Truss Structual Optimization using NetSolve System

NetSolve

Truss Structual Optimization using NetSolve System

! #"!$!% &!' (!) * !"!$!%

+-, ./ 01 ! #"!2!% , 3!4 !5 * !"!2!%

, 67 8!9

! #"!2!%

& 7 :!; =<!>!"#2!% ? 7 @ =<!>#"!2%

,

Jack Dongarra

Department of Computer Science,University of Tennessee

%

Tomoyuki HIROYASU, Doshisha University, [email protected]

Mitsunori MIKI, Doshisha University, Tatara Miyakodani 1-3, Kyo-Tanabe, Kyoto Hisashi SHIMOSAKA, Graduate School of Engineering, Doshisha University Masaki SANO, Graduate School of Engineering, Doshisha University Yusuke TANIMURA, Graduate School of Engineering, Doshisha University Yasunari MIMURA, Graduate School of Frontier Sciences, The University of Tokyo Shinobu YOSHIMURA, Graduate School of Frontier Sciences, The University of Tokyo Jack Dongarra, Department of Computer Science,University of Tennessee

In this paper, the APIs for general optimization system in the Grid is proposed. The system is supposed to be constructed with the Grid RPC. The proposed APIs consists of three modules. One of them is the API for searching for the data from the optimizer and analyzer. The others are the APIs for the optimizer and analyzer.

In this paper, the optimization system is implemented with the proposed APIs and the NetSolve that is one of the Grid RPC systems. To discuss the effectiveness the proposed APIs and the implemented system, the structural optimization problem of truss structure is solved by the implemented system. Through the optimization, it is found that the proposed APIs are very useful to construct the optimization system in the Grid. At the same time, since there is an overhead time of NetSove, the calculation time of analyzing the problem should be longer to shorten the total execution time.

Key word: Grid Computing, Grid RPC, NetSolve System, Optimization Computation

1

ACBEDF

GIHKJMLINIOQPIRTSIUWVTXZY\[\RI]_^\`KacbTdce

Grid

fhgjilknmpolqKrhs

Grid^thuwv

bwxly

f_z{o}|l~€wrhƒ‚

^…„_†c‡…ˆc‰‹Š_Œ o JK ‚WŽ R’‘W“_^…”_•’‡W–_—’b 

r 1)^{sw˜Z™Iš Jl›IœI Ž\žIŸ} „_†T‡\ I¡£¢¤J kI¥

bQ›Iœ



ŽQžIŸc¦

“I§I¨_^Q”I•ª©¬«I­T®_¯

r\s

Grid

‡Q°_±M¢\y

²_³ §ª´…‘¤µ_¶‹·¹¸¤º‹RªY¤»ƒN ¦ N…¼ Žc½ bW¾_± k¤mcr

¿

U_©¤À‹Á

k¤m’rWs

›_œ‹

ž_Ÿ

U¤›_œ‹c‹RcS¤Ã’R_“_^

¦ Rw¨£´¬“ Ž  I¡£©

Grid^{^ i¬ÄZ U ¿\Å ®K¯ ÅTÆ s ¿ m}

¥ ^

Grid

bKÇ\È

_r\É_ÊTË

dW‰\œIÌ

t\Í

b

Global Grid Forum

b

oWÎ_Ϫk¤mco_qIr

2)^s

Ð

ÉKÊ

^

iWÑ

t J

Grid^ÒIÓIÔ

b Ë Œ

rWÕIÖI×c¦

“K§I¨

^W”_•cb‹¯

r…s_Õ_Ö_גØWÙ

^Ú¹Û

t

J_Ü_Ýcb…Þ_ߒb

ž_Ÿ

L…“ªY fWà ÛáJZâ㊅J_ä_å’^ ØWÙ

tWæ

¡ Ñ_Ž‹ØWْf Ú q

yIçQJ

Grid^ÒIÓIÔ

b

ËIq_oIè

^\é_êZ‡\ë

ÆI¿

U\®WìKí

Ñ

Žw¦ “K§w¨K^}”K• f «K­p®K¯ r ¿ U f ÀKÁ kîmZr}sðï

Grid^ñ

U

t\ò

‚

Power Grid(^óãôKõ )

b è

^Qö_†

f ¯

rWsZ_ŽQ÷

ø

JcùcNcú{û_RWü

tWó_ý‹þcÿ

‡ƒL_NW»’NcY{b

_r

Œ_®WJ

ó_ô

^’‡’Œ r ¿ U f «‹­c®‹¯ rWs û_RWü

t

k¤m

y

ó_ô

f

ä_å’b

½ ¿ ®

þ k¤m

y_^_â…J

½

^cd

Æ

b k¤m y_^_â…Jcx_yWJ èWmª¥ ^

ÿ

f’½ ^c®

¯ r

^Iâ_‡

r

t

Ž_q\s

Grid^ÒKÓIÔ

b Ë Œ

rWÕKÖI×

¦

“_§_¨‹b

ˋq_o

©{JcùcNcú{û_RWü

t

ê Ñ b ½ ¿ ® Õ

Ö_× · f ë ÷_mco‹q_r ^‹â_‡ r ¿ U f_Ž ۅU_© Õ

Ö_×

^…ˆ

fª¥¤m’r

¿ U f

®_¯

rWs x_y…J

Õ_Ö_×

^’^  ·

t

Jcº _P_R! תk{m" # b$

%

«_­c®K¯

r ¿ U f

U

Ž_rWs

Ð

ÉKÊ

® t ¦

“I§_¨_^W”_•T‡

Grid RPC

U&

š\mcr('

ú)!+*…ù,_^…°_±c‡- ™_r…s. bJ_ä_åc^…ä/cb Ë_q_o

t J

Netsolve³⁾

U&

š\mZr

Grid

^ '

ú0!1*324,I‡Q±

qwo

¦

“_§‹¨_‡W”‹•

_r…s56c‹rWÕãÖ_×c¦

“_§_¨‹®

t

J’x

7 J Õ_Ö_× b Ë Œ r89: ; ‡<ãÌ _r

Optimizer

U\¯

r=_ž> ? b‹¾

_r’iWÑ

Ç

?@

·ABC D

@

‡<_Ì



r

Analyzer (Function Call)^{EGF b FGH ¢ o - ™wrQs} Op- timizer

U

Analyzer

‡… _¡ ¢Wy

¦

“ã§_¨

t . ̒^

؅Ù

b

ª¢

o t ì_í

Ñ_ŽI

¡ª©W¯

r_f

J

Õ_Ö_×

_·

ØWÙ

^$

% J

b ؅Ùcf ¯ªeðJK Lc^C Dc‡NMcy k…Ž_q…sO8 b

56’¦

“㧋¨

t

JOP

‚ ^ 

RQƒb ±R

k mƒo‹qEr

Opti- mizer^{‡S‹ê¢ J ¦} “‹§E¨…°‹± Tâ ¥UWV ‡ƒŒãŠ‹Œ r

Optimizer Agent

U¤JP

‚ ^

Optimizer

â

¥X V

‡’Œ

ŠKŒhJ4P

‚ ^

Analizer

b

žIŸ

‡YOZ

Ir

Analizer Agent

^

2

ŠI^

Agent

â ¥}ŽIr\s…ÕIÖw×T¦ “w§I¨O[Q`\TTy ø t J

P\]Tb

Optimizer

U

Analizer

‡O[\`

Ir\s€¦

“I§K¨\°_±

T t J è\mTf

Grid

KT^ ½ ¿ b |w~ ¢¬J ½ Æ

°I±T®_^ r ^

â_‡`ª¢’Û)

¥

7cU‹©¤J

"#

b ¦

“ã§_¨_‡…°_±

‹r

¿ U

Ð É_Ê ® t J 56’_rWՋÖ_׃¦ “_§‹¨_‡Wä ‹r y_ç

b

’Ž

Grid RPC

^‹y_ç‹^

API

U

Netsolve

‡…±

q y

ä/’b_Š

q_ocrWs

q_o

J

56c¦

“‹§_¨‹^WÜ

E

‡

Ñ

,! ’¸_¨…J

Ë

dW‰

’‡W±

q

yªY

“\”

OTbcdTe äO/£¢

o?O@

äT‡\ë

z

y\ˆOT‡£¢îJ

-c‡\ë

Æ s

2 Grid

2.1 Grid

Grid

U

tQu_v

VTXTY}[\R_]IbO k¤mToIqIr ¯ ¥Tr

ž_Ÿ

„_†c‰ _„_†’‡W ã¡ª¢

o

Jcû_RWü

f

!’bW°_±

® ^ r ¿

U‡ i Ä ¢ J ¿ m ‡…ä" ãr#%$&"'

(^{(*) Y}

* 2\,îJ,+TRMú*2\,îJ€VZXTY\[QRI]

)

U¬¾I±-/.I¼0I¨

‡R1

_r 4) ^s

Grid

® t

M‚

^ 

R’‘ “ f

uEv bN

k{m

y2%3’® J

45câ_Š…ì_í Ñ b ž_Ÿ „_†ƒ‡W°_± _r b t

Jcû_R…ü67

·T“,8O lPwRª¸}N\¼ J:90;IJã»0<lPM¸¬§>= J

)@?

R! YKY

´AcNW“…J ½ ¿ b èB㎠fW|‹~ ¢ o_q_r âcU qcz y



R’‘W“

Ž’½

J_ÚC’b

÷ y r;

‡-D ¢ Ž Œ

mcš_Ž

¥WŽ_q\s

¿

^cd

Æ

Ž_؅Ù

‡N-Dª¢

Ž_f’¥

J ,- ¸8‹R

¦FE N…^

-._¼ ' NW¼‹‡Wë

Ƌ¿

U t

Þ_ߒbGH’®_¯ r…scè ^_y

çWJ

Grid

b_Ç

IrWÉIÊ

tJI

ÀT^0KcdTeðJ\KLT^

Ø\Ù

‡

< ‹r

Grid

^ ' ú ! *2O,ã^W[` f? ÚÛ Žƒk mƒoãq

r\s*L/M_˜

UI¢

o J

Globus⁵⁾

·

Legion⁶⁾

J

NetSolve³⁾

J

Ninf^{7) U qEz y ¦ “I§w¨ fONQPZ¥ mTrQs ¿ mM¥} Grid^{^ '}

ú0! * 2G,I‡\°w± Ir

¿

U\®QJ "O# b

Grid

KZ^\„I†Z‡Q°

±

_rI¦

“_§_¨_‡W”_•

_r

¿ U f

®(^

rWs

2.2 Grid RPC

~

Grid

^ ' ú ! * 24,

t k xQRQx

Ž

©\^

f

[Q`

k m

o‹Ë eðJ

q Û Š_â‹^ ,- ¸A8_R

¦E

NST-.‹¼

'

N…¼

º‹Â!’‡

5

_r…s

Grid RPC ^{t è}

^

LM’Ñ‹Ž

©…^_^…Ü

Š_®_¯

rWs

RPC(RPC:Remote Procedure Call)

U t

¸îºwRMY

¦ “

§_¨Kcb |_~’_rWVU !cR%WcN…‡_VcXcYW[WR‹]XVY_®&

‰Z

:[

­’®ã¯ªeðJ

Grid

KĨ:\

™‹r

RPC^t Grid RPC

U&

šQmToIqKr\s

RPC

b

#Q]

Û

žKŸT¦

“I§K¨

t

ûIR\üM‘

¸¬§,=KbJ^

m J

uIv

žIŸ

ÒKÓ

®I¯

r

Gird

KT^

'

ú0!1*2

,‹^…Ü_%` U‹¢

o

%a’®‹¯

r…s

Grid RPC^{t }

®‹b…

‚

Ž

Fb

®W䋱 Ñ b:\_± k¤m Š_Š‹¯ªeáJcd ×e ^:f eTg

h 8) ^©îë ÷ImToIq_rQs

¿

^Iy_ç\J

Grid RPC^tJi/j

x 

x  `/k _r U - ™c¥¤mcrQs

2.3 NetSolve System NetSolve System^3)t

J

Jack Dongarra^l

bZd

z o [}`

k¤m y

Grid RPC ^{b #] Û ¦ “K§_¨_®_¯ rWs} NetSolve^t

8 ^

3

ŠI^

Í L_NnmIR_VTNTY¤â ¥ ”0o k

mco_qIrWs

1. The NetSolve Client 2. The NetSolve Agent

3. The NetSolve computational resorces (or servers)

p

1

^cd

Æ

bWJ

NetSolve^¦

“_§_¨_‡\_±¢

Grid

Kc^

q U ª¸ ·+cRªú*2,_‡\‹±ª¢Wy q û‹RWü_· ,-ª¸

8KR

¦rE N t J

NetSolve Client

U

ŽIrQs

NetSlove Client

t NetSolve Client Interface(^{¿ m t} API^{UI¢ o 5 k¬m}

r

)

‡J\‹±ª¢¤J

NetSolve Agent

bª¢

o

Jq

U ¸¤^

\_±c‡

\Vc_rWs

NetSolve Client

â ¥ ^ V

‡O£ŒcŠcŒTy

NetSolve Agent^t

Jã¯

¥

âQs\çnt/u

k¬mToKqIrJv

NetSolve Server

^._Rªú Q’NW“W·\_±_«‹­

Ž

%q

U ¸¤‡-Dª¢{J

Ö

w Ž

Server

e U¤J

NetSolve Client

^ V ‡xO _rWs

NetSolve Agent^{b€d z o Jzy,{ kîm y} NetSolve Server

t J

NetSolve Client

â ¥ ^

OV

‡ /|cŒ J}q

U

M¸¬‡

ä_몢¤J

NetSolve Client^{^ OV} ‡Wé_ꪢ¤J_ˆc‡J~  s

NetSolve Agent

NetSolve Server

NetSolve Server NetSolve Server

NetSolve Client

Data Grid RPC Choice

Result Registration

p

1: Use of The NetSolve System

3 Grid RPC

€‚„ƒ†…„‡„ˆŠ‰„‹‚ŒŠƒ D„ŒV

Ž’‘”“Š•

Ð

ÉKÊ

® t Grid RPC

‡W°_±ª¢WyWåZ^

Õ_Ö_×c¦

“_§_¨

^ )

´{R_¨c[…R_]‹‡ 56c_r…s x‹y…JIä‹åcb…J

NetSolve

‡W°_±ª¢

o

ä/ª¢¤J

"#cŽJ–—

äc‡Wë

Æ s

3.1 ^{˜™š›œž}

Ð É_Ê ® t J ÕI֋×_ž‹Ÿ b Ë Œ r89: ; ‡<‹Ì _r

Optimizer

U G€‡hë

q€i}Ñ Ç ?

·4AGB4CGD€^

@

‡3<KÌ



r

Analyzer^{‡ FOH ¢ o - ™KrWs0Ÿ b ycU Ë eðJ} Op- timizer

U

Analyzer

‡… ‹¡ª¢…y

¦

“ã§_¨

t J . Ìc^

ØWÙ

b £¢

o t

ÞIßZbJ^

m y ¦

“_§K¨TU

Ž_r\s

¢\âT¢

Ž_fZ¥

J

äKåTb

t J

Optimizer

U

Analyzer

^O[\`TKJ ÉIÊ T

t 

ŽKr\spè

^Iy_ç\J

Optimizer

U

Analyzer

U

FH

Kr

¿ U

®}J èQm¢¡Qm ^G[}`Z‡n£>¤Zb}ë

Æl¿

U f ® ^îJ à J ­ Žw¦

“_§_¨_‡W”_•

‹r

¿ U f

«_­ªU

Ž_rWs x_y

Optimizer^t

Analyzer Wrapper ^¿ Analyzer

^

;

‡

‹r

¿ U f ® ^ rWs

¿ ^ ¿ U t Jc¯

r

ÜpŠK^

Analyzer^{b Y ¥QŽKq ± ÑKŽ} Optimizer^{^4[Q`Z‡}

–_—cb

_rWs

3.2 Grid RPC J˜/™0šQ›/œ//ž0

Ð É_Ê ® t J

Grid RPC

‡…± q y Õ_Öãג¦ “‹§‹¨_‡ 5

6cIrWs56cKr

Grid RPC

‡W±

q y

Õ_Ö_×c¦

“_§_¨_‡

p

2^{bJ  s p} 2^{® t JIÜ ˜ U_¢ o J} Optimizer^{U_¢ o} Gradient Method

J

GA

J

SA

^ v

Server

f ¯eáJ

Analyzer

U‹¢

o

FEM

J

CFD

J

MDA

^ v

Server^f

¯ r ¿ U…‡

̪¢ oIq_rWs

Agent for Optimizer

Gradient Method Server

GA Server

SA Server

Agent for Analyzer FEM Server

CFD Server

MDA Server

Client Grid RPC

Grid RPC

p

2: Optimization System using Grid RPC

¿ ^ ¦

“_§_¨_^

.

t 8

^9¹e¤®_¯

rWs

• Optimizer

±c^

Agent(Agent for Optimizer)

U

An- alyzer^{±c^} Agent(Agent for Analyzer)fW|_~c_rWs

•

v

Optimizer^t Agent for Optimizer^{J v} Analyzer

t Agent for Analyzer

btu

_rWs

•

v

Optimizer

^

Server^t Agent for Analyzer

^

~

‡

zQo_Ë

eáJG

Ir_Ø\Ù bcd

z\o

Agent

‡9

s o

Grid RPC

bcd r äKë OV bcdTe ÖcŽ K±

^

Server

‡y{c®(^

r\s

Ð ¦

“l§K¨l®

t

Jp®

0 y€d

Æ b

ÕlÖw×wžwŸ

‡}ë

Æ

Optimizer

U OT‡}ë

Æ

Analyzer

‡

FOH

¢ o - ™ J

Grid

Ò_Ó_Ô

®‹^

¦

“‹§_¨_‡…”_•

_r

¿

UW®WJ

8

^cd

Æ Ž ° ;

‡ cŠ s

•

ù€NpúnûKRhü

t Agent for Optimizer

f"!#pwr} R

Q€^

h ‡

z{owqhm€š

d€ÛðJ

½ ¿ b

Optimizer

·

An- alyzer^J Agent for Analyzer^{fZ½ ¿ bw¯ r ^Kâw‡ r}

t

Ž_qWs

•

ûKR}ü

t Optimizer

U

Analyzer

^}äKë

)$

q !&% '

(*)+,-*./01

• Optimizer² Analyzer³ Grid4657869;:=<

+>@?

AB*C !#

)+DE

2GFHI*3JK

/LM*N

%'

(

5

/PO

7*QR2ST*U

>?*.VW CX +1

Y

1: Arguments of API for Optimizer Name of Optimizer Specification of Optimizer

Tag Specification of Tag Name of Analyzer Specification of Analyzer

Inputfile Path to Input File Outputfile Path to output File

3.3 Grid RPCZ[\]^_*`abcdefd API Grid RPC Uhgjilknmno

Mqpsrst %unv )n+sDqE

Uw2

xnynz C 3

0{O}|s~q

API

%' ( 5

Dw1€q

Uq‚<„ƒ

U

|0

7…†

)+1‡Mpr6tˆ‰Š

3‹



API^UŒP

D

Optimizer^Ž*g Analyzer

%uv

5

/

<*‘

/ ƒ /01

3.3.1 ^’“”*d•–

86—

Mp@rt

C

32

€‡

Calling Sequence^{% '}

(

5˜2

™q' VqWš/

Optimizer

X +n0

3 Analyzer

w›qœ %w +

‹ž4

. CGŸ + ¡G¢ %

89 )q+1

‹R<G3 NetSolve

C  2

£

H¤

¥¦

%§P¨

Calling Ssequence^{%© U*57}

0+1

[opt analyzers]

ª«¬

' VW*/

Analyzer

­® %Y¯ )+

[opt analyzers show Analyzer Name]

°@±

5 D

Analyzer(Analyzer Name)



…&†

%"Y6¯ )

+

[opt analyzers subanalyze Analyzer Name]

°±

5 D

Analyzer

.

2² ³

C ¬ ' )@+

£‡´

Analyzer

%Y¯

)+

[opt optimizers]

ª«¬

' VW*/

Optimizer

­6® %Y¯ )+

[opt optimizers show Optimizer Name]

°±

5 D

Optimizer(Optimizer Name)



…‡†

% Y¯

)+

3.3.2 Optimizer^d API Optimizer



API³

Mprt ™' Š*µ6¶·¸*N¹R~

ƒ

¢6º

»*:¼<

+1

Optimizer



API^3½*U

¯ ) Q 

C 2 Y

1^U

¯ )

5

|¾¿ % 4

+1

optimize(Name of Optimizer,Tag, Name of Analyzer,Inputfile,Outpufile);

Optimizer

À § )+žÁPÂ

´  °ž±*% )ž+

Tag^3½



3

| °± . C*Ÿ +1

[Analyze Init]

°@±

:¼<

D

Analyzer

@Ã6Ä&ÅlÆǹ˜È

K Ä $

·"Æ %

Analyzer^UÉ

+1

Y

2: Arguments of API for Analyzer Name of Analyzer Specification of Analyzer

Tag Specification of Tag Inputfile Path to Input File Outputfile Path to output File

[Run]

°±

: < D

Optimizer

È

K Ä $

·"Æn~

ƒ Optimizer

% À §

5R2*»K Ä $

·Æ

U

*% )1

[Analyze Config]

°s±

:"<

D

Analyzer ^U Input Config

Ä $

·Æ

4

Output Config

Ä $

·Æ % É +1

Input Config

Ä $

·Æ

43 analyzer

È

K Ä $

·Æ

4

> ¿

4



žY 

‹ž4

C

2P‹ <

%©

U Analyzer



wrapper³ Analyzer

È

K Ä $

·Æ % *)+1

Output Con- fig

Ä $

·Æ

43 analyzer



» K Ä $

· Æ

4

¿

Ž*g

 Y 

‹4

C 2 ‹R<

%©

U An- alyzer



»K Ä $

·Æ*~

ƒ wrapper³

¿

Ž

g

%

»

)+1

3.3.3 Analyzer^d API Analyzer



API^3*U Optimizer

~ ƒ º

»*:R<

+1

|

i 2 Optimizer ^{3 !}

C¯

5 D

Optimizer

"WG

0jO=|~ %

Analyzer^ API^{%"' 0 7} À#s)@+"1

Analyzer



API^{3 ½&U}

¯ ) Q  C 2 Y

2^U

¯ )

4

|@@¾ ¿ % 4 +"1

analyze(Name of Analyzer,Tag, Inputfile,Outputfile);

Optimizer

À § )+žÁPÂ

´  °±G% )+

Tag^3½



4

| °± . C*Ÿ +1

[Initialize]

°±

:R<

D

Analyzer

 ± Ä $

·Æ %

Analyzer^U

É +

[Config]

°s±

:"<

D

Analyzer ^U Input Config

Ä $

·Æ

4

Output Config

Ä $

·Æ % É +

[Solve]

°ž±

:R<

D

Analyzer

È

K Ä $

·ÆG~

ƒ Analyzer

% À §

5R2*»K Ä $

·Æ

U

*% )

[Result]

°6±

:˜<

D

Analyzer

%$

7

%& ' &%

»K Ä $ ·

Æ U )

3.4 ^()*`abcd*+

FžHI*3½

 g ¨ / ,- C

8—

Mpžrt C Á  ´ % À

§ )+1

1. opt analyzers ^5R2 Analyzer 2. opt optimizers

/1. % ¬ ' 5 2 Optimizer

%2/30 )+

3. opt analyzers subanalyze Analyzer Name

/. % ¬

' 5 2 /0

5 D

Analyzer



£G´

Analyzer

%4526

µ)+1

4.

$ 7 

Analyzer

 ± Ä $

·Æ %' ()+1

5.

$ 7 

Analyzer



Input Config

Ä $

·Æ

4 Output Config

Ä $

·Æ %' (*)+1

6.

$ 7 

Analyzer^U

57 2

± Ä $

·Æ %

É7 )@+

optimize(Name of Optimizer,Analyze Init, Name of Analyzer,Inputfile,Dummy);

7. Analyzer^U

5"72 Input Config

Ä $

·Æ

4 Output Config

Ä $

·Æ %

É7 )+

optimize(Name of Optimizer,Analyze Config, Name of Analyzer,

Input config file,Output Config file);

8.

/ 0

5 D

Optimizer

È

K Ä $

· Æ %' (

5¼2 Opti- mizer

% À § )+1

optimize(Name of Optimizer,Run, Name of Analyzer,Inputfile,Outputfile);

‹

8

3 Outputfile^{U :¼<}

+1

3.5 NetSolve 9:<;e]^_*`abcd*=

>?



Optimizer^Ž*g Analyzer



API

%©

U Net- Solve^{%' 0 7 À# %§  D1À # 3@ E 7A B C 2} Net- Solve

C' (

:R<G7 0ž+

¿ %' 0 7 API^{UŒ }

Dž¶·

´

¶DC %uv )

<*‘*g 01

EF

‘2 Optimize()

% À#)+G H

U3

€&

g ¨ U

/+1

netsl(Name of Optimizer,Tag,

Name of Analyzer,Inputfile,Outpufile);

Analyzer()

% À#*)+GH

U3

€*

g ¨ U /+1

netsl(Name of Analyzer,Tag, Inputfile,Outputfile);

‹R<

~

ƒžQJI

~+

g ¨ U2

À#

U3 K

L

7

¿

%™

' )+

‹64Q

VW CX +1

M

2



8—

M6p6rt

UŽ

0

7 2 Optimizer^U GA² An- alyzer^U CFD

%/0

5

DGH

2N Client² Sever^U6Ž

ž+

Optimizer^OG Analyzer² NetSolve



API

.žº



»*:R<

+,-

U

|0

7 M

3^U

¯ )}1

optimize(GA,Run,CFD,Inputfile1,Outputfile1) Client

Optimizer

analyze(CFD,Solve,Inputfile2,Outputfile2)

Analyzer

netsl(CFD,Solve,inputfile2,outputfile2) Grid RPC

netsl(GA,Run,CFD,inputfile1,outputfile1) Grid RPC

CFD Server

M

3: Flow Chart of API Call

4

4.1 qc"!#%$'&qa(*)*+, 9: ;že

()*`abcd*=

NetSolve ^%w'

0nDÀ # 5 D

8n—sksmso

Mspsrsts.-

W %/0 )ž+žDžE

U2 Optimizer^U213

¸Æ4 C65qt

(GA:Genetic Algorithm)⁹⁾² Analyzer^U

¹˜¶@p u87 &'

%' 0 7

À#*)ž+1:9;2<

3

¹R¶p

u:7=



kžm

>

;2< CX +1

GA^3?>

= A@

o

%AB8C

5

DAD

< D k@m@o

,FE&"­&| C

X +1

GA^3?G

¿&IHKJ

(^L:M )^Ug

+ONIPKQ¦ %§ ¨ 1

¹R¶p u7:= 

km

>R;"<

432

X + ! P

UK

%2S

F

72:G

¿G

%2T F D 4 Ÿ

2*‚<Pƒ

%2U D)

k

V

M:W

 ¹R¶p

u:7:=G%

>*)+

‹4 CžX

+1

U32N !

P&:X

2N6³KY



KKZI[:\

.O] F

ƒ¼<

+1

xyz C' 0+ ¹R¶žp u:7 &'

32N³:Y

:^_

W %

>@¿

4

)@+1

Optimizer

CX +

GA^{3 2NFLIM}

 >

¿

%

¹˜¶6p

uI7I=



³8Y

I^O_

W 4657 Analyzer^U

&

' %O`:a )+1

Analyzer³²

¹R¶p uI7 & ' %§ 0 2 ¹

¶p u:7= 

M:W2

$ 7  !

P*X

2 $ 7 

³Y

2b

KP4c[:\

K

0:d

< ~ % )"1

GA³

­:e

žU

%*f

UgI

/ž0qDžE

2

xžyžz

C

32

¹R¶žp u:7=

km

>R;"< % & O > C

%

g ¨ DE

U2:hRi Æ*r:j

¿:ERk

m ' 5 D1

hi Æ*r:j

¿:E

3 U D :}<

/ž0

k2l:m

oP5

D2nRk

¿ 4

¡*¢H

I:o D

¿k

kmžo )+

4 0 ¨

p E CX +1

4572

$ 7  !

P*:Xq* >

r 2 $ 7 

³Y

"b K  > r k  2

žD$

7 

³Y

U[\

.ž/q0

‹ž4ž4ž5 D1

‹

žDE

2 GA

.„¹ ¶žp&'

UGg7

¿k  +

U32

¹ ¶p u:7= 

MWP4

!

P*:X:q

2³:Y

2b

K2:[:\

K .,-

U

/+1

x ! C 2 K9

45

DP¹R¶p u:7K=

32

M

4^U

¯ ) g ¨

/ 2 22^{! P} 50^{³8Y  ¹˜¶p uI7I=‡C6X +1 ‹  ¹¼¶6p u}

7:=

32 5^³:Y

 © x uI7  ¹R¶p:k

10^s:t:u

D u:7

4

/

7

0ž+1žD

‹  ¹R¶p u:7=

3k

:vG:w|ž

!

PRkKx ±

5R2@k

> v­&|

! P U2yKz

S t 1000.0N

k

{ S )+1

¬ '

Y|G3}*~

DžE€*-ž¿

1.0GPa



‚:ƒ €:-

YP4 )+1

š32

:„G)+

!

P*:X*.

0.003m

€

2 $

³IY

 K .

2.0MPa

€

2 $

³KYU6Ž

M

4: 10 Stage truss

Y

3: Machine Spec.

CPU PentiumIII 800MHz

Memory 256MB

Network 100Mbps(Fast Ethernet)

0

72[:\

.

Ž*‹ƒ

/0

‹4

†…R| CX +1

g72 22^!

P

50^³Y 144

P¹ ¶qp u7=  k

m

>R;?<

CX +1

4.2 ^*:‡:ˆ:‰

À‹Š

U ' 0‡DsL‡M N@p

h 6 µ‹k Y

3 ^U

¯ )"1$

7 

Client² Agent² Server^U

Y

3

RŒ

H†

k2Ž

i7 D1

4.3 ‘:’R“d:”2•R:–K&:—2˜*Z[\2™:š:›Kœ

Grid^Iž*UŽ

0

732

>? AB .OŸ: 

UO¡

«

)6+

‹4

.?] F

ƒ¼<@2¢

­&'£k § ¨

Analyzer ^QcG

¿K¤

¡ « )

+

‹4 .2] F

ƒR<

+1

‚

GH

Uq32*‚< ƒ

k2¥¦:§:¨

U7

¬ ' )+

‹4U*g*i 2k6mo



STo .2©<

<

+1

M

5^U

¯ )

kžmžo

MžpžrtžG H 2 Optimizer



GA Server^32G

¿*

L:M

&'Rk2ª:«

U2g

F

+žDE

2

¥:¦

U

§:¨&)+

‹4

. CŸ +1

M

5

C 3

¹R¶p&'Rk §P¨

¢R¬

­:® 

Analyzer

.

2

| X i 2 2^L:M

 ¹¼¶p&'¯k

¢ 8

U

§:¨ C&Ÿ +1

M

5^U

¯ ) 2 NetSolve

k ' 0

2

| ¹=¶p& '£k?¥K¦

U À § )n+

kžmqo

Mžpžrqt

4 2 NetSolve^{k ' 0d 2} GA

U*g72°½GU

¹R¶žp&'Rk § 0P¹R¶qp u7:=

kžm

>

;2<Rk& O

kžmo

MžprtžÀ § 8:±>:²Rk § 0

2:³

´ k § ¨ 1Kµ¶ž*9;2<

32·

B . V : 0žDžE

U2

&'

U

,6-‡/8I±

3?¸8¹&U V :

01

‚

6DE

2

&' 88±

U&gw

72

. .

g ¨ U o

)ž+~*k¥:ºG)+DžE

2

­R¶

 ¹R¶žp&'

U 0ms² 1500ms² 3000ms

8:±k

‚<R»