def generate_tree_vs_bassic(tree_file, bassic_file, output_name=None):
results_tree = parse_file(tree_file)
results_bassic = parse_file(bassic_file)
x_values = results_tree['x'] #it has 5, bassic only 4
times_tree = results_tree['y1']
times_bassic = results_bassic['y1']
distances_tree = results_tree['y2']
distances_bassic = results_bassic['y2']
axis_x, axis_y1, axis_y2 = create_axis_group(
"Algorithm's depth (n)",
"Tolerance (%1)", BLUES[0],
"Time (s)", GREENS[0] )
trace_accurance_bassic = create_scatter(
x=x_values,
y=distances_bassic,
mode='markers+lines',
name='bassic tolerance',
color=BLUES[0]
)
trace_time_bassic = create_scatter(
x=x_values,
y=times_bassic,
mode='markers+lines',
name='bassic time',
color=GREENS[0],
yaxis='y2'
)
trace_accurance_tree = create_scatter(
x=x_values,
y=distances_tree,
mode='markers+lines',
name='tree tolerance',
color=BLUES[5],
symbol='square'
)
trace_time_tree = create_scatter(
x=x_values,
y=times_tree,
mode='markers+lines',
name='tree time',
color=GREENS[5],
yaxis='y2',
symbol='square'
)
layout = create_layout(title=go.layout.Title(
text="Results kd-tree finder vs bassic (largest group)",
xref='paper',
y=0
), xaxis=axis_x,
yaxis=axis_y1, yaxis2=axis_y2, legend={'x': 1.1,'xanchor': 'left', 'y': 1 })
traces = [trace_accurance_bassic, trace_time_bassic, trace_accurance_tree, trace_time_tree]
finish(tree_file, output_name, traces, layout)
def generate_n_from_error(errors_dict, errors_dict2={}):
# n = ln( ln(eph * 32 )/ln(0.14 * 32 ) / 0.405
x_values, y_values = get_values(errors_dict)
x_values_2, y_values_2 = get_values(errors_dict2)
axis_x, axis_y1, _ = create_axis_group(
"Error reached (%1)",
"Algorithm's depth (n)",
x_args={'type': 'log'},
y1_color=BLUES[2],
y1_args={'rangemode': 'nonnegative'})
trace = dots_trace(x_values,
y_values,
name="n level large group",
color=BLUES[2])
trace_2 = dots_trace(x_values_2,
y_values_2,
name="n level other groups",
color=GREENS[2])
x_lin = np.linspace(5e-8, 0.05, 500)
trace_real = create_scatter(
x=x_lin,
y=np.ceil(np.log(np.log(x_lin * 32) / np.log(0.014 * 32)) / 0.405),
color=REDS[2],
yaxis='y1',
mode="lines",
name="theorical n(error) 0.014")
trace_real2 = create_scatter(
x=x_lin,
y=np.ceil(np.log(np.log(x_lin * 32) / np.log(0.009 * 32)) / 0.405),
color=REDS[6],
yaxis='y1',
mode="lines",
name="theorical n(error) 0.009")
layout = create_layout("Error reached for depth levels",
xaxis=axis_x,
yaxis=axis_y1,
width=1000,
legend={
'x': 1.1,
'xanchor': 'left',
'y': 1
})
output_name = "n_from_error"
finish("", output_name, [trace, trace_2, trace_real, trace_real2], layout)
def generate_time_from_error(errors, times, errors_2, times_2):
# n = ln( ln(eph * 32 )/ln(0.14 * 32 ) / 0.405
y_values, x_values = times, errors
y_values_2, x_values_2 = times_2, errors_2
axis_x, axis_y1, _ = create_axis_group("Error reached (%1)",
"Time required (s)",
x_args={'type': 'log'},
y1_color=BLUES[2],
y1_args={
'type': 'log',
'rangemode': 'nonnegative'
})
trace = dots_trace(x_values,
y_values,
name="error large group",
color=BLUES[2])
trace_2 = dots_trace(x_values_2,
y_values_2,
name="error other groups",
color=GREENS[2])
x_lin = np.linspace(1e-7, 1, 100)
trace_real = create_scatter(x=x_lin,
y=np.log(1 / x_lin)**2.71,
color=REDS[2],
yaxis='y1',
mode="lines",
name="theorical time(error)")
trace_real2 = create_scatter(x=x_lin,
y=0.003 * np.log(1 / x_lin)**2.71,
color=REDS[6],
yaxis='y1',
mode="lines",
name="theorical time(error) x0.003")
layout = create_layout("Time required for reached errors",
xaxis=axis_x,
yaxis=axis_y1,
width=1000,
legend={
'x': 1.1,
'xanchor': 'left',
'y': 1
})
output_name = "time_required_give_error"
finish("", output_name, [trace, trace_2, trace_real, trace_real2], layout)
def _complete_stuff(results,
traces,
sk_type,
color_scale,
symbol,
text=False,
pos=None):
densities, times, distances = results
studing_densities = densities if pos == None else densities[pos:pos +
1]
for jj, density in enumerate(studing_densities):
index = jj if pos == None else p
x_values = [t[index] for ene, t in times.iteritems()]
y_values = [err[index] for ene, err in distances.iteritems()]
traces.append(
create_scatter(
x=x_values,
y=y_values,
name="%s d=%d" % (sk_type, density),
mode="lines+markers+text",
text=['n=%s' % (nn + 1)
for nn in range(len(x_values))] if text else [],
textposition='top center',
textfont=dict(color=color_scale[index]),
color=color_scale[index],
symbol=symbol))
def generate_error_from_n(errors_dict, errors_dict2={}):
# n = ln( ln(eph * 32 )/ln(0.14 * 32 ) / 0.405
y_values, x_values = get_values(errors_dict)
y_values_2, x_values_2 = get_values(errors_dict2)
axis_x, axis_y1, _ = create_axis_group("Algorithm's depth (n)",
"Error reached (%1)",
y1_args={'type': 'log'},
y1_color=BLUES[2],
x_args={'rangemode': 'nonnegative'})
trace = dots_trace(x_values,
y_values,
name="error large group",
color=BLUES[2])
trace_2 = dots_trace(x_values_2,
y_values_2,
name="error other groups",
color=GREENS[2])
x_lin = np.linspace(1, 6, 100)
import pdb
pdb.set_trace()
trace_real = create_scatter(x=x_lin,
y=0.0325 *
np.power(0.014 * 32, np.power(1.5, x_lin)),
color=REDS[2],
yaxis='y1',
mode="lines",
name="theorical error(n) 0.014")
trace_real2 = create_scatter(x=x_lin,
y=0.0325 * (0.009 * 32)**((1.5)**x_lin),
color=REDS[6],
yaxis='y1',
mode="lines",
name="theorical error(n) 0.009")
layout = create_layout("Level of recursion required for given errors",
xaxis=axis_x,
yaxis=axis_y1,
width=1000,
legend={
'x': 1.1,
'xanchor': 'left',
'y': 1
})
output_name = "error_from_n"
finish("", output_name, [trace, trace_2, trace_real, trace_real2], layout)
def dots_trace(x_values, y_values, name, color):
return create_scatter(
x=x_values,
y=y_values,
name=name,
mode="markers",
#text=['n=%s'%(nn+1) for nn in range(len(x_values))],
color=color,
yaxis='y1')
def generate_distances_and_times(filesnames, output_name=None):
x_values, times, distances = parse_several(filesnames)
traces_times = []
traces_distances = []
axis_x, axis_y1, axis_y2 = create_axis_group(
"Group density (#items)",
y1_color=BLUES[2],
y1_title="Error for each density (%1)",
y2_color=GREENS[2],
y2_title="Time for each density (s)",
y1_args={'type': 'log'},
y2_args={'type': 'log'})
ns = times.keys()
for nn in ns:
traces_times.append(
create_scatter(x=x_values,
y=times[nn],
mode='markers+lines+text',
name="time for n=%s" % nn,
symbol='square',
text=[''] * 3 + ['n=%s' % nn],
textposition='top center',
color=GREENS[int(nn) - 1],
yaxis='y2'))
traces_distances.append(
create_scatter(x=x_values,
y=distances[nn],
mode='markers+lines+text',
name="distance for n=%s" % nn,
text=['n=%s' % nn],
textposition='top center',
color=BLUES[int(nn) - 1]))
layout = create_layout("Distace and time increasing group density",
xaxis=axis_x,
yaxis=axis_y1,
yaxis2=axis_y2,
width=1000,
legend={
'x': 1.1,
'xanchor': 'left',
'y': 1
})
_finish('distance_and_time', output_name, traces_times + traces_distances,
layout)
def get_rate_trace(x_values, y_values, color=REDS[0], width=4):
trace = create_scatter(x=x_values,
y=y_values,
mode='markers+lines',
color=color,
name='theoretical',
line=dict(color=color, width=width))
return trace
def _random_vs_shorter_stuff(directory, output_name, yaxis_title, key_results, legend_text, plot_title, color1, color2):
random_files = search_files(directory, "*finder-random-kdtree*.csv")
shorter_files = search_files(directory, "*finder-kdtree*.csv")
results = several_random_shorter_parse(random_files, shorter_files)
x_values = results['x']
axis_x, axis_y1, _ = create_axis_group(
"Group density (#items)",
yaxis_title, y1_args={'type': 'log'})
traces_shorter = []
traces_random = []
for nn, values in results[key_results + '_random'].items():
traces_random.append(
create_scatter(
x=x_values,
y=values,
mode='markers+lines',
name='random %s n=%s'%(legend_text, nn),
color=color1[int(nn)-1],
symbol='square'
)
)
for nn, values in results[key_results + '_shorter'].items():
traces_shorter.append(
create_scatter(
x=x_values,
y=values,
text=['n=%s'%nn],
textposition='top center',
mode='text+markers+lines',
name='shorter %s n=%s'%(legend_text, nn),
color=color2[int(nn)-1],
symbol='circle'
)
)
layout = create_layout(plot_title,
xaxis=axis_x, yaxis=axis_y1, width=1000,
legend={'x': 1.1,'xanchor': 'left', 'y': 1 }
)
traces = traces_random + traces_shorter
finish(directory, output_name, traces, layout)
def generate_time_vs_error(directory, output_name=None):
filesnames = search_files(directory, "*results*.csv")
all_results = parse_several_error_time(filesnames)
#random_results = all_results['random'] #[n_items] , {N: [times]}, {N: [distances]}
#bassic_results = all_results['bassic']
tree_results = all_results['tree']
axis_x, axis_y1, _ = create_axis_group("Time (s)",
"Error (%1)",
y1_args={'type': 'log'},
x_args={
'type':
'log',
'range':
[np.log10(0.02),
np.log10(150)]
})
densities = tree_results[0]
times = tree_results[1]
distances = tree_results[2]
traces_tree = []
for index in range(len(densities)):
density = densities[index]
x_values = [tim[index] for ene, tim in times.iteritems()]
y_values = [err[index] for ene, err in distances.iteritems()]
traces_tree.append(
create_scatter(
x=x_values,
y=y_values,
name="tree d=%d" % (density),
mode="lines+markers+text",
text=['n=%s' % (nn + 1) for nn in range(len(x_values))],
textposition='top center',
textfont=dict(color=GREENS[index]),
color=GREENS[index],
symbol='star',
yaxis='y1'))
traces = traces_tree
layout = create_layout(
"Compare error and time for tree searcher as n increases for several densities",
xaxis=axis_x,
yaxis=axis_y1,
width=1000,
legend={
'x': 1.1,
'xanchor': 'left',
'y': 1
})
#traces = traces_random + traces_bassic + traces_tree
output_name = "new_only_tree"
finish(directory, output_name, traces, layout)
def generate_basic_example(filepath,
output_name=None,
rate=False,
exp_time=None,
exp_acc=None):
results_data = parse_file(filepath)
x_values = results_data['x']
times = results_data['y1']
distances = results_data['y2']
axis_x, axis_y1, axis_y2 = create_axis_group(
"Algorithm's depth (n)",
"Tolerance (%1)",
BLUES[0],
"Time (s)",
GREENS[0],
x_args=dict(domain=[0, 0.8]) if rate else {},
y2_args={'anchor': 'x'})
trace_accurance = create_scatter(x=x_values,
y=distances,
mode='markers+lines',
name='tolerance',
color=BLUES[0])
trace_time = create_scatter(x=x_values,
y=times,
mode='markers+lines',
name='time',
color=GREENS[0],
yaxis='y2')
traces = [trace_accurance, trace_time]
y3axis = {}
if rate:
y3axis['yaxis3'] = create_axis("Rate (s^-1)",
side='right',
overlaying='y',
zeroline=False,
color=REDS[0],
anchor='free',
position=0.9)
traces.append(get_rate_trace(x_values, times, distances))
if exp_time:
trace_e = create_scatter(yaxis='y2', **exp_time)
traces.append(trace_e)
if exp_acc:
trace_e = create_scatter(yaxis='y1', **exp_acc)
traces.append(trace_e)
layout = create_layout("Results bassic finder for largest group",
xaxis=axis_x,
yaxis=axis_y1,
yaxis2=axis_y2,
legend={
'x': 1,
'xanchor': 'left',
'y': 1
},
**y3axis)
finish(filepath, output_name, traces, layout)
]
teo_y = [3**l for l in x_values]
x = np.linspace(1, 16, num=20)
f = create_pow_function(x, 1, 3)
axis_x, axis_y1, _ = create_axis_group("Operator max lenght",
"Group size (#items)",
y1_args={'type': 'log'},
x_args=dict(range=[0, 17],
dtick=1,
tickangle=0))
trace_teo = get_rate_trace(x_values, teo_y)
trace_emp = create_scatter(x=x_values,
y=emp_y,
mode='markers+lines',
name='empirical',
color=GREENS[0])
traces = [trace_teo, trace_emp]
layout = create_layout("Theorical and empirical group size",
xaxis=axis_x,
yaxis=axis_y1,
legend={
'x': 1,
'xanchor': 'left',
'y': 1
})
finish('filepath', 'group_len', traces, layout)
