4);
r = size(LL6);
[U y, Sy, V y] = svd(LL2);
q = size(Sy);
[ny, ay, vy] = svd(LL4);
z = size(ay);
82
[oy, by, py] = svd(LL6);
h = size(by);
I w = imread(0 zk.jpg 0 );
I w = Iw (:, :, 1);
I1 w = imresize(I w, p);
J w = imread(0 zvr.jpg 0 );
J w = J w(:, :, 2);
J1 w = imresize(J w, z);
f igure(15); imshow(I1 w); title(0 original watermark7a0 );
f igure(13); imshow(J1 w); title(0 original watermark70 );
[U w, Sw, V w] = svd(double(I1 w));

[nw, sw, vw] = svd(double(J1 w));
Smark = Sy + alpha ∗ Sw;

nmark = ay + alpha ∗ sw;
LL2 1 = U y ∗ Smark ∗ V y 0 ;
LL2 2 = ny ∗ nmark ∗ vy 0 ;
LL1 1 = idwt2(LL2 1, HL2, LH2, HH2,0 haar0 );

I 1 = idwt2(LL1 1, HL1, LH1, HH1,0 haar0 );
LL2 3 = idwt2(LL2 2, HL4, LH4, HH4,0 haar0 );
J 1 = idwt2(LL2 3, HL3, LH3, HH3,0 haar0 );
f igure(3); imshow(uint8(I 1)); title(0 W atermarked Image0 );
f igure(10); imshow(uint8(J 1)); title(0 W atermarked Image100 );
[LL1 wmv, HL1 wmv, LH1 wmv, HH1 wmv] = dwt2(I 1,0 haar0 );
[LL2 wmv, HL2 wmv, LH2 wmv, HH2 wmv] = dwt2(LL1 wmv,0 haar0 );
[U y wmv, Sy wmv, V y wmv] = svd(LL2 wmv);
Swrec = (Sy wmv − Sy)/alpha;
W M y = U w ∗ Swrec ∗ V w0 ;
[LL3 wmv, HL3 wmv, LH3 wmv, HH3 wmv] = dwt2(J 1,0 haar0 );
[LL4 wmv, HL4 wmv, LH4 wmv, HH4 wmv] = dwt2(LL3 wmv,0 haar0 );
[nyw mv, sy wmv, vy wmv] = svd(LL4 wmv);
83
swrec = 2 ∗ alpha ∗ (sy wmv − ay);
W M y1 = nw ∗ swrec ∗ vw0 ;
f igure(4); imshow(uint8(W M y)); title(0 Extracted W atermark 0 );

f igure(12); imshow(uint8(W M y1)); title(0 Extracted W atermark120 );
f igure(5); imshow(uint8(LL1 1)); title(0 W avelet20 );
f igure(6); imshow(H); title(0 Original Image0 );
f igure(6); imshow(R1); title(0 Original Image(a)0 );
84
Bibliography
[1] D.M. Arnold, Abelian Groups and Representations of Finite Partially Ordered Sets,
CMS Books in Mathematics, vol. 2, Springer, 2000. 244 p.
[2] I. Assem, C. Reutenauer, and D. Smith, Friezes, Adv. Math 225 (2010), 3134–3165.
[3] I. Assem, D. Simson, and A. Skowroński, Elements of the Representation Theory of
Associative Algebras, Cambridge University. Press, 2006.
[4] M. Auslander, I. Reiten, and Sv. Smalo, Representation Theory of Artin Algebras,
Cambridge Univ. Press, Cambridge, 1995.
[5] C. Blundo, A. de Santis, and D. R. Stinson, On the Contrast in Visual Cryptography
Schemes, Journal of Cryptology 12 (1999), no. 4, 261–289.
[6] H.S. Baird, M.A. Moll, and S.Y. Wang, ScatterType: A legible but hard-to-segment
CAPTCHA, ICDAR’05: Proc. 8th Int. Conference on Document Analysis and Recog-
nition, IEEE Computer Society (2005), 935–939. doi: 10.1109/ICDAR.2005.205.
[7] R. Bautista, On algebras of strongly unbounded type, Comm. Math. Helv. 60 (1985),
392-399.
[8] R. Bautista, P. Gabriel, A.V. Roiter, and L. Salmeron, Representation-finite algebras
and multiplicative bases, Invent. Math. 81 (1985), 277-285.
[9] G. Bravo, H. Giraldo, and A.M. Cañadas, Integer Sequences Arising From Auslander-
Reiten Quivers of Some Hereditary Artin Algebras, JAA (submitted).
[10] P. Bubenik and J. Scott, Categorification of persistent homology, arXiv: 1205.3669 3
(2014).
[11] I. Çanakçi and R. Schiffler, Cluster algebras and continued fractions, Compositio
Mathematica 154 (2018), 565-593.
[12] G. Carlsson and V. De Silva, Zig Zag persistence, Found Comp Math 10 (2010),
367-405.
[13] A. M. Cañadas, H. Giraldo, and P.F.F. Espinosa, Categorification of some integer
sequences, FJMS 92 (2014), no. 2, 125–139.
[14] , Categorification of some integer sequences and Higher Dimensional Parti-
tions, FJMS 92 (2014), no. 2, 125–139.
85
[15] A.M. Cañadas, P.F.F. Espinosa, and I.D.M. Gaviria, Categorification of some integer
sequences via Kronecker modules, JPANTA 38 (2016), no. 4, 339-347.
[16] A.M. Cañadas, H. Giraldo, and G.B. Rios, On the number of sections in the
Auslander-Reiten quiver of algebras of Dynkin type, FJMS 101 (2017), no. 8, 1631-
1654.
[17] A.M. Cañadas, H. Giraldo, and R.J. Serna, Matrix problems induced by visual cryp-
tography schemes, FJMS 102 (2017), no. 6, 1223-1241.
[18] A.M. Cañadas, V. Cifuentes, and A.V. Gonzalez, On the number of two-point an-
tichains in the powerset of an n-element set ordered by inclusion, JPANTA 102
(2016), no. 3, 1223-1241.
[19] A. M. Cañadas and N. P. P. Vanegas, Extended Visual Cryptography Scheme
with an Artificial Cocktail Party Effect, IEEE Digital Library (2011). DOI:
10.1049/ic.2011.0114.
[20] A. M. Cañadas, N. P. P. Vanegas, and M.H. Quitian, Visual cryptography
schemes based in k-linear maps, LNCS; Springer-Verlag 8389 (2014), 288-302. DOI:
10.1007/978-3-662-43886-2 21.
[21] G. Carlson and V. de Silva, Zigzag persistence, Foundations of computational math-
ematics 10 (2010), 367-405. DOI: 10.1007/978-3-662-43886-2 21.
[22] S-K. Chen, A visual cryptography based system for sharing multiple secret images,
Proc. 7th WSEAS int. conf. on Signal Proccesing, Computational Geometry and
Artificial Vision, Athens, Greece 40 (2007), 117–121.
[23] S. Cimato and C.-N. Yang, Visual Cryptography and Secret Image Sharing, CRC
Press, 2011.
[24] D. Cvetković, Applications of Graph Spectra: An Introduction to the Literature,
preprint 101.
[25] B.A. Davey and H.A. Priestley, Introduction to Lattices and Order, 2nd ed., Cam-
bridge University Press, 2002.
[26] M. Dokuchaev, N. Gubareni, V.M. Futorny, M.A. Khibina, and V.V. Kirichenko,
Dynkin diagrams and spectra of graphs, The São Paulo Journal of Mathematical Sci-
ences 7 (2013).
[27] S. Droste, New Results on Visual Cryptography, Advances in cryptology - CRYPTO
’96. Lecture Notes in Computer Science 1109 (1998), 401–415.
[28] E.G. Escolar and Y. Hiraoka, Persistence modules on commutative ladders of finite
type, arXiv: 1404.7588 2 (2015).
[29] P. Fahr and C. M. Ringel, A partition formula for Fibonacci numbers, Journal of
integer sequences 11 (2008), no. 08.14.
[30] E. Fernández, Gráficas y matrices de adyacencia, Actas del VII congreso Dr. Antonio
A.R. Monteiro (2005), 151-179.
86
[31] P. Gabriel and A.V. Roiter, Representations of Finite Dimensional Algebras, Algebra
VIII, Encyclopedia of Math. Sc., vol. 73, Springer-Verlag, 1992. 177p.
[32] P. Gabriel, Représentations indécomposables des ensemblés ordonnés, Semin. P.
Dubreil, 26 annee 1972/73, Algebre, Expose 13 (1973), 301–304.
[33] N. J. Mitra, H. K. Chu, T. Y. Lee, L. Wolf, H. Yeshurun, and D. Cohen-Or, Emerging
Images, ACM Transactions on Graphics 28 (December, 2009), no. 5, 8 pp.
[34] Y. Xu, G. Reynaga, S. Chiasson, J-M. Frahm, F. Monrose, and P. Van Oorschot, Se-
curity and usability challenges of moving-object CAPTCHAs: decoding codewords in
motion, Proceeding Security’12 Proceedings of the 21st USENIX conference on Secu-
rity symposium. IEEE TDSC 11 (2014), no. 5, 480–493. doi: 10.1109/TDSC.2013.52.
[35] J.A. De la Peña, Representaciones de Algebras y Ecuaciones Diferenciales, Facultad
de Ciencias, Universidad de los Andes Merida (1999). Decimosegunda escuela vene-
zolana de Matemáticas.
[36] P. Fahr and C. M. Ringel, Categorification of the Fibonacci numbers, Journal of integer
sequences 15 (2012), no. 12.2.1.
[37] S. Fomin, L. Wiliams, and A. Zelevinsky, Introduction to cluster algebras, Arxiv 2
(2017), no. 1608.05735.
[38] B. Fontaine and P.G. Plamondon, Counting friezes, preprint.
[39] M. Hirschhorn, On recurrences of Fahr and Ringel arising in graph theory, Journal
of integer sequences 12 (2009), no. 09.6.8.
[40] M. Hirschhorn and J. Sellers, Partitions into three triangular numbers, Australasian
Journal of Combinatorics 30 (2004), 307-318.
[41] A. Hubery and H. Krause, A categorification of non-crossing partitions, arXiv:
1310.1907.
[42] C. Ingalls and H. Thomas, Noncrossing partitions and representations of quivers,
arXiv: 0612219.
[43] J.B. Feng, H.C. Wu, C.S. Tsai, and Y.P. Chu, A new multi-secret images sharing
scheme using Lagrange’s interpolation, The Journal of Systems and Software 76
(2005), no. 3, 327–339.
[44] M. Hazewinkel, N. Gubareni, and V.V. Kirichenko, Algebras, Rings and Modules,
First Edition, Vol. 2, Springer, 2007.
[45] A. Klein and M. Wessler, Extended Visual Cryptography Schemes, Information and
Computation 205 (2007), no. 5, 716–732.
[46] H. Koga and H. Yamamoto, Proposal of a lattice-based visual secret sharing scheme
for color and gray scale images, IEICE Trans. Fundamentals E 81-A (1998), no. 6,
1262–1269.
[47] A. Klein and M. Wessler, Extended Visual Cryptography Schemes, Information and
Computation 205 (2007), no. 5, 716–732.
87
[48] I.D. Marin-Gaviria, The Auslander- Reiten Quiver of Equipped Posets of Finite
Growth Representation Type, Preprint, National University of Colombia, 2017.
[49] M. Nakajima and Y. Yamaguchi, Extended Visual Cryptography for Natural Images,
WSCG 10 (2002), no. 2, 303–310.
[50] M. Naor and A. Shamir, Visual Cryptography, Advances in Cryptology Eurocrypt’94.
Lecture Notes in Computer Science 950 (1995), 1–12.
[51] L.A. Nazarova and A.G. Zavadskij, Partially ordered sets of finite growth, Function.
Anal. i Prilozhen., 19 (1982), no. 2, 72–73 (in Russian); English transl., Functional.
Anal. Appl., 16 (1982), 135–137.
[52] X. Peng and J. Zhang, Cluster algebras and Markoff numbers, CaMUS 3, 19–26.
[53] M.I. Platzeck and E. Fernández, A note on the spectral properties of cluster algebras,
Arxiv 1 (2010), 1011.5520.
[54] M.A. Osorio, Pruebas de Interacción con Humanos Basadas en Imágenes Emergentes
y Multiestables, Preprint, National University of Colombia, 2017.
[55] J. Propp, The combinatorics of frieze patterns and Markoff numbers, arXiv: 05116334
(2008).
[56] C.M. Ringel, Tame Algebras and Integral Quadratic Forms, LNM, vol. 1099, Springer-
Verlag, 1984. 1–371.
[57] C. M. Ringel, Catalan combinatorics of hereditary algebras, arXiv: 1107.1858 2 (2011),
1-12.
[58] N.M. Riviere, Recursive formulas on free distributive lattices, Journal of Combinato-
rial Theory 5 (1968), 229-234.
[59] A. Ross and A. A. Othman, Visual Cryptography for Face Privacy, SPIE Conference
on Biometrics Technology for Human identification VII (April 2010).
[60] A. De Santis, A.L. Ferrara, and B. Masucci, Visual cryptography schemes with revers-
ing, In Visual Cryptography and Secret Image Sharing (2011), 255–278. Edited by S.
Cimato and C.-N. Yang, CRC Press.
[61] D. Simson, Linear Representations of Partially Ordered Sets and Vector Space Cate-
gories, Gordon and Breach, London, 1992.
[62] R. Schiffler, A geometric model for cluster categories of type Dn , J. Alg. Comb 29
(2008), no. 1, 1–21.
[63] B. Schröder, Ordered Sets An Introduction, Birkhäuser, Boston, 2002.
[64] L.D. Servi, Nested square roots of 2, MAA 110 (2003), no. 4, 326-330.
[65] S.J. Shyu, S.Y. Huang, Y.K. Lee, R.Z. Wang, and K. Cheng, Sharing multiple secrets
in visual cryptography, Pattern Recognition 40 (2007), 3633–3651.
[66] D. Simson, Linear Representations of Partially Ordered Sets and Vector Space Cate-
gories, Gordon and Breach, London, London, 1992.
88
[67] C.S. Tsai, C.C. Chang, and T.S. Chen, Sharing multiple secrets in digital images, The
Journal of Systems and Software 64 (2002), no. 2, 163–170.
[68] L. Williams, Cluster algebras; An introduction, Arxiv 3 (2013), no. 1212.6263.
[69] L. Wolf, A. Zaritsky, and E. Rotman, EI CAPTCHA;
Emerging images CAPTCHA description page (2010).
http://www.cs.tau.ac.il/haimnach/eicaptcha/description/index.php.
[70] C.C. Wu and L.H. Chen, A study on visual cryptography, Master Thesis, National
Chiao Tung University, Taiwan, R.O.C (1998).
[71] H.C. Wu and C.-C. Chang, Sharing visual multi-secrets using circle shares, Comput.
Stand. Interfaces 134 (2005), no. 28, 123–135.
[72] T.J. Yusun, Dedekind Numbers and Related Sequences, Simon Fraser University, Lon-
don, 2011. Master Thesis.
[73] A.G. Zavadskij, The Auslander-Reiten quiver for posets of finite growth, Topics in
Algebra, Banach Center Publ. 26 (1990), Part 1, 569–587.
[74] , Tame equipped posets, Linear Algebra Appl. 365 (2003), 389–465.
[75] , Equipped posets of finite growth, Representations of Algebras and Related
Topics, AMS, Fields Inst. Comm. Ser. 45 (2005).
[76] , On two point differentiation and its generalization, Algebraic Structures and
their Representations, AMS, Contemporary Math. Ser. 376 (2005).
[77] A.G. Zavadskij and V.V. Kiričenko, Semimaximal rings of finite type, Mat Sbornik
103 (1977), 323-345 (in Russian).
[78] , Torsion-free modules over primary rings, Journal of Soviet Mathematics 11
(1979), no. 4, 598-612.
[79] N.J.A. Sloane, On-Line Encyclopedia of Integer Sequences, The OEIS Foundation,
http://oeis.org/A052558, A132262, A110122, A00732.
89

Lecture CIMPA School

Uploaded by

Copyright:

Available Formats

Lecture CIMPA School

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture CIMPA School

Uploaded by

Copyright:

Available Formats

Some Applications of the Theory of

Agustı́n Moreno Cañadas and Isaı́as David Marı́n Gaviria

2 Some Applications of the Theory of Representation of Posets

The impact of the theory of representation of algebras and in particular of

Categorification of integer sequences is an example of the application of the

The Auslander-Reiten quiver of quivers 2-Kronecker and 3-Kronecker can

Recently the theory of homological persistence has arose as a new source of

In this chapter we describe the theoretical setting necessary to describe ap-

Let P be an ordered set and let x, y ∈ P we say x is covered by y if x < y

Let P be a finite ordered set. We can represent P by a configuration of

(1) To each point x ∈ P, associate a point p(x) of the Euclidean plane R2 ,

(3) Carry out (1) and (2) in such a way that

(a) if x < y, then p(x) is lower than p(y),

A configuration satisfying (1)-(3) is called a Hasse diagram or diagram of P.

An ordered set C is called a chain (or a totally ordered set or a linearly

1.2 Representations of Posets

where U0 is a finite dimensional k-vector space, Ux is a subspace of U0 for

A map f : U → V ∈ rep P between representations U, V ∈ rep P is a k-linear

Attached to each representation U there exists its matrix presentation M =

M = Mx1 Mx2 ... Mxn (2)

with entries in k, partitioned horizontally into n = |P| (vertical) blocks (also

Two representations A and B of a poset (P, ) are isomorphic if and only

1. elementary transformations of rows of the whole matrix A;

2. elementary transformations of columns within each vertical strip;

3. additions of columns of a strip Ai to columns of a strip Aj if i  j in

1.3 Algorithms of Differentiation

The following is the definition of the algorithm of differentiation with respect

Let (P, ≤) be an ordinary poset then a maximal point x ∈ max P is suitable

The algorithm of differentiation with respect to a suitable point b ∈ max P

with a partial order induced by P \ b and N

P = N + {6 < 7 < b} P0b = N

The functor of differentiation F b with respect to a maximal point b is defined

induces an equivalence between quotient categories

and mutually inverse bijections between classes of indecomposable objects

Ind P \ [N ]  Ind P0b .

Often, posets K1 , . . . , K5 are called the Kleiner’s critical.

Theorem 1.1 was also used by Fernandéz et al to categorify the sequence

The algorithm of differentiation with respect to a suitable pair of points

A pair of points (a, b) in a poset P is suitable for differentiation DI if poset

where C = c1 < c2 < · · · < cn is a chain and for each i, 1 ≤ i ≤ n, points

being considered as a subposet of the free lattice generated by P, we write

U0 -subspaces in a representation of the derived poset P0(a,b) are given by the

Ux0 = Ux for the remaining points x ∈ P0

DI allowed to Zavadskij and Nazarova to give a finite growth representation

As an example, we consider the equipped poset P whose diagram is illustrated

Such partitioned matrices M are called matrix representations of P over

For example the representation (1.3.6) is a sincere representation of the poset

1.3.1 Some indecomposable objects

If P is an equipped poset and A ⊂ P then P (A) = P (min A) = (F ; Px |

If a, b ∈ P⊗ then T (a) and T (a, b) denote indecomposable objects with matrix

On the other hand, representation T (a, b) may be described in such a way

If a ∈ P⊗ and B ⊂ P is a subset completely weak such that a ≺ B then

In particular, T (a, ∅) = T (a).

where {a ≺ c1 ≺ · · · ≺ cn } is a completely weak chain incomparable with

If P is an equipped poset with a pair of points (a, b), VII-suitable then

where a− ∈ (P0 )⊗ , a+ ∈ (P0 )◦ , C − and C + are completely

(2) Order relations in P \ {a + C}, together with the relations described

Figure 6 shows the diagrams of P and the corresponding P0(a,b) .

Uc0+ = Uci + F (Ua ),

Ux0 = Ux for the remaining points x ∈ P0(a,b) ,

Two representations A and B of a poset (P, ) are isomorphic if and only

3. additions of columns of a strip Ai to columns of a strip Aj if i j in

Ind P \ [N ] Ind P0b .

where a− ∈ (P0 )⊗ , a+ ∈ (P0 )◦ , C − and C + are completely

τ.1 -τ.2 .τ3