0 ratings0% found this document useful (0 votes) 96 views5 pagesUnrestricted Search
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content,
claim it here.
Available Formats
Download as PDF or read online on Scribd
- Na TE ONU-DIMUNSIONAL MINIMIZATION Mion
c
ar eaoonanns rey at a
we
oo | |
ae ens: Mpc = Or pat deal >
Fae 0) tet tee
vigure 5.4 Outcome Of first two experiments,
would have been discarded (Fig. 5.5b), while if”
10, 3) 4, To, a1 and [xa 1] can be dismissed (Fig. 5.50). Further,
sont (> on ¢xperiments* remains within the new interval, as will
= the Ori i" Figs. 5.5a and 5.5b, only one other experiment need.
oF Rin it in order that the process be repeated. In situations
: we {wo more experiments are to be placed in order to find a
Fig. 5. 2yal of wacertainty-
Suced int tw
oo
om
or tsp ow
pigure 5.5 Current interval of uncertainty: (a) current interval of uncertainty
S"to. xl; (b) current interval of uncertainty is [xs 1
© | current interval of uncertainty is (x2, x2]-
The assumption of unimodality is made in all the climination
techniques. Ifa function is known to be multimodal (i.c., having several
valleys or peaks), the range of the function has to be subdivided into
feveral parts and each part treated separately as a unimodal function.
ELIMINATION METHODS
3.3 UNRESTRICTED SEARCH
In most of the practical problems, the optimum solution is known to lic
within restricted ranges of the design variables.
rnge is not known, and hence the search
In some cases, this
wetrictions on the values of the variables.
has to be made with no 220
Search with fixed step size. ‘The most clementary
101 ap ee
sucha problem is) to use a fixed step size and to move troerOtey
guess point in favourable direction (positive or negativey the j,
‘ negative), ay
incthod is very simple to programme, it is not efficient in many 2° ah
method is described in the following steps.
1. Start with an initial guess point, say, 2.
2. Find ff, =S 04)
3. Assuming a step size s, find x,= x, +5.
4. Find fp=S@)-
$. Mf, 1 can be tay
en ag
7. Iff.is originally > f;, the scarch should be carried in the
direction x», xs, where x_j;= x, —(j—1)s. Teverse
8. Iff,—f, the desired minimum lies in between x, and x.
minimum point can be taken as either x, or xe. % and the
9.
Irit happens that both f, and f-, are greater than f;, it implies
the desired optimum will lie in the double interval x, < x24) 0%
4;
‘This procedure is illustrated through the following example.
Example 5.2
Find the maximum of the function
reo={ x2 for x2
—x4+3 for x>2
by using the unrestricted search method with
1=—1.0 and 520.4,
(Solution
» +The function is shown L
i - Corresponding to the initial
guess point 1, we find the function value as f, —— 0.5. Assuming. that we do 201
the location of the optimum point, let ‘ i ive direction |
fe te eee dae point, let us try to move in the negative directio
7 Xe = —1.4 and this gives the function value /
fz 28 —0.7.' Sine : gives the func
$a ein eae 42 Fr Hence by using th
ihe search fo nade inna we conclude that the maximum cannot lie to the left oy
bare © Positive direction of x by considering x; = x1 + @— Di.
vee until i i : :
following results: | “PONE ¥€ is found at which fi< fia. This leads to ™
graphically in Fig. 5.6. gam PROORANMING 13 ONT-DIMENSIONAL Mt
an
IMIZATION MUTHODS — 221
r : ® ny
Figure 5.6 Graph of s(x).
; mom t Ds fem Seo Wh
‘Sten,
xopr Hes: between
way 0nd *r
ves,
Stop,
Xppr tes between
acim) 278
‘Es
‘igure 5.7 Flowchart for unrestricted search with accelerated step size.
Example 5.3
Find the maximum of f=x
pes she of f« (1.5—x) by starting from 0.0 with an initial step size
Solution
The function value at x, is ee
an abe function value at a is F.=0.0. If we try to start moving im the negative *
tion of unimodality inaiteles, tons Se eae ee ie a ae ya
ea that the maximum cannot lie towards the left of #1:
‘Thus, we start moving is i
ing in the positive x direction and obtain the following results if
Lee
| 1 0.05 0:05 0.0725 No
2 0.10 0-10 0.140 No
3 0.20 0.20 0.260 No
4 0.40 0.40 0.440 No
% 0.80 0.80 0.560 No
6 1.60 1.60 —0.160 Yes
1 ee a
sults, the optimum point can be seen to be xopt xe—0.8. In this
wd xz do not really bracket the maximum point, but provide
If a better approximation to the optimum is desired, the
ted from xg with a smaller step size.
can be restal
