Decomposition Methods in algorithms



Download 54,26 Kb.
bet6/7
Sana26.01.2022
Hajmi54,26 Kb.
#410998
1   2   3   4   5   6   7
Bog'liq
decomposition

An LP example

To illustrate the idea of decomposition, we consider the LP



minimize cT u + c˜T v

subject to Au ¹ b



A˜v ¹ ˜b

Fu + F˜v ¹ h (7)

¹
with variables u and v. The two subproblems are coupled by the constraints Fu + F˜v h, which might represent limits on some shared resources. Of course we can solve the problem as one LP; decomposition allows us solve it with a master algorithm, and solving two LPs (possibly in parallel) at each iteration. Decomposition is most attractive in the case when the two subproblems are relatively large, and there are just a few coupling constraints, i.e., the dimension of h is relatively small.

    1. Primal decomposition (allocation)


We introduce the complicating variable z to decouple the constraint between u and v. The coupling constraint Fu + F˜v ¹ h becomes


Fu ¹ z,

F˜v ¹ h z.

The original problem is equivalent to the problem



minimizez φ(z) + φ˜(z)


where

φ(z) = infu {cT u | Au ¹ b, Fu ¹ z}

(8)

φ˜(z) = infv {c˜T v | A˜v ¹ b, F˜v ¹ h z}.

The values of φ(z) and φ˜(z) can be evaluated by solving two LP subproblems. Let λ(z) be an optimal dual variable for the constraint Fu ¹ z, and λ˜(z) be an optimal dual variable for the constraint F˜v ¹ h z. A subgradient of the function φ(z) + φ˜(z) is then given by



g(z) = −λ(z) + λ˜(z).

We assume here that for any z, the two subproblems are feasible. It’s not hard to extend the ideas to the case when one, or both, is infeasible.

Primal decomposition combined with the subgradient method for the master problem gives the following algorithm:
repeat

Solve the subproblems.

Solve the LPs (8) to obtain optimal u, v, and associated dual variables λ, λ˜.



Master algorithm subgradient. g := −λ + λ˜.

Master algorithm update. z := z αkg.

Here k denotes the iteration number, and αk is the subgradient step size rule, which is any nonsummable positive sequence that converges to zero. In the primal decomposition algorithm, we have feasible u and v at each step; as the algorithm proceeds, they converge to optimal. (We assume here that the two subproblems are always feasible. It’s not hard to modify the method to work when this isn’t the case.)

The algorithm has a simple interpretation. At each step the master algorithm fixes the allocation of each resource, between the two subproblems. The two subproblems are then


solved (independently). The optimal Lagrange multiplier −λi tells us how much worse the

objective of the first subproblem would be, for a small decrease in resource i; λ˜i tells us

how much better the objective of the second subproblem would be, for a small increase in





resource i. Thus, gi = λi + λ˜i tells us how much better the total obective would be if we transfer some of resource i from the first to the second subsystem. The resource allocation update zi := zi αkgi simply shifts some of resource i to the subsystem that can make better use of it.

Now we illustrate primal decomposition with a specific LP problem instance, with nu = nv = 10 variables, mu = mv = 100 private inequalities and p = 5 complicating inequalities. The problem data A, A˜, F , and F˜ were generated from a unit normal distribution, while c, c˜, b, ˜b, and h were generated from a unit uniform distribution.

We use the subgradient method with diminishing step size rule to solve master prob- lem. Figure 1 shows master problem’s objective function value φ(z) + φ˜(z) versus iteration

number k when αk = 0.1/ k. Figure 2 shows convergence of the primal residual.


−1.5

−2


φ(z(k)) + φ˜(z(k))
−2.5

−3

−3.5

−4

0 5 10 15 20 25 30

k

Figure 1: Objective function value versus iteration number k, when master problem

is solved with subgradient method using diminishing rule αk = 0.1/ k.



101


100



f (x(k)) − p?
10−1


10−2


10−3

0 5 10 15 20 25 30

k

Figure 2: Primal residual versus iteration number k, when master problem is solved

with subgradient method using diminishing rule αk = 0.1/ k.




    1. Download 54,26 Kb.

      Do'stlaringiz bilan baham:
1   2   3   4   5   6   7




Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©hozir.org 2024
ma'muriyatiga murojaat qiling

kiriting | ro'yxatdan o'tish
    Bosh sahifa
юртда тантана
Боғда битган
Бугун юртда
Эшитганлар жилманглар
Эшитмадим деманглар
битган бодомлар
Yangiariq tumani
qitish marakazi
Raqamli texnologiyalar
ilishida muhokamadan
tasdiqqa tavsiya
tavsiya etilgan
iqtisodiyot kafedrasi
steiermarkischen landesregierung
asarlaringizni yuboring
o'zingizning asarlaringizni
Iltimos faqat
faqat o'zingizning
steierm rkischen
landesregierung fachabteilung
rkischen landesregierung
hamshira loyihasi
loyihasi mavsum
faolyatining oqibatlari
asosiy adabiyotlar
fakulteti ahborot
ahborot havfsizligi
havfsizligi kafedrasi
fanidan bo’yicha
fakulteti iqtisodiyot
boshqaruv fakulteti
chiqarishda boshqaruv
ishlab chiqarishda
iqtisodiyot fakultet
multiservis tarmoqlari
fanidan asosiy
Uzbek fanidan
mavzulari potok
asosidagi multiservis
'aliyyil a'ziym
billahil 'aliyyil
illaa billahil
quvvata illaa
falah' deganida
Kompyuter savodxonligi
bo’yicha mustaqil
'alal falah'
Hayya 'alal
'alas soloh
Hayya 'alas
mavsum boyicha


yuklab olish