def do_diffusion_plots_old():
import utils; utils.DEFAULT_H5_FILE = "data.h5"
import specs, models, analysis
for boundary_layer in [3,6,9,12,15]:
for light in [1,2,4,6,10,16,24,32]:
query = specs.make_query(boundary_layer='==%s'%boundary_layer,
light_penetration='==%s'%light)
plt.clf()
analysis.heights.phase_diagram_2d(
'diffusion_constant', 'uptake_rate',
spec_query=query, num_cells=15, statistic='max',
cmap=plt.get_cmap('hot'), vmin=0, vmax=64)
save_plot(path='diffusion/heights/max/boundary{0}_light{1}'.format(boundary_layer, light),
xlabel='Diffusion Constant',
ylabel='Uptake Rate')
plt.clf()
analysis.convex_density.phase_diagram_2d(
'diffusion_constant', 'uptake_rate',
spec_query=query, num_cells=15, statistic='mean',
cmap=plt.get_cmap('hot'), vmin=0, vmax=1)
save_plot(path='diffusion/convex_density/boundary{0}_light{1}'.format(boundary_layer, light),
xlabel='Diffusion Constant',
ylabel='Uptake Rate')
def do_coverage_plots():
import utils; utils.DEFAULT_H5_FILE = "coverages.h5"
import specs, models, analysis
for mass in [500, 1000, 1500, 2000, 2500]:
for spacing in [1, 2, 4, 6, 8, 10, 16, 24, 32, 64, 128]:
plt.clf()
for light in [0, 8, 16]:
query = specs.make_query(initial_cell_spacing='==%s'%spacing,
stop_on_mass='==%s'%mass,
light_penetration='==%s'%light)
analysis.coverages.average_curve_plot(spec_query=query, lw=3,
label='Light Penetration Depth = %s'%light)
plt.legend()
save_plot(path='coverages/mass%s_spacing%s'%(mass, spacing),
xlabel='Depth', ylabel='Coverage')
def do_diffusion_plots():
import utils; utils.DEFAULT_H5_FILE = "diffusion.h5"
import specs, models, analysis
for light in [0, 8]:
query = specs.make_query(light_penetration='==%s'%light)
plt.clf()
analysis.heights.phase_diagram_2d(
'diffusion_constant', 'uptake_rate',
spec_query=query, num_cells=40, statistic='max',
cmap=plt.get_cmap('hot'), vmin=5, vmax=35)
save_plot(path='diffusion/heights/max/light{0}'.format(light),
xlabel='Diffusion Constant',
ylabel='Uptake Rate')
plt.clf()
analysis.heights.phase_diagram_2d(
'diffusion_constant', 'uptake_rate',
spec_query=query, num_cells=40, statistic='mean',
cmap=plt.get_cmap('hot'), vmin=5, vmax=18)
save_plot(path='diffusion/heights/mean/light{0}'.format(light),
xlabel='Diffusion Constant',
ylabel='Uptake Rate')
plt.clf()
#levels = [0.6, 0.7, 0.8, 0.9, 1.0]
#analysis.convex_density.contour_plot(
# 'diffusion_constant', 'uptake_rate', levels=levels,
# spec_query=query, num_cells=100, statistic='mean',
# smoothing=0.5, cmap=plt.get_cmap('hot'), vmin=0.0, vmax=1.0)
analysis.convex_density.phase_diagram_2d(
'diffusion_constant', 'uptake_rate',
spec_query=query, num_cells=40, statistic='mean',
cmap=plt.get_cmap('hot'), vmin=0.6, vmax=1.0)
save_plot(path='diffusion/convex_density/light{0}'.format(light),
xlabel='Diffusion Constant',
ylabel='Uptake Rate')
def find_diffusion_line():
import utils; utils.DEFAULT_H5_FILE = "diffusion3.h5"
import specs, models, analysis
import random
# boundary = 5
# slope = 4.40
# query = '(boundary_layer=={0})&(uptake_rate >= {1}*diffusion_constant)'\
# .format(boundary, slope)
# high_growth = [models.Result.where('spec_uuid=="%s"'%spec.uuid)
# for spec in specs.Spec.where(query)
# if analysis.convex_density.get_by_spec(spec)['mean'] < 0.9]
# print len(high_growth)
# for result_set in high_growth:
# result = random.choice(list(result_set))
# spec = result.spec
# path = "fig/images/weird/b{0}_ur{1}_d{2}.png"\
# .format(boundary, spec.uptake_rate, spec.diffusion_constant)
# cv2.imwrite(path, result.int_image*255)
boundary = 5
query = specs.make_query(boundary_layer='==%s'%boundary,
uptake_rate=('>0.49', '<0.52'),
diffusion_constant='<0.25')
line = [models.Result.where('spec_uuid=="%s"'%spec.uuid)
for spec in sorted(specs.Spec.where(query),
key=lambda s:s.diffusion_constant)]
for i, result_set in enumerate(line):
result = random.choice(list(result_set))
spec = result.spec
path = "fig/images/weird/{0}_b{1}_ur{2}_d{3}.png"\
.format(i, boundary, spec.uptake_rate, spec.diffusion_constant)
cv2.imwrite(path, result.int_image*255)
def do_biomass_vs_light_plots_spaced():
import utils; utils.DEFAULT_H5_FILE = "biomass.h5"
import specs, models, analysis
from numpy.random import randint
plt.clf()
query = specs.make_query(light_penetration='==8', stop_on_mass='<4000')
x8, y8 = analysis.mass.scatter_plot(analysis.light_exposure, spec_query=query,
marker='o', mfc='none', ms=8,
label='Light Dependent (Penetration Depth = 8)')
query = specs.make_query(light_penetration='==0', stop_on_mass='<4000')
x0, y0 = analysis.mass.scatter_plot(analysis.light_exposure, spec_query=query,
mec='r', mfc='r', ms=4,
label='Light Independent', marker='.')
plt.semilogx()
plt.xlim([400, 2400])
powers = [2.6, 2.8, 3.0, 3.2, 3.4]
plt.xticks([10**x for x in powers], ['$10^{%s}$'%x for x in powers])
plt.legend(loc='lower right')
save_plot(path='new_spaced_biomass_vs_light_exposure_all',
xlabel='Biomass', ylabel='Light Exposure')
def dump_to_file(path, xs, ys):
with open(path, "w") as f:
f.write("biomass,light_exposure\n")
for x, y in zip(xs, ys):
f.write("%s,%s\n"%(x, y))
dump_to_file("fig/new_spaced_light_indep_data.txt", x0, y0)
dump_to_file("fig/new_spaced_light_dep_dep8_data.txt", x8, y8)
x_range = [400, 2400]
points = np.linspace(*x_range)
log_points = np.log10(points)
plt.xlim(x_range)
plt.ylim([220, 670])
def do_fit(x, y, label, **plot_args):
in_range = (x >= x_range[0])*(x <= x_range[1])
log_x = np.log10(x[in_range])
y = y[in_range]
fit = np.polyfit(log_x, y, 1, full=True)
p = np.poly1d(fit[0])
r2 = 1 - fit[1]/(y.size*y.var())
print label, p, r2
plt.plot(points, p(log_points), label=label, **plot_args)
return p
p0 = do_fit(x0, y0, 'Light Independent', c='r', lw=3)
p8 = do_fit(x8, y8, 'Light Dependent (Penetration Depth = 8)', ls='--', lw=3, c='k')
save_plot(path='new_spaced_biomass_vs_light_exposure_upper',
xlabel='Biomass', ylabel='Light Exposure')
def get_percent_diff(p1, p2):
return np.abs((p1(log_points)-p2(log_points))/p1(log_points)).mean()
real_diff = get_percent_diff(p0, p8)
print "Real diff", real_diff
pool = np.array(zip(x0, y0) + zip(x8, y8))
def sample_percent_diff():
def sample(size): return pool[randint(pool.shape[0], size=size), :]
subpool0, subpool8 = sample(x0.size), sample(x8.size)
def fit(subpool): return np.poly1d(np.polyfit(np.log10(subpool[:, 0]), subpool[:, 1], 1))
p0, p8 = fit(subpool0), fit(subpool8)
return get_percent_diff(p0, p8)
