318
A. S. Semenov
Fig. 21.8 Component-container model of the distributed gradient ascent algorithm: 1/r = 3, N =
2, m ≈ 429 s, n = 2
by setting a capacity curve. Let m ≈ 429 s and u[m] = 7. Then a number of steps k
is calculated by Eq. 21.9.
There is no loss of generality in assuming that each container E runs A1 with next
rule for each container program in accordance with operation of prototyping ≡ (1/r,
N):
T = [0, . . . , τ n ] = ≡ (1/r, N )[0, . . . , τ n ].
(21.12)
CCM for gradient ascent algorithm and pattern 1/r = 3, N = 2 (see Fig. 21.3a) is
presented by Eq. 21.13.
E = f
n
(E
0
= {e 0 }
S
= A1,
R
{c}
S
,
[n + +, ∀{e}
S
∈ E
n−1
{e}
S
↓ (E
n
= ≡ (3, 2){e}
S
)|0 < n ≤ k] ,
[n − −, ∀{e}
S
∈ E
n−1
{e}
S
↓ (E
n
= ≡
−1
(3, 2){e}
S
)|0 < n ≥ k] ) (21.13)
When elastic component-container tree is folded, the result will be in the initial
container component. One of the potential problems is that elasticity takes time.
21.5 Conclusions
The contribution of the chapter is the application of elastic CCM to design the
distributed algorithms. As a result of this current study, several conclusions can be
drawn:
• Distributed algorithm can be presented as an “elastic object” which is transformed
dynamically at runtime.
• CCM provides the following advantages: the ability to automatically select a
distributed configuration of the data processing organization, building a visual
model of elastic computing organization, and evaluation of its effectiveness.