Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Applied Mathematical Sciences, Vol. 13, 2019, no. 17, 845 - 857
HIKARI Ltd, www.m-hikari.com
https://doi.org/10.12988/ams.2019.98115
Project Management and Optimization
Processes Choices to Maximize Resource
Allocation Results
Giuseppe Caristi
Department of Economics
University of Messina, Italy
Vera Fiorani
R.F.I., Italy
Sabrina Lo Bosco
UniPegaso, Italy
Alberto Vieni
UniPegaso, Italy
This article is distributed under the Creative Commons by-nc-nd Attribution License.
Copyright c© 2019 Hikari Ltd.
Abstract
Integer Linear Programming (ILP) deals in a systematic way with
the minimization (maximization) problem of a linear function with sev-
eral variables (some also of a random nature), subject to equality con-
straints and linear inequality and to the constraint that one or more
846 Giuseppe Caristi, Vera Fiorani, Sabrina Lo Bosco and Alberto Vieni
variables has integer values. This mathematical approach is essential to
address and solve a very large number of real problems, typical of the ap-
plied sciences, when there is an indivisibility of the good to be produced
or of the resource to be used and thus the consequent need to repre-
sent the problem through models of Linear Programming with integer
variables. The applications of the method concern many applications of
modern society and extend from industrial analysis for the distribution
of goods and the sequencing of productive activities, to economic prob-
lems aimed at the optimal management of a securities portfolio, to those
of planning and optimal planning of public investments, to the problems
inherent in biology, high energy physics, engineering and robotics. in
this paper we consider a cutting plane problem for an irregular lattice
represented in fig. 2.
1 Introduction
In this work a mathematical methodology is developed that refers to the ge-
ometric analysis of the variables in a suitable Rm space, with a polyhedral
approach, useful for optimizing strategic choices in public decision-making
processes, ie in corporate ones. The criterion also lends itself to being ap-
plied in the various sectors of the business economy, also to solve problems
of resource allocation (investments, allocation of personnel to strategic areas,
etc.). Through specific geometric probability techniques elaborated below, we
tried to provide the decision maker with the formal basis of the strategic choice
aimed at optimizing the overall results of the action under consideration, tak-
ing into account both the different constraints posed by the problem, and
the context scenario, proceeding from a Life-Cycle Cost Analysis perspective.
Project Management and optimization processes choices to maximize resource
allocation results. In this work a mathematical methodology is developed that
refers to the geometric analysis of the variables in a suitable Rm space, with a
polyhedral approach, useful for optimizing strategic choices in public decision-
making processes, ie in corporate ones. The criterion also lends itself to being
applied in the various sectors of the business economy, also to solve problems
of resource allocation (investments, allocation of personnel to strategic areas,
etc.). Through specific geometric probability techniques elaborated below, we
tried to provide the decision maker with the formal basis of the strategic choice
Project management and optimization processes 847
aimed at optimizing the overall results of the action under consideration, tak-
ing into account both the different constraints posed by the problem, and the
context scenario, proceeding from a Life-Cycle Cost Analysis perspective. In
practice, it is often necessary to tackle problems in which the decision variables
represent indivisible quantities (such as the number of trains on a relationship
OD of a railway network, of people assigned to specific company tasks, etc.) or
others characterized by the need to choose between a finite number of different
alternatives. In the latter case, in particular, Linear Programming problems
01 will arise, ie problems in which the variables are binary and take the value
0 or 1. The methodology proposed here allows the construction of a math-
ematical model based on PLI integrated linear programming and studies the
so-called ”cutting planes” in Rm: the resolution of the problem is carried out
by eliminating the ”integrity” constraint of the variables. If the solution found
at a generic step of the process is continuous, then it is also excellent for the
original problem with integer variables; otherwise, the method continues by
generating characteristic cutting planes, or new inequalities to add to the re-
laxed problem, which must render the current solution inadmissible and not
discard any of the solutions of the original problem in the study (see fig. 1)
Programmazione Lineare Intera PLI.1
Programmazione Lineare Intera (PLI)
zPLI = min cTx
Ax ∏ b
x ∏ 0
x intero
• vincoli di interezza non lineari: es. sin(ºx) = 0
• Rimuovendo il vincolo di interezza
) PL (rilassamento continuo di PLI), tale che
zPL ∑ zPLI (lower bound per zPLI)
Dim. si cerca il minimo in un insieme piu grande. 2
-
6
d d d d d dd d d d dd d d d dd d d d d
⇢⇢⇢⇢⇢⇢⇢⇢
⇢⇢
HHHHHHHHH
AAAAAAAAA
f� xPL (ottimo di PL)v
v
¢¢¢¢¢Æ
xPLI (ottimo di PLI)
XXXXXXy bxPLc (arrotondamento di xPL)°°°µ
@@
P = {x ∏ 0 : Ax ∏ b} X = {x 2 P : x intero}
Se la soluzione ottima xPL di PL e ammissibile per PLI allora ne e
la soluzione ottima
Dim. zPL ∑ zPLI ;
xPL ammissibile per PLI ) zPL = cTxPL ∏ zPLI 2
PLI Optimum
PL Optimum
Rounding of xPL
fig. 1
The problems of the cutting plan are increasingly important in these economic-
productive sectors, because they allow an appropriate choice of the truncation
plan in each iteration of the whole Linear Programming algorithm, ensuring in
the mathematical treatment of the specific cases-study the rapidity of conver-
gence of the characteristic functions. Cutting plan problems are increasingly
848 Giuseppe Caristi, Vera Fiorani, Sabrina Lo Bosco and Alberto Vieni
important for economic applications, in particular the choice of the cutting
plane in each iteration of the related ILP algorithm is very important in terms
of convergence speediness. Several studies have been done on the matter, based
on the study of the feasible cutting planes polynomial structures. Ralph Ed-
ward Gomory [10] proved that the optimal cutting plane is the one that max-
imizes the number of feasible integer points the cut touches. A theorem intro-
duced by Georg Alexander Pick [13] allows calculating the area of each polygon
whose vertices belong to a bi-dimensional lattice, as a function of the number
of its internal and boundary lattice points. In 1957, John Edmund Reeve pro-
posed a generalization of Pick’s theorem to the three dimensional case. Many
empirical problems, such as those typical of the economic-financial sector, or
in studies of investment optimization, or for the optimal choice of alternatives
for the construction of network infrastructures (tangible and intangible), can
be represented by means of particular programming models linear integer, in
which the objective function and the constraints are integer variables. In this
case, however, the constraints of wholeness must be added to the constraints
of non-negativity, representing the problem in the m-dimensional space of the
decision variables (see fig. 2). Motivated by above facts in the present pa-
per starting from results obtained in recent years [1], [2],..., [8] we consider a
cutting plane problems for an irregular lattice (see in fig. 3), we compute the
probability that a segment of random position and constant length intersects
(cut) a side of lattice.
Project management and optimization processes 849
fig. 2 – Logical phases of the project management process to identify the optimal decision.
Optimum Solution
Definitionproblem
(Life-Cycle Cost Analysis)
Objective Function
and constraint analysis
Mathematics models
(Cutting plane in Rm)
Evaluation and analysis of
possible alternatives
Analysis and consistency of
results
2 Cutting probabilities
Considering the fundamental results obtained by Poincare [14], Stoka [15],
we want to consider a Laplace type problem for a particular irregular lattice
composed by three triangles, showing the geometric point of view also.
Let < (a;α, β), the lattice with fundamental cell C0 rappresented in figure 2
850 Giuseppe Caristi, Vera Fiorani, Sabrina Lo Bosco and Alberto Vieni
A
B H L G C
fig.3
where α and β are angles with
0 < β < α ≤ π
2. (1)
By this figure we obtain the following relations
|BE| = a sinα
sin β, |CF | = |EG| = a sinα,
|BG| = a sinαctgβ, |GC| = a (sinαctgβ − 2 cosα) ,
AE = a (sinαctgβ − cosα) ; (2)
b = 2a (sinαctgβ − cosα) ; (3)
ctgα = ctgβ − 2ctgα; (4)
|CE| = a sinα
sinα(5)
areaC0 = 2a2 sinα (sinαctgβ − cosα) . (6)
We want to compute the probability that a segment s with random position
and of constant length l < a2
intersects a side of lattice <, i.e. the probability
Pint that a segment s intersects a side of the fundamental cell C0.
The position of the segment s is determinated by its middle point and by
the angle ϕ that it formed with the line BC (o AD).
Project management and optimization processes 851
To compute the probability Pint we consider the limiting positions of seg-
ment s, for a specified value of ϕ, of segment s .
Thus we obtain the fig. 4
A
B C
DE
ϕ
C01(ϕ )
A1
A2
A3
B1
B2
B3
B4
C1
C2
C3
C5
C4
C6
D1
E1
E3
E2
E4
E5
E6
E7
a1
a2
a3
a4a
5
c1
c2 c
3
c4
c5
b1
b2
b3
b4
b5
C03(ϕ )
C04(ϕ )
fig. 4
and the relations
areaC01 (ϕ) = areaC01 −5∑i=1
areaai (ϕ) , (7)
areaC02 (ϕ) = areaC02 −5∑i=1
areabi (ϕ) , (8)
areaC03 (ϕ) = areaC03 −3∑i=1
areaci (ϕ) . (9)
By fig. 2 we have that:
areaa1 (ϕ) =l2 sinϕ sin (α− ϕ)
2 sinα,
areaa2 (ϕ) =al
2sin (α− ϕ)− l2 sinϕ sin (α− ϕ)
2 sinα,
areaa4 (ϕ) =l2 sinϕ sin (ϕ− β)
2 sinα,
areaa3 (ϕ) =al sinα
2 sin βsin (ϕ− β)− l2 sinϕ sin (ϕ− β)
2 sin β,
areaa5 (ϕ) =al
2(sinαctgβ − cosα) sinϕ− l2
4(ctgβ − ctgα) (1− cos 2ϕ) .
852 Giuseppe Caristi, Vera Fiorani, Sabrina Lo Bosco and Alberto Vieni
We obtain that:
A1 (ϕ) =5∑i=1
areaai (ϕ) =al
2(sinαctgβ − cosα)
sinϕ− l2
4(ctgβ − ctgα) (1− cos 2ϕ) , (10)
with this value the relation (7) is written
areaC01 (ϕ) = areaC01 − A1 (ϕ) . (11)
In order to compute areaC02 (ϕ) we have that:
areab1 (ϕ) =l2 sinϕ sin (ϕ+ α)
2 sinα,
areab5 (ϕ) =al
2(sinαctgβ − cosα) sinϕ− l2 sinϕ sin (ϕ− β)
2 sin β,
areab3 (ϕ) =l2 sin (α− ϕ) sin (ϕ+ α)
2 sin (α + α),
areab4 (ϕ) =al
2sin (α− ϕ)− l2 sin (α− ϕ) sin (ϕ+ α)
2 sin (α + α),
areab2 (ϕ) =al sinα
2 sinαsin (ϕ+ α)− l2 sinϕ sin (ϕ+ α)
2 sinα
− l2 sin (α− ϕ) sin (ϕ+ α)
2 sin (α + α).
We obtain that:
A2 (ϕ) =5∑i=1
areabi (ϕ) = al sinα [cosϕ+ (ctgβ − ctgα) sinϕ]−
l2
4 (ctgβ − ctgα){(ctgβ − 3ctgα) sin 2ϕ+ [1 + ctgα (ctgβ − 2ctgα)] (12)
cos 2ϕ+ 1− ctgα (ctgβ − 2ctgα)} .
Project management and optimization processes 853
With this value the formula (8) become
areaC02 (ϕ) = areaC02 − A2 (ϕ) . (13)
In order to compute areaC03 (ϕ) we have that:
areac1 (ϕ) =l2 sin (ϕ− β) sinϕ
2 sin β, (14)
areac3 (ϕ) =l2 sinϕ sin (ϕ+ α)
2 sinα,
areac2 (ϕ) = al (sinαctgβ − cosα) sinϕ− l2 sin (ϕ− β) sinϕ
2 sin β
− l2 sinϕ sin (ϕ+ α)
2 sinα,
areac5 (ϕ) =al sinα
2(ctgβ sinϕ− cosϕ)− l2
4
[ctgβ (1− cos 2ϕ)− sin 2ϕ] ,
areac4 (ϕ) =al sinα
2[(ctgβ − 2ctgα) sinϕ+ cosϕ]− l2
4
[(ctgβ − 2ctgα) (1− cos 2ϕ) + sin 2ϕ] .
We obtain that:
A3 (ϕ) =5∑i=1
areaci (ϕ) =3al
2sinα (ctgβ − ctgα) sinϕ
− l2
2(ctgβ − ctgα) (1− cos 2ϕ) . (15)
With this value the (9) is written
areaC03 (ϕ) = areaC03 − A3 (ϕ) . (16)
By relations (10), (12) and (15) we obtain
854 Giuseppe Caristi, Vera Fiorani, Sabrina Lo Bosco and Alberto Vieni
3∑i=1
Ai (ϕ) = al sinα [3 (ctgβ − ctgα) sinϕ+ cosϕ]− l2
4 (ctgβ − ctgα)
[(ctgβ − 3ctgα) sin 2ϕ+
(1− ctg2α + 7ctgαctgβ − 3ctg2β
)cos 2ϕ+
1 + 5ctg2α− 7ctgαctgβ + 3ctg2β]
. (17)
Denoting by Mi (i = 1, 2, 3) the set of segments s which have their center
in C0i and with Ni the set of segments s all contained in the cell C0i we have
[15]:
Pint = 1−∑3
i=1 µ (Ni)∑3i=1 µ (Mi)
, (18)
where µ is the Lebesgue measure in the Euclidean plane.
To compute the above measure µ (Mi) and µ (Ni) we use the Poincare
kinematic measure [14]:
dk = dx ∧ dy ∧ dϕ,
where x, y are the coordinate of middle point of s and ϕ the fixed angle.
3 Main Result
Taking in account that ϕ ∈ [β, α] , we have that:
µ (Mi) =
∫ α
β
dϕ
∫ ∫{(x,y)εC0i}
dxdy =
∫ α
β
(areaC0i) dϕ = (α− β) areaC0i,
then
3∑i=1
µ (Mi) = (α− β) areaC0. (19)
Then, considering relation (11), (13) and (16) we can write
Project management and optimization processes 855
µ (Ni) =
∫ α
β
dϕ
∫ ∫{(x,y)εC0i(ϕ)}
dxdy =
∫ α
β
(areaC0i (ϕ)
)dϕ =
∫ α
β
[areaC0i − Ai (ϕ)] dϕ =
(α− β) areaC0i −∫ α
β
[Ai (ϕ)] dϕ,
therefore
3∑i=1
µ (Ni) = (α− β) areaC0 −∫ α
β
[3∑i=1
A1 (ϕ)
]dϕ. (20)
Replacing relations (19) and (20) in the (18) we obtain
Pint =1
(α− β) areaC0
∫ α
β
[3∑i=1
Ai (ϕ)
]dϕ. (21)
By formula (17) we have
∫ α
β
[3∑i=1
Ai (ϕ)
]dϕ = al sin [sin β − sinα + 3 (cos β − cosα) (ctgβ − ctgα)]−
l2
8 (ctgβ − ctgα)
[(1 + ctg2α + 7ctgαctgβ − 3ctgβ
)(sin 2α− sin 2β)−
(ctgβ − 3ctgα) (cos 2α− cos 2β) + 2 (α− β)(1 + 5ctg2α− 7ctgαctgβ + 3ctg2β
)].
(22)
The relations (6), (21) and (22) give us
Pint =1
2a2 (α− β) sin2 α (ctgβ − ctgα)
{al sinα [sin β − sinα + 3 (cos β − cosα) (ctgβ − ctgα)]− l2
8 (ctgβ − ctgα)
[(1− ctg2α + 7ctgα− ctgβ − 3ctg2β
)− (ctgβ − 3ctgα) (cos 2α− cos 2β)
856 Giuseppe Caristi, Vera Fiorani, Sabrina Lo Bosco and Alberto Vieni
2 (α− β)(1 + 5ctg2α− 7ctgα− ctgβ − 3ctg2β
)]}.
In particular for α = π2, β = π
4, l = a
3we obtain
Pint = 0, 7.
References
[1] D. Barilla, G. Caristi, A. Puglisi, A Buffon-Laplace type problems for
an irregular lattice and with maximum probability, Applied Mathematical
Sciences, 8, no. 165 (2014), 8287 - 8293.
https://doi.org/10.12988/ams.2014.411916
[2] D. Barilla, G. Caristi, A. Puglisi, M. Stoka, Laplace Type Problems for
a Triangular Lattice and Different Body Test, Applied Mathematical Sci-
ences, 8, no. 103 (2014), 5123 - 5131.
[3] D. Barilla, G. Caristi, E. Saitta, M. Stoka, A Laplace type problem for
lattice with cell composed by two quadrilaterals and one triangle, Applied
Mathematical Sciences, 8, no. 16 (2014), 789 - 800.
https://doi.org/10.12988/ams.2014.46423
[4] G. Caristi, M. Stoka, A Buffon - Laplace type problem for an irregular
lattice and ”body test” rectangle, Applied Mathematical Sciences, 8, no.
168 (2014), 8395 - 8401. https://doi.org/10.12988/ams.2014.411927
[5] G. Caristi, A. Puglisi, M. Stoka, A Laplace type problem for a regular
lattices with octagonal cell, Far East Journal Mathematical Sciences, 48,
no. 1 (2011), 103 - 118.
[6] G. Caristi, M. Stoka, A Laplace type problem for lattice with the ax-
ial symmetry and different types of obstacles (II), Far East Journal of
Mathematical Sciences, 64, no. 2 (2012), 281 - 295.
[7] G. Caristi, M. Stoka, A Laplace type problem for regular lattices with
irregular hexagonal cell, Far East Journal Mathematical Sciences, 50, no.
1 (2011), 23 - 36.
Project management and optimization processes 857
[8] G. Caristi, A. Puglisi, E. Saitta, A Laplace type for an regular lattices
with convex-concave cell and obstacles rhombus, Applied Mathematical
Sciences, 7, no. 82 (2013), 4049 - 4065.
https://doi.org/10.12988/ams.2013.211618
[9] Dornbusch R., Fischer S., Startz R., Macroeconomia, McGraw-Hill Com-
panies, Milano, Italy, 2010. https://doi.org/10.1524/9783486808100
[10] Gomory R.E., Outline of an algorithm for integer solutions to linear pro-
grams, Bull. Amer. Math. Soc., 64 (1958), no. 5, 275 - 278.
https://doi.org/10.1090/s0002-9904-1958-10224-4
[11] Lo Bosco S., Un criterio matematico per lanalisi dellutilit globale di
un intervento pubblico sulle reti di mobilit, Collana scientifica Aracne,
Gioacchino Onorati Editore Srl, Roma, 2018.
[12] Masera R. (a cura di), Saggi sulla metodologia della ricerca in economia,
Universit degli Studi G. Marconi, Roma: Gangemi Editore, 2010.
[13] Pick, G. A., Geometrisches zur Zahlenlehre, Sitzungsberichte des
Deutschen Naturwissenschaftlich-Medicinischen Vereines fur Bohmen
”Lotos” in Prag, 19 (1899), 311 - 319.
[14] H. Poincare, Calcul des probabilites, ed. 2, Gauthier Villars, Paris, 1912.
[15] M. Stoka, Probabilites geometriques de type Buffon dans le plan euclidien,
Atti Acc. Sci. Torino, 110 (1975-1976), 53 - 59.
[16] Turner G., Handley D., Newcombe J., Ozdemiroglu E., Valuation of the
external costs and benefits to health and environment of waste manage-
ment options, final report for Defra by Enviros Consulting Limited in
association with EFTEC, 2004.
Received: August 27, 2019; Published: September 11, 2019