next up previous contents
Next: Lazy Learning, Artificial Neural Networks... Up: The METHOD project Previous: Preliminary numerical results   Contents

Subsections


Interface between CONDOR and XFLOS / Pre-Solve phase

We want to optimize optimizes a non-linear function

$\displaystyle y=$$\displaystyle \mbox{$\cal F$}$$\displaystyle (x), \; x \in \Re^n \; y \in \Re \ $   Subject to: $\displaystyle \begin{cases}
 b_l \leq x \leq b_u, \; & b_l,b_u \in \Re^n
 \\  A...
...Re^{m \times n}, b \in \Re^m \\
 c_i(x) \geq 0, \; & i=1,\ldots,l \end{cases}$ (11.1)

For the Method project, the objective function $ \mbox{$\cal F$}$$ (x)$ is an external code (an executable program): XFLOS. I developed a simple interface between CONDOR and XFLOS. This interface is configured via a configuration text file: ``optim.cfg''. I will now describe the content of ``optim.cfg''.
Here is a simple example of a ``server configuration file'':
; number of CPU's (not used currently)
;1

; IP's (not used currently)
;127.0.0.1

; blackbox objective function
;/home/andromeda_2/fld_user/METHOD/TD21/splitter
/home/fvandenb/L6/splitter
;testoptim

; objective function: input file:
/home/fvandenb/L6/000/optim.out

; objective function: output file:
/home/fvandenb/L6/000/xflos.out

; number of input variables (x vector) for the objective function
31

; The objective function is ofunction= sum_over_i ( w_i * ( O_i - C_i)^(e_i) )
; of many variables. The weights (w_i) are:
; 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19     20
  0 0 0 0 0 0 0 0 0  0  0  0  0  0  0  0  20  0 -1     20

; C_i are:
; 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17       18 19     20
  0 0 0 0 0 0 0 0 0  0  0  0  0  0  0  0  0.0956  0  0  0.521

; e_i are:
; 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19     20
  2 2 2 2 2 2 2 2 2  2  2  2  2  2  2  2  2  2  2      2

; optimization of only a part of the variables:
;dz zh1 rh2 zs1 rs2 sleh sles sh1 sh2 th1 th3 ss1 ss2 ts0 ts1 ts3 tkh tks b0 b2 r2  r0 delt bet2 str coel dle el11 el12 el21 el22
; 1  2   3   4   5   6    7   8   9   10  11  12  13  14  15  16  17  18  19 20 21  22   23  24   25  26   27   28   29   30   31
  0  1   1   1   1   0    0   1   1   1   1   1   1   1   1   1   1   1   0  0   0   0    1   1   1    1   1    1    1    1    1

; a priori estimated x (starting point)-
;     1     2     3     4     5     6   7   8      9       10  11  12   13     14       15    16    17      18     19    20
;   21   22       23       24    25   26    27    28    29    30   31
;    dz   zh1   rh2   zs1   rs2   sleh  sles   sh1 sh2    th1      th3 ss1 ss2  ts0    ts1      ts3   tkh   tks      b0     b2    r2
;   r0  delt     bet2      str coel  dle   el11   el12   el21   el22
0.1068 0.0010 0.055555 0.0801 0.2245 0.0096 0.000 0.3 0.37 -0.974 -1.117010721 0.297361 0.693842 -0.301069296  -0.8 -1.117010721  0.004 0.0051 0.07647 0.035 0.225 0.07223 0.0 -0.99398 0.00 4.998956  0.000  0.000  0.000  0.000  0.0000

; use previous line as starting point:
; - 1: yes
; - 0: no, use best point found in database.
1

; lower bounds for x
; 1   2   3   4   5   6   7   8   9   10   11  12   13   14   15   16   17   18  19  20  21   22    23   24   25   26   27   28   29   30   31
;dz zh1 rh2 zs1 rs2 sleh sles sh1 sh2  th1  th3 ss1 ss2  ts0  ts1  ts3  tkh   tks  b0  b2  r2  r0 delt  bet2  str coel  dle el11 el12 el21 el22
  0   0   0   0   0   0   0   0   0 -1.7 -1.7   0  0  -1.7 -1.7 -1.7 .002  .002   0   0   0 .05    0  -1.0  -.5  0.1 -0.05    0    0    0    0

; upper bounds for x
; 1   2   3   4   5   6   7   8   9   10   11  12   13   14   15   16   17   18  19  20  21   22    23   24   25   26   27   28   29   30   31
;dz zh1 rh2 zs1 rs2 sleh sles sh1 sh2  th1  th3 ss1 ss2  ts0  ts1  ts3  tkh   tks  b0  b2  r2  r0 delt  bet2  str coel  dle el11 el12 el21 el22
  1   1   2   1   2   1   1   1   1  1.7  1.7   1  1   1.7  1.7  1.7  .05   .05   2   1   2   1    2  -0.4   .5  10   0.05  0.02  0.02  0.02  0.02

; number of inequalities
15
;0

; here would be the matrix for inequalities definition if they were needed
; 1   2   3   4   5   6   7    8   9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24   25    26   27   28   29   30   31
;dz zh1 rh2 zs1 rs2 sleh sles sh1 sh2 th1 th3 ss1 ss2 ts0 ts1 ts3 tkh tks  b0  b2  r2  r0 delt bet2 str coel  dle el11 el12 el21 el22  RHS
 -1   0   0   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   0   1   0   0   0   0   1    0     0    0    0    0    0    0
  0   0   1   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   0   0  -1   0   1   0   0    0     0    0    0    0    0    0
  0   0   0   0  -1   0   0    0   0   0   0   0   0   0   0   0   0   0   1   0   0   1   0   0   0    0     0    0    0    0    0    0
  0   0   0   0   1   0   0    0   0   0   0   0   0   0   0   0   0   0   0   0  -1   0   1   0   0    0     0    0    0    0    0    0
  0   0   0   0   0   1   0    0   0   0   0   0   0   0   0   0   0   0   0   0  -1   0   0   0   0    0     0    0    0    0    0    0
  0   0   0   0   0   0   1    0   0   0   0   0   0   0   0   0   0   0   0   0  -1   0   0   0   0    0     0    0    0    0    0    0
  0   0   0   0   0   0   0    1  -1   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0    0     0    0    0    0    0    0
  0   0   0   0   0   0   0    0   0   0   0   1  -1   0   0   0   0   0   0   0   0   0   0   0   0    0     0    0    0    0    0    0
  0   0   0   0   0   0   0    0   0   0   1   0   0   0   0  -1   0   0   0   0   0   0   0   0   0    0     0    0    0    0    0  .35
  0   0   0   0   0   0   0    0   0   0  -1   0   0   0   0   1   0   0   0   0   0   0   0   0   0    0     0    0    0    0    0  .35
  0   0   0   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   1   0  -1   1   1   0   0    0     0    0    0    0    0    0
  0   0   0   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   0  -1   0   0   0   0   0    0     0    1    0    0    0    0
  0   0   0   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   0  -1   0   0   0   0   0    0     0    0    1    0    0    0
  0   0   0   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   0  -1   0   0   0   0   0    0     0    0    0    1    0    0
  0   0   0   0   0   0   0    0   0   0   0   0   0   0   0   0   0   0   0  -1   0   0   0   0   0    0     0    0    0    0    1    0

; scaling factor for the normalization of the variables.
; 1     2    3    4    5    6    7    8    9   10   11   12   13   14   15   16   17   18   19   20   21   22   23   24   25   26   27   28   29   30   31
; dz  zh1  rh2  zs1  rs2 sleh sles  sh1  sh2  th1  th3  ss1  ss2  ts0  ts1  ts3  tkh  tks   b0   b2   r2   r0 delt bet2  str coel  dle el11 el12 el21 el22
1e-3 1e-3 1e-3 1e-3 1e-3 1e-3 1e-3 1e-2 1e-2 1e-2 1e-2 1e-2 1e-2 1e-2 1e-2 1e-2 5e-4 5e-4 1e-3 1e-3 1e-3 1e-3 1e-3 1e-2 1e-3 1e-2 1e-3 1e-3 1e-3 1e-3 1e-3


; stopping criteria \rho_end=
1e-4
;1e-8

; bad value of the objective function
3

; Nuovo Pignone non-linear constaints hard coded into code must be
; - activated : 1
; - desactivated: 0
;1
1

; the data are inside a file called:
;/home/andromeda_2/fld_user/METHOD/TD21/data.ll
/home/fvandenb/L6/dataNEW.ll

; when waiting for the result of the evaluation of the objective
; function, we check every xxx seconds for an arrival of the file
; containing the results
3

Config file on client node

When running CONDOR inside a cluster of computers (to do parallel optimization), the user must start on each client-node the CONDOR-client-software. This software is performing the following:
  1. Wait to receive from the server a sampling site (a point) (using nearly no CPU time).
  2. Evaluate the objective function at this site and return immediately the result to the server. Go to step 1.
Each client node will use a different ``client configuration file''. These files contain simply the parameters 1 to 7 of the ''server configuration file'' (described in Section 11.4). The IP addresses and the port numbers of the client nodes are currently still hard coded inside the code (at the beginning of file ``parallel.cpp'').

next up previous contents
Next: Conclusions Up: The METHOD project Previous: Preliminary numerical results   Contents
Frank Vanden Berghen 2004-04-19