Calling Matlab from a Scipt
Jump to navigation
Jump to search
It is possible to call matlab from a script to perform some calculation needed in the script. It is made possible by using the -r flag among others when invoking matlab from the command line (dos shell). The following is a little example demonstrating how this is done.
The example is a simple truss example where I am trying to determine the value of the parameter A needed to hit a target displacement.
- I start off with setting some variables, my target displacement, a tolerance, and two bounding values on my parameter of interest Amin and Amax.
- I then define a proc getA(). In this procedur I write the two bounding values to a file, I invoke a matlab script to determine a new A, and then I read the result from a file.
- Next I create my truss model and the analysis and perform a single analysis step
- Then I iterate until i have converged on a solution or hit a bounding value. In the iteration I call the matlab function to determine a new A, set the parameter in the elements, do an analysis step to obtain new displacement, and check versus target dispacement
- Finally I print a message indicating succcess or failure
#
# set some variables
#
set targetDisp 1.2; # target displacement
set tol 0.01; # tolerance
set Amin 1.0
set Amax 10.0
#
# define a procedure: getA() which calls matlab script getA.m in which our new A is computed
#
proc getA {Amin Amax} {
# write current min and max values
set fileID [open data1 w]
puts $fileID "$Amin $Amax"
close $fileID
# invoke matlab
if {[catch {exec matlab -nosplash -nodesktop -r "getA; quit"}]} {
puts "Ignore this $msg"
}
# read matlab result
set fileIN [open data2 r]
set fileData [read $fileIN]
close $fileIN
#cleanup
file delete data1
file delete data2
# return result
set lines [split $fileData "\n"]
set A [lindex $lines 0]
}
set A [getA $Amin $Amax]
#
# create the model & perform one analysis step
#
model BasicBuilder -ndm 2 -ndf 2
node 1 0.0 0.0
node 2 144.0 0.0
node 3 168.0 0.0
node 4 72.0 96.0
fix 1 1 1
fix 2 1 1
fix 3 1 1
uniaxialMaterial Elastic 1 3000
element truss 1 1 4 $A 1
element truss 2 2 4 $A 1
element truss 3 3 4 $A 1
timeSeries Constant 1
pattern Plain 1 1 {
load 4 100 -50
}
#create analysis
system BandSPD
numberer RCM
constraints Plain
integrator LoadControl 1.0
algorithm Linear
analysis Static
analyze 1
#
# iterate until convergence is reached, or failure
#
set xDisp [nodeDisp 4 1]
set done 0
while {$done == 0} {
if {$xDisp > $targetDisp} {
set Amin $A
} else {
set Amax $A
}
set diffA [expr $Amax-$Amin]
if {$diffA < $tol} {
set done 2
break
}
set A [getA $Amin $Amax]
setParameter -value $A -ele A
analyze 1
set xDisp [nodeDisp 4 1]
set diff [expr abs($xDisp - $targetDisp)]
if {$diff < $tol} {
set done 1
}
puts "$A $xDisp"
}
#
# print success or failure
#
if {$done == 1} {
puts "SUCCESS: A = $A gives Displacement of $xDisp versus target of $targetDisp"
} else {
puts "FAILURE: Hit limit A = $A gives Displacement of $xDisp versus target of $targetDisp"
}
The matlab scipt contained in the file getA.m is a simple divide-and-conquer.
load data1;
a=(data1(1,1)+data1(1,2))/2.0;
dlmwrite('data2', a);