def stats_gnuplot(tm, num_pods, num_tors_per_pod):
num_tors = num_pods * num_tors_per_pod
idx = 0
pod_tr = {}
sid = 0
for spod in xrange(num_pods):
for stor in xrange(num_tors_per_pod):
sid += 1
did = 0
for dpod in xrange(num_pods):
for dtor in xrange(num_tors_per_pod):
did += 1
key = (sid, did)
if key not in tm:
continue
bw = tm[key]
idx += 1
key = (spod, dpod)
if spod == dpod:
key = (0, 0)
if key not in pod_tr:
pod_tr[key] = []
pod_tr[key].append(bw)
data = sorted(pod_tr[(0, 0)])
dlen = len(data)
data_max = data[-1]
y = map(lambda x: (float(x) / dlen), xrange(dlen))
x = data
g = Gnuplot.Gnuplot(debug=0)
g('set terminal png')
g('set output "test.png"')
for i in range(1, 9):
dm = data_max * i / 8
print dm
g("set arrow from %d,0 to %d,1 nohead lc rgb 'red'" % (dm, dm))
g.plot(Gnuplot.Data(zip(x, y)))
exit(1)
def demo():
"""Demonstrate the Gnuplot package."""
# A straightforward use of gnuplot. The `debug=1' switch is used
# in these examples so that the commands that are sent to gnuplot
# are also output on stderr.
g = gnuplot.Gnuplot(debug=1)
g.title('A simple example') # (optional)
g('set style data linespoints') # give gnuplot an arbitrary command
# Plot a list of (x, y) pairs (tuples or a numpy array would
# also be OK):
g.plot([[0, 1.1], [1, 5.8], [2, 3.3], [3, 4.2]])
input('Please press return to continue...\n')
g.reset()
# Plot one dataset from an array and one via a gnuplot function;
# also demonstrate the use of item-specific options:
x = arange(10, dtype='float_')
y1 = x**2
# Notice how this plotitem is created here but used later? This
# is convenient if the same dataset has to be plotted multiple
# times. It is also more efficient because the data need only be
# written to a temporary file once.
d = gnuplot.Data(x, y1, title='calculated by python', with_='points 3 3')
g.title('Data can be computed by python or gnuplot')
g.xlabel('x')
g.ylabel('x squared')
# Plot a function alongside the Data PlotItem defined above:
g.plot(gnuplot.Func('x**2', title='calculated by gnuplot'), d)
input('Please press return to continue...\n')
# Save what we just plotted as a color postscript file.
# With the enhanced postscript option, it is possible to show `x
# squared' with a superscript (plus much, much more; see `help set
# term postscript' in the gnuplot docs). If your gnuplot doesn't
# support enhanced mode, set `enhanced=0' below.
g.ylabel('x^2') # take advantage of enhanced postscript mode
g.hardcopy('gp_test.ps', enhanced=1, color=1)
print('\n******** Saved plot to postscript file "gp_test.ps" ********\n')
input('Please press return to continue...\n')
g.reset()
# Demonstrate a 3-d plot:
# set up x and y values at which the function will be tabulated:
x = arange(35) / 2.0
y = arange(30) / 10.0 - 1.5
# Make a 2-d array containing a function of x and y. First create
# xm and ym which contain the x and y values in a matrix form that
# can be `broadcast' into a matrix of the appropriate shape:
xm = x[:, newaxis]
ym = y[newaxis, :]
m = (sin(xm) + 0.1 * xm) - ym**2
g('set parametric')
g('set style data lines')
g('set hidden')
g('set contour base')
g.title('An example of a surface plot')
g.xlabel('x')
g.ylabel('y')
# The `binary=1' option would cause communication with gnuplot to
# be in binary format, which is considerably faster and uses less
# disk space. (This only works with the splot command due to
# limitations of gnuplot.) `binary=1' is the default, but here we
# disable binary because older versions of gnuplot don't allow
# binary data. Change this to `binary=1' (or omit the binary
# option) to get the advantage of binary format.
g.splot(gnuplot.GridData(m, x, y, binary=0))
input('Please press return to continue...\n')
# plot another function, but letting GridFunc tabulate its values
# automatically. f could also be a lambda or a global function:
def f(x, y):
return 1.0 / (1 + 0.01 * x**2 + 0.5 * y**2)
g.splot(gnuplot.funcutils.compute_GridData(x, y, f, binary=0))
input('Please press return to continue...\n')
def mainFollow(self, x):
"""
Main following algorithm [Nesterov, p. 202]
Args:
x (Matrix): good approximation of the analytic center, used as the starting point of the algorithm
Returns:
Matrix: found optimal solution of the problem
"""
if self.drawPlot:
x0All = [x[0, 0]]
x1All = [x[1, 0]]
self.xMF = x
# Main path-following scheme [Nesterov, p. 202]
self.logStdout.info('\nMAIN PATH-FOLLOWING')
# initialization of the iteration process
t = 0
k = 0
# print the input condition to verify that is satisfied
if self.logStdout.isEnabledFor(logging.INFO):
Fd, Fdd, _ = Utils.gradientHessian(self.AAll, x)
self.logStdout.info('Input condition = ' +
str(Utils.LocalNormA(Fd, Fdd)))
while True:
k += 1
if self.logStdout.isEnabledFor(logging.INFO):
self.logStdout.info('\nk = ' + str(k))
# gradient and hessian
Fd, Fdd, A = Utils.gradientHessian(self.AAll, x)
# iteration step
t = t + self.gamma / Utils.LocalNormA(self.c, Fdd)
x = x - solve(Fdd, (t * self.c + Fd))
if self.drawPlot:
x0All.append(x[0, 0])
x1All.append(x[1, 0])
self.xMF = hstack((self.xMF, x))
if self.logStdout.isEnabledFor(logging.INFO):
self.logStdout.info('t = ' + str(t))
self.logStdout.info('x = ' + str(x))
# print eigenvalues
if self.logStdout.isEnabledFor(logging.INFO):
for i in range(0, len(A)):
eigs, _ = eig(A[i])
eigs.sort()
self.logStdout.info('EIG[' + str(i) + '] = ' + str(eigs))
# breaking condition
if self.logStdout.isEnabledFor(logging.INFO):
self.logStdout.info('Breaking condition = ' +
str(self.eps * t))
if self.eps * t >= self.nu + (
self.beta + sqrt(self.nu)) * self.beta / (1 - self.beta):
break
self.solved = True
self.result = x
self.resultA = A
self.iterations += k
# plot main path
if self.drawPlot:
self.mainPathPlotFile = tempfile.NamedTemporaryFile()
self.gnuplot.replot(
gp.Data(x0All,
x1All,
title='Main path',
with_='points pt 1',
filename=self.mainPathPlotFile.name))
print('\nPress enter to continue')
input()
return x
def dampedNewton(self, start):
"""
Damped Newton method for analytics center [Nesterov, p. 204]
Args:
start (Matrix): the starting point of the algorithm
Returns:
Matrix: approximation of the analytic center
"""
k = 0
# starting point
y = start
if self.drawPlot:
y0All = [y[0, 0]]
y1All = [y[1, 0]]
self.xAC = y
# gradient and hessian
Fd, Fdd, _ = Utils.gradientHessian(self.AAll, y)
FdLN = Utils.LocalNormA(Fd, Fdd)
self.logStdout.info('AUXILIARY PATH-FOLLOWING')
while True:
k += 1
if self.logStdout.isEnabledFor(logging.INFO):
self.logStdout.info('\nk = ' + str(k))
# iteration step
y = y - solve(Fdd, Fd) / (1 + FdLN)
if self.logStdout.isEnabledFor(logging.INFO):
self.logStdout.info('y = ' + str(y))
if self.drawPlot:
y0All.append(y[0, 0])
y1All.append(y[1, 0])
self.xAC = hstack((self.xAC, y))
# gradient and hessian
Fd, Fdd, A = Utils.gradientHessian(self.AAll, y)
FdLN = Utils.LocalNormA(Fd, Fdd)
# print eigenvalues
if self.logStdout.isEnabledFor(logging.INFO):
for i in range(0, len(A)):
eigs, _ = eig(A[i])
eigs.sort()
self.logStdout.info('EIG[' + str(i) + '] = ' + str(eigs))
# breaking condition
if FdLN <= self.beta:
break
# save number of iterations
self.iterations = k
# plot auxiliary path
if self.drawPlot:
self.dampedNewtonPlotFile = tempfile.NamedTemporaryFile()
self.gnuplot.replot(
gp.Data(y0All,
y1All,
title='Damped Newton',
with_='points pt 1',
filename=self.dampedNewtonPlotFile.name))
return y
def auxFollow(self, start):
"""
Auxiliary path-following algorithm [Nesterov, p. 205]
Warning: This function appears to be unstable. Please use dampedNewton instead.
Args:
start (Matrix): the starting point of the algorithm
Returns:
Matrix: approximation of the analytic center
"""
t = 1
k = 0
# starting point
y = start
if self.drawPlot:
y0All = [y[0, 0]]
y1All = [y[1, 0]]
# gradient and hessian
Fd, Fdd, A = Utils.gradientHessian(self.AAll, y)
Fd0 = Fd
FddInv = inv(Fdd)
# print eigenvalues
if self.logStdout.isEnabledFor(logging.INFO):
for i in range(0, len(A)):
eigs, _ = eig(A[i])
eigs.sort()
self.logStdout.info('EIG[' + str(i) + '] = ' + str(eigs))
self.logStdout.info('AUXILIARY PATH-FOLLOWING')
while True:
k += 1
self.logStdout.info('\nk = ' + str(k))
# iteration step
t = t - self.gamma / Utils.LocalNorm(Fd0, FddInv)
y = y - dot(FddInv, (t * Fd0 + Fd))
#self.logStdout.info('t = ' + str(t))
self.logStdout.info('y = ' + str(y))
if self.drawPlot:
y0All.append(y[0, 0])
y1All.append(y[1, 0])
# gradient and hessian
Fd, Fdd, A = Utils.gradientHessian(self.AAll, y)
FddInv = inv(Fdd)
# print eigenvalues
if self.logStdout.isEnabledFor(logging.INFO):
for i in range(0, len(A)):
eigs, _ = eig(A[i])
eigs.sort()
self.logStdout.info('EIG[' + str(i) + '] = ' + str(eigs))
# breaking condition
if Utils.LocalNorm(
Fd, FddInv) <= sqrt(self.beta) / (1 + sqrt(self.beta)):
break
# prepare x
x = y - dot(FddInv, Fd)
# save number of iterations
self.iterations = k
# plot auxiliary path
if self.drawPlot:
self.gnuplot.replot(
gp.Data(y0All,
y1All,
title='Auxiliary path',
with_='points pt 1',
filename='tmp/auxiliaryPath.dat'))
return x
def main():
"""Exercise the Gnuplot module."""
print (
'This program exercises many of the features of Gnuplot.py. The\n'
'commands that are actually sent to gnuplot are printed for your\n'
'enjoyment.'
)
wait('Popping up a blank gnuplot window on your screen.')
g = gnuplot.Gnuplot(debug=1)
g.clear()
# Make two temporary files:
if hasattr(tempfile, 'mkstemp'):
(fd, filename1,) = tempfile.mkstemp(text=1)
f = os.fdopen(fd, 'w')
(fd, filename2,) = tempfile.mkstemp(text=1)
else:
filename1 = tempfile.mktemp()
f = open(filename1, 'w')
filename2 = tempfile.mktemp()
try:
for x in numpy.arange(100.)/5. - 10.:
f.write('%s %s %s\n' % (x, math.cos(x), math.sin(x)))
f.close()
print ('############### test Func ###################################')
wait('Plot a gnuplot-generated function')
g.plot(gnuplot.Func('sin(x)'))
wait('Set title and axis labels and try replot()')
g.title('Title')
g.xlabel('x')
g.ylabel('y')
g.replot()
wait('Style linespoints')
g.plot(gnuplot.Func('sin(x)', with_='linespoints'))
wait('title=None')
g.plot(gnuplot.Func('sin(x)', title=None))
wait('title="Sine of x"')
g.plot(gnuplot.Func('sin(x)', title='Sine of x'))
wait('axes=x2y2')
g.plot(gnuplot.Func('sin(x)', axes='x2y2', title='Sine of x'))
print ('Change Func attributes after construction:')
f = gnuplot.Func('sin(x)')
wait('Original')
g.plot(f)
wait('Style linespoints')
f.set_option(with_='linespoints')
g.plot(f)
wait('title=None')
f.set_option(title=None)
g.plot(f)
wait('title="Sine of x"')
f.set_option(title='Sine of x')
g.plot(f)
wait('axes=x2y2')
f.set_option(axes='x2y2')
g.plot(f)
print ('############### test File ###################################')
wait('Generate a File from a filename')
g.plot(gnuplot.File(filename1))
wait('Style lines')
g.plot(gnuplot.File(filename1, with_='lines'))
wait('using=1, using=(1,)')
g.plot(gnuplot.File(filename1, using=1, with_='lines'),
gnuplot.File(filename1, using=(1,), with_='points'))
wait('using=(1,2), using="1:3"')
g.plot(gnuplot.File(filename1, using=(1,2)),
gnuplot.File(filename1, using='1:3'))
wait('every=5, every=(5,)')
g.plot(gnuplot.File(filename1, every=5, with_='lines'),
gnuplot.File(filename1, every=(5,), with_='points'))
wait('every=(10,None,0), every="10::5"')
g.plot(gnuplot.File(filename1, with_='lines'),
gnuplot.File(filename1, every=(10,None,0)),
gnuplot.File(filename1, every='10::5'))
wait('title=None')
g.plot(gnuplot.File(filename1, title=None))
wait('title="title"')
g.plot(gnuplot.File(filename1, title='title'))
print ('Change File attributes after construction:')
f = gnuplot.File(filename1)
wait('Original')
g.plot(f)
wait('Style linespoints')
f.set_option(with_='linespoints')
g.plot(f)
wait('using=(1,3)')
f.set_option(using=(1,3))
g.plot(f)
wait('title=None')
f.set_option(title=None)
g.plot(f)
print ('############### test Data ###################################')
x = numpy.arange(100)/5. - 10.
y1 = numpy.cos(x)
y2 = numpy.sin(x)
d = numpy.transpose((x,y1,y2))
wait('Plot Data against its index')
g.plot(gnuplot.Data(y2, inline=0))
wait('Plot Data, specified column-by-column')
g.plot(gnuplot.Data(x,y2, inline=0))
wait('Same thing, saved to a file')
gnuplot.Data(x,y2, inline=0, filename=filename1)
g.plot(gnuplot.File(filename1))
wait('Same thing, inline data')
g.plot(gnuplot.Data(x,y2, inline=1))
wait('Plot Data, specified by an array')
g.plot(gnuplot.Data(d, inline=0))
wait('Same thing, saved to a file')
gnuplot.Data(d, inline=0, filename=filename1)
g.plot(gnuplot.File(filename1))
wait('Same thing, inline data')
g.plot(gnuplot.Data(d, inline=1))
wait('with_="lp 4 4"')
g.plot(gnuplot.Data(d, with_='lp 4 4'))
wait('cols=0')
g.plot(gnuplot.Data(d, cols=0))
wait('cols=(0,1), cols=(0,2)')
g.plot(gnuplot.Data(d, cols=(0,1), inline=0),
gnuplot.Data(d, cols=(0,2), inline=0))
wait('Same thing, saved to files')
gnuplot.Data(d, cols=(0,1), inline=0, filename=filename1)
gnuplot.Data(d, cols=(0,2), inline=0, filename=filename2)
g.plot(gnuplot.File(filename1), gnuplot.File(filename2))
wait('Same thing, inline data')
g.plot(gnuplot.Data(d, cols=(0,1), inline=1),
gnuplot.Data(d, cols=(0,2), inline=1))
wait('Change title and replot()')
g.title('New title')
g.replot()
wait('title=None')
g.plot(gnuplot.Data(d, title=None))
wait('title="Cosine of x"')
g.plot(gnuplot.Data(d, title='Cosine of x'))
print ('############### test compute_Data ###########################')
x = numpy.arange(100)/5. - 10.
wait('Plot Data, computed by Gnuplot.py')
g.plot(
gnuplot.funcutils.compute_Data(x, lambda x: math.cos(x), inline=0)
)
wait('Same thing, saved to a file')
gnuplot.funcutils.compute_Data(
x, lambda x: math.cos(x), inline=0, filename=filename1
)
g.plot(gnuplot.File(filename1))
wait('Same thing, inline data')
g.plot(gnuplot.funcutils.compute_Data(x, math.cos, inline=1))
wait('with_="lp 4 4"')
g.plot(gnuplot.funcutils.compute_Data(x, math.cos, with_='lp 4 4'))
print ('############### test hardcopy ###############################')
print ('******** Generating postscript file "gp_test.ps" ********')
wait()
g.plot(gnuplot.Func('cos(0.5*x*x)', with_='linespoints 2 2',
title='cos(0.5*x^2)'))
g.hardcopy('gp_test.ps')
wait('Testing hardcopy options: mode="eps"')
g.hardcopy('gp_test.ps', mode='eps')
wait('Testing hardcopy options: mode="landscape"')
g.hardcopy('gp_test.ps', mode='landscape')
wait('Testing hardcopy options: mode="portrait"')
g.hardcopy('gp_test.ps', mode='portrait')
wait('Testing hardcopy options: eps=1')
g.hardcopy('gp_test.ps', eps=1)
wait('Testing hardcopy options: mode="default"')
g.hardcopy('gp_test.ps', mode='default')
wait('Testing hardcopy options: enhanced=1')
g.hardcopy('gp_test.ps', enhanced=1)
wait('Testing hardcopy options: enhanced=0')
g.hardcopy('gp_test.ps', enhanced=0)
wait('Testing hardcopy options: color=1')
g.hardcopy('gp_test.ps', color=1)
# For some reason,
# g.hardcopy('gp_test.ps', color=0, solid=1)
# doesn't work here (it doesn't activate the solid option), even
# though the command sent to gnuplot looks correct. I'll
# tentatively conclude that it is a gnuplot bug. ###
wait('Testing hardcopy options: color=0')
g.hardcopy('gp_test.ps', color=0)
wait('Testing hardcopy options: solid=1')
g.hardcopy('gp_test.ps', solid=1)
wait('Testing hardcopy options: duplexing="duplex"')
g.hardcopy('gp_test.ps', solid=0, duplexing='duplex')
wait('Testing hardcopy options: duplexing="defaultplex"')
g.hardcopy('gp_test.ps', duplexing='defaultplex')
wait('Testing hardcopy options: fontname="Times-Italic"')
g.hardcopy('gp_test.ps', fontname='Times-Italic')
wait('Testing hardcopy options: fontsize=20')
g.hardcopy('gp_test.ps', fontsize=20)
print ('******** Generating svg file "gp_test.svg" ********')
wait()
g.plot(gnuplot.Func('cos(0.5*x*x)', with_='linespoints 2 2',
title='cos(0.5*x^2)'))
g.hardcopy('gp_test.svg', terminal='svg')
wait('Testing hardcopy svg options: enhanced')
g.hardcopy('gp_test.ps', terminal='svg', enhanced='1')
print ('############### test shortcuts ##############################')
wait('plot Func and Data using shortcuts')
g.plot('sin(x)', d)
print ('############### test splot ##################################')
wait('a 3-d curve')
g.splot(gnuplot.Data(d, with_='linesp', inline=0))
wait('Same thing, saved to a file')
gnuplot.Data(d, inline=0, filename=filename1)
g.splot(gnuplot.File(filename1, with_='linesp'))
wait('Same thing, inline data')
g.splot(gnuplot.Data(d, with_='linesp', inline=1))
print ('############### test GridData and compute_GridData ##########')
# set up x and y values at which the function will be tabulated:
x = numpy.arange(35)/2.0
y = numpy.arange(30)/10.0 - 1.5
# Make a 2-d array containing a function of x and y. First create
# xm and ym which contain the x and y values in a matrix form that
# can be `broadcast' into a matrix of the appropriate shape:
xm = x[:,numpy.newaxis]
ym = y[numpy.newaxis,:]
m = (numpy.sin(xm) + 0.1*xm) - ym**2
wait('a function of two variables from a GridData file')
g('set parametric')
g('set style data lines')
g('set hidden')
g('set contour base')
g.xlabel('x')
g.ylabel('y')
g.splot(gnuplot.GridData(m,x,y, binary=0, inline=0))
wait('Same thing, saved to a file')
gnuplot.GridData(m,x,y, binary=0, inline=0, filename=filename1)
g.splot(gnuplot.File(filename1, binary=0))
wait('Same thing, inline data')
g.splot(gnuplot.GridData(m,x,y, binary=0, inline=1))
wait('The same thing using binary mode')
g.splot(gnuplot.GridData(m,x,y, binary=1))
wait('Same thing, using binary mode and an intermediate file')
gnuplot.GridData(m,x,y, binary=1, filename=filename1)
g.splot(gnuplot.File(filename1, binary=1))
wait('The same thing using compute_GridData to tabulate function')
g.splot(gnuplot.funcutils.compute_GridData(
x,y, lambda x,y: math.sin(x) + 0.1*x - y**2,
))
wait('Same thing, with an intermediate file')
gnuplot.funcutils.compute_GridData(
x,y, lambda x,y: math.sin(x) + 0.1*x - y**2,
filename=filename1)
g.splot(gnuplot.File(filename1, binary=1))
wait('Use compute_GridData in ufunc and binary mode')
g.splot(gnuplot.funcutils.compute_GridData(
x,y, lambda x,y: numpy.sin(x) + 0.1*x - y**2,
ufunc=1, binary=1,
))
wait('Same thing, with an intermediate file')
gnuplot.funcutils.compute_GridData(
x,y, lambda x,y: numpy.sin(x) + 0.1*x - y**2,
ufunc=1, binary=1,
filename=filename1)
g.splot(gnuplot.File(filename1, binary=1))
wait('And now rotate it a bit')
for view in range(35,70,5):
g('set view 60, %d' % view)
g.replot()
time.sleep(1.0)
wait(prompt='Press return to end the test.\n')
finally:
os.unlink(filename1)
os.unlink(filename2)
gnuplot = gp.Gnuplot()
gnuplot('set xlabel "Dimension"')
gnuplot('set ylabel "Time [s]"')
gnuplot('set title "Performance of the POP solver based on the number of variables."')
# plot data
tempFiles = []
for df in range(0, shape[1]):
for dr in range(0, shape[2]):
dataSliced = data[:, df, dr]
noNan = nonzero([~isnan(dataSliced)])[1]
a = dataSliced[noNan]
if a.shape[0] != 0:
plotFile = tempfile.NamedTemporaryFile()
tempFiles.append(plotFile)
plot = gnuplot.replot(gp.Data(noNan, a, title = 'df = ' + str(df) + ', relaxOrder = ' + str(dr), with_ = 'lines lw 1.5', filename=plotFile.name))
print('\nPress enter to continue')
input()
# close temp file
for tempFile in tempFiles:
tempFile.close()
# based on relaxation order
if args.relaxOrder:
# start GnuPlot
gnuplot = gp.Gnuplot()
gnuplot('set xlabel "Relaxation order"')
gnuplot('set ylabel "Time [s]"')
gnuplot('set title "Performance of the POP solver based on the relaxation order."')
parts = path[:-4].split('_')
print(parts)
date = parts[1][0:4] + '.' + parts[1][4:6] + '.' + parts[1][
6:8] + ' ' + parts[2][0:2] + ':' + parts[2][2:4]
# plot them
dims = []
plotData = []
for n in range(0, shape[0]):
if not isnan(data[n]):
dims.append(n)
plotData.append(data[n])
# plot
plotFile = tempfile.NamedTemporaryFile()
tempFiles.append(plotFile)
plot = gnuplot.replot(
gp.Data(dims,
plotData,
title=date + ' ' + parts[3],
with_='lines linewidth 1.5',
filename=plotFile.name))
print('\nPress enter to continue')
input()
# close temp files
for tempFile in tempFiles:
tempFile.close()