def show_reduction_steps(self, report, max_nsteps):
nsteps, nplans, ncplans = self.compute_reduction_steps(max_nsteps)
report.data('nsteps', nsteps)
report.data('nplans', nplans)
report.data('ncplans', ncplans)
f = report.figure()
caption = """
Efficient plan generations.
Let L be the length, N the number of commands,
and P the number of plans
The naive grows exponentially.
P_naive ~= N ^ L
The reduced grows according to the ambient space.
If the topology is that of Reals^K, then it grows like
P_smart ~= L ^ K
this is the *volume* of the area.
Usually the number of commands is N = 2*K
So we have:
N^L vs L^(N/2)
The logarithm is
log(P_naive) = log(N) * L
log(P_smart) = N * log(L)
So fixing N and plotting as a function of L and in logarithmic
coordinates, we have that the first plot should be a line,
and the second one a logarithm.
"""
report.text('explanation', caption)
xp = range(1, 6)
params = dict(figsize=(4.5, 4.5))
with f.plot('reduction', **params) as pylab:
pylab.semilogy(nsteps, nplans, 's-', label='naive')
pylab.semilogy(nsteps, ncplans, 'o-', label='reduced')
pylab.legend(loc='upper left')
pylab.xticks(xp, xp)
pylab.xlabel('plan length (L)')
pylab.ylabel('number of plans (P)')
x_axis_extra_space(pylab)
with f.plot('reduction2', **params) as pylab:
pylab.semilogy(nsteps, nplans, 's-', label='naive')
pylab.semilogy(nsteps, ncplans, 'o-', label='reduced')
pylab.xticks(xp, xp)
pylab.xlabel('plan length (L)')
pylab.ylabel('number of plans (P)')
x_axis_extra_space(pylab)
def show_reduction_steps(self, report, max_nsteps):
nsteps, nplans, ncplans = self.compute_reduction_steps(max_nsteps)
report.data('nsteps', nsteps)
report.data('nplans', nplans)
report.data('ncplans', ncplans)
f = report.figure()
caption = """
Efficient plan generations.
Let L be the length, N the number of commands,
and P the number of plans
The naive grows exponentially.
P_naive ~= N ^ L
The reduced grows according to the ambient space.
If the topology is that of Reals^K, then it grows like
P_smart ~= L ^ K
this is the *volume* of the area.
Usually the number of commands is N = 2*K
So we have:
N^L vs L^(N/2)
The logarithm is
log(P_naive) = log(N) * L
log(P_smart) = N * log(L)
So fixing N and plotting as a function of L and in logarithmic
coordinates, we have that the first plot should be a line,
and the second one a logarithm.
"""
report.text('explanation', caption)
xp = range(1, 6)
params = dict(figsize=(4.5, 4.5))
with f.plot('reduction', **params) as pylab:
pylab.semilogy(nsteps, nplans, 's-', label='naive')
pylab.semilogy(nsteps, ncplans, 'o-', label='reduced')
pylab.legend(loc='upper left')
pylab.xticks(xp, xp)
pylab.xlabel('plan length (L)')
pylab.ylabel('number of plans (P)')
x_axis_extra_space(pylab)
with f.plot('reduction2', **params) as pylab:
pylab.semilogy(nsteps, nplans, 's-', label='naive')
pylab.semilogy(nsteps, ncplans, 'o-', label='reduced')
pylab.xticks(xp, xp)
pylab.xlabel('plan length (L)')
pylab.ylabel('number of plans (P)')
x_axis_extra_space(pylab)
def figure_style1(pylab):
f = pylab.gcf()
size = 1.57 * 3
ratio = 1
f.set_size_inches((size, size * ratio))
ieee_spines(pylab)
yield pylab
y_axis_extra_space(pylab)
x_axis_extra_space(pylab)
def figure_style1(pylab):
f = pylab.gcf()
size = 1.57 * 3
ratio = 1
f.set_size_inches((size, size * ratio))
ieee_spines(pylab)
yield pylab
y_axis_extra_space(pylab)
x_axis_extra_space(pylab)
def plot_style_sensels(pylab):
y_axis_set(pylab, -0.1, 1.1)
x_axis_extra_space(pylab)
def plot_style_sensels_deriv(pylab):
y_axis_set(pylab, -0.1, 0.1)
x_axis_extra_space(pylab)
