def test_commented_header(self):
"Check that names can be retrieved even if the line is commented out."
data = StringIO("""
#gender age weight
M 21 72.100000
F 35 58.330000
M 33 21.99
""")
# The # is part of the first name and should be deleted automatically.
test = iopro.genfromtxt(data, names=True, dtype=None)
ctrl = np.array([('M', 21, 72.1), ('F', 35, 58.33), ('M', 33, 21.99)],
# JNB: changed test because iopro defaults to object string
# instead of fixed length string, and unsigned long int
# instead of int.
dtype=[('gender', 'O'), ('age', 'u8'), ('weight', 'f8')])
#dtype=[('gender', '|S1'), ('age', int), ('weight', float)])
assert_equal(test, ctrl)
# Ditto, but we should get rid of the first element
data = StringIO("""
# gender age weight
M 21 72.100000
F 35 58.330000
M 33 21.99
""")
test = iopro.genfromtxt(data, names=True, dtype=None)
assert_equal(test, ctrl)
def test_usecols_with_named_columns(self):
"Test usecols with named columns"
ctrl = np.array([(1, 3), (4, 6)], dtype=[('a', float), ('c', float)])
data = "1 2 3\n4 5 6"
kwargs = dict(names="a, b, c")
test = iopro.genfromtxt(StringIO(data), usecols=(0, -1), **kwargs)
assert_equal(test, ctrl)
test = iopro.genfromtxt(StringIO(data),
usecols=('a', 'c'), **kwargs)
assert_equal(test, ctrl)
def test_usecols_as_css(self):
"Test giving usecols with a comma-separated string"
data = "1 2 3\n4 5 6"
test = iopro.genfromtxt(StringIO(data),
names="a, b, c", usecols="a, c")
ctrl = np.array([(1, 3), (4, 6)], dtype=[(_, float) for _ in "ac"])
assert_equal(test, ctrl)
def test_with_masked_column_uniform(self):
"Test masked column"
data = StringIO('1 2 3\n4 5 6\n')
test = iopro.genfromtxt(data, dtype=None,
missing_values='2,5', usemask=True)
control = ma.array([[1, 2, 3], [4, 5, 6]], mask=[[0, 1, 0], [0, 1, 0]])
assert_equal(test, control)
def test_gft_using_generator(self):
def count():
for i in range(10):
yield "%d" % i
res = iopro.genfromtxt(count())
assert_array_equal(res, np.arange(10))
def test_user_filling_values(self):
"Test with missing and filling values"
ctrl = np.array([(0, 3), (4, -999)], dtype=[('a', int), ('b', int)])
data = "N/A, 2, 3\n4, ,???"
kwargs = dict(delimiter=",",
dtype=int,
names="a,b,c",
missing_values={0:"N/A", 'b':" ", 2:"???"},
filling_values={0:0, 'b':0, 2:-999})
test = iopro.genfromtxt(StringIO(data), **kwargs)
ctrl = np.array([(0, 2, 3), (4, 0, -999)],
dtype=[(_, int) for _ in "abc"])
assert_equal(test, ctrl)
test = iopro.genfromtxt(StringIO(data), usecols=(0, -1), **kwargs)
ctrl = np.array([(0, 3), (4, -999)], dtype=[(_, int) for _ in "ac"])
assert_equal(test, ctrl)
def test_userconverters_with_explicit_dtype(self):
"Test user_converters w/ explicit (standard) dtype"
data = StringIO('skip,skip,2001-01-01,1.0,skip')
test = iopro.genfromtxt(data, delimiter=",", names=None, dtype=float,
usecols=(2, 3), converters={2: bytes})
control = np.array([('2001-01-01', 1.)],
dtype=[('', '|S10'), ('', float)])
assert_equal(test, control)
def test_tricky_converter_bug1666(self):
"Test some corner case"
assert_equal(True, False)
s = StringIO('q1,2\nq3,4')
cnv = lambda s:float(s[1:])
test = iopro.genfromtxt(s, delimiter=',', converters={0:cnv})
control = np.array([[1., 2.], [3., 4.]])
assert_equal(test, control)
def test_names_auto_completion(self):
"Make sure that names are properly completed"
data = "1 2 3\n 4 5 6"
test = iopro.genfromtxt(StringIO(data),
dtype=(int, float, int), names="a")
ctrl = np.array([(1, 2, 3), (4, 5, 6)],
dtype=[('a', int), ('f1', float), ('f2', int)])
assert_equal(test, ctrl)
def test_missing_with_tabs(self):
"Test w/ a delimiter tab"
txt = "1\t2\t3\n\t2\t\n1\t\t3"
test = iopro.genfromtxt(StringIO(txt), delimiter="\t",
usemask=True,)
ctrl_d = np.array([(1, 2, 3), (np.nan, 2, np.nan), (1, np.nan, 3)],)
ctrl_m = np.array([(0, 0, 0), (1, 0, 1), (0, 1, 0)], dtype=bool)
assert_equal(test.data, ctrl_d)
assert_equal(test.mask, ctrl_m)
def test_with_masked_column_various(self):
"Test masked column"
data = StringIO('True 2 3\nFalse 5 6\n')
test = iopro.genfromtxt(data, dtype=None,
missing_values='2,5', usemask=True)
control = ma.array([(1, 2, 3), (0, 5, 6)],
mask=[(0, 1, 0), (0, 1, 0)],
dtype=[('f0', bool), ('f1', bool), ('f2', int)])
assert_equal(test, control)
def test_names_with_usecols_bug1636(self):
"Make sure we pick up the right names w/ usecols"
data = "A,B,C,D,E\n0,1,2,3,4\n0,1,2,3,4\n0,1,2,3,4"
ctrl_names = ("A", "C", "E")
test = iopro.genfromtxt(StringIO(data),
dtype=(int, int, int), delimiter=",",
usecols=(0, 2, 4), names=True)
assert_equal(test.dtype.names, ctrl_names)
#
test = iopro.genfromtxt(StringIO(data),
dtype=(int, int, int), delimiter=",",
usecols=("A", "C", "E"), names=True)
assert_equal(test.dtype.names, ctrl_names)
#
test = iopro.genfromtxt(StringIO(data),
dtype=int, delimiter=",",
usecols=("A", "C", "E"), names=True)
assert_equal(test.dtype.names, ctrl_names)
def test_skip_footer(self):
data = ["# %i" % i for i in range(1, 6)]
data.append("A, B, C")
data.extend(["%i,%3.1f,%03s" % (i, i, i) for i in range(51)])
data[-1] = "99,99"
kwargs = dict(delimiter=",", names=True, skip_header=5, skip_footer=10)
test = iopro.genfromtxt(StringIO("\n".join(data)), **kwargs)
ctrl = np.array([("%f" % i, "%f" % i, "%f" % i) for i in range(41)],
dtype=[(_, float) for _ in "ABC"])
assert_equal(test, ctrl)
def test_empty_file(self):
"Test that an empty file raises the proper warning."
warn_ctx = WarningManager()
warn_ctx.__enter__()
try:
warnings.filterwarnings("ignore", message="genfromtxt: Empty input file:")
data = StringIO()
test = iopro.genfromtxt(data)
assert_equal(test, np.array([]))
finally:
warn_ctx.__exit__()
def test_skip_footer_with_invalid(self):
warn_ctx = WarningManager()
warn_ctx.__enter__()
try:
basestr = '1 1\n2 2\n3 3\n4 4\n5 \n6 \n7 \n'
warnings.filterwarnings("ignore")
# Footer too small to get rid of all invalid values
assert_raises(ValueError, iopro.genfromtxt,
StringIO(basestr), skip_footer=1)
a = iopro.genfromtxt(StringIO(basestr), skip_footer=1, invalid_raise=False)
assert_equal(a, np.array([[1., 1.], [2., 2.], [3., 3.], [4., 4.]]))
#
a = iopro.genfromtxt(StringIO(basestr), skip_footer=3)
assert_equal(a, np.array([[1., 1.], [2., 2.], [3., 3.], [4., 4.]]))
#
basestr = '1 1\n2 \n3 3\n4 4\n5 \n6 6\n7 7\n'
a = iopro.genfromtxt(StringIO(basestr), skip_footer=1, invalid_raise=False)
assert_equal(a, np.array([[1., 1.], [3., 3.], [4., 4.], [6., 6.]]))
a = iopro.genfromtxt(StringIO(basestr), skip_footer=3, invalid_raise=False)
assert_equal(a, np.array([[1., 1.], [3., 3.], [4., 4.]]))
finally:
warn_ctx.__exit__()
def test_replace_space(self):
"Test the 'replace_space' option"
txt = "A.A, B (B), C:C\n1, 2, 3.14"
# Test default: replace ' ' by '_' and delete non-alphanum chars
test = iopro.genfromtxt(StringIO(txt),
delimiter=",", names=True, dtype=None)
ctrl_dtype = [("AA", int), ("B_B", int), ("CC", float)]
ctrl = np.array((1, 2, 3.14), dtype=ctrl_dtype)
assert_equal(test, ctrl)
# Test: no replace, no delete
test = iopro.genfromtxt(StringIO(txt),
delimiter=",", names=True, dtype=None,
replace_space='', deletechars='')
ctrl_dtype = [("A.A", int), ("B (B)", int), ("C:C", float)]
ctrl = np.array((1, 2, 3.14), dtype=ctrl_dtype)
assert_equal(test, ctrl)
# Test: no delete (spaces are replaced by _)
test = iopro.genfromtxt(StringIO(txt),
delimiter=",", names=True, dtype=None,
deletechars='')
ctrl_dtype = [("A.A", int), ("B_(B)", int), ("C:C", float)]
ctrl = np.array((1, 2, 3.14), dtype=ctrl_dtype)
assert_equal(test, ctrl)
def test_dtype_with_object(self):
"Test using an explicit dtype with an object"
assert_equal(True, False)
from datetime import date
import time
data = """ 1; 2001-01-01
2; 2002-01-31 """
ndtype = [('idx', int), ('code', np.object)]
func = lambda s: strptime(s.strip(), "%Y-%m-%d")
converters = {1: func}
test = iopro.genfromtxt(StringIO(data), delimiter=";", dtype=ndtype,
converters=converters)
control = np.array([(1, datetime(2001, 1, 1)), (2, datetime(2002, 1, 31))],
dtype=ndtype)
assert_equal(test, control)
#
ndtype = [('nest', [('idx', int), ('code', np.object)])]
try:
test = iopro.genfromtxt(StringIO(data), delimiter=";",
dtype=ndtype, converters=converters)
except NotImplementedError:
errmsg = "Nested dtype involving objects should be supported."
raise AssertionError(errmsg)
def test_gft_using_filename(self):
# Test that we can load data from a filename as well as a file object
wanted = np.arange(6).reshape((2,3))
if sys.version_info[0] >= 3:
# python 3k is known to fail for '\r'
linesep = ('\n', '\r\n')
else:
linesep = ('\n', '\r\n', '\r')
for sep in linesep:
data = '0 1 2' + sep + '3 4 5'
f, name = mkstemp()
# We can't use NamedTemporaryFile on windows, because we cannot
# reopen the file.
try:
os.write(f, asbytes(data))
assert_array_equal(iopro.genfromtxt(name), wanted)
finally:
os.close(f)
os.unlink(name)
def test_iopro():
# this is kind of stupid right now because it does a copy,
# but Tight IOPro integration will be a priority...
h = Heap()
s = StringIO(','.join(letters))
data = iopro.genfromtxt(s, dtype='c', delimiter=",")
addr, block = allocate_numpy(h, data.dtype, data.shape)
block[:] = data[:]
assert not block.flags['OWNDATA']
assert block.ctypes.data == addr
assert len(h._arenas) == 1
assert block.nbytes < h._lengths[0]
finalize(h)
def test_iopro():
# this is kind of stupid right now because it does a copy,
# but Tight IOPro integration will be a priority...
h = Heap()
s = StringIO(','.join(letters))
data = iopro.genfromtxt(s, dtype='c', delimiter=",")
addr, block = allocate_numpy(h, data.dtype, data.shape)
block[:] = data[:]
assert not block.flags['OWNDATA']
assert block.ctypes.data == addr
assert len(h._arenas) == 1
assert block.nbytes < h._lengths[0]
finalize(h)
def test_usecols_with_integer(self):
"Test usecols with an integer"
test = iopro.genfromtxt(StringIO("1 2 3\n4 5 6"), usecols=0)
assert_equal(test, np.array([1., 4.]))
import iopro
import numpy as np
a_0p28125 = iopro.genfromtxt('cached_kde.ada.0p28125.csv')
a_0p50 = iopro.genfromtxt('cached_kde.ada.0p5.csv')
N, D = a_0p50.shape
#a = np.hstack((a0[:,0:2], a5[:,2:]))
#np.savetxt("cached_kde.ada.hack1.csv", a, delimiter=" ")
#a = np.hstack((a5[:,0:2], a0[:,2:]))
#np.savetxt("cached_kde.ada.hack2.csv", a, delimiter=" ")
#a = np.hstack((a5[:,0].reshape((N, 1)), a0[:,1].reshape((N,1)), a5[:,2:]))
#np.savetxt("cached_kde.ada.hack3.csv", a, delimiter=" ")
# !!
# mildly adapting c0?
#a = np.hstack((a_0p0[:,0].reshape((N, 1)),
# a_0p50[:,1].reshape((N, 1)),
# a_0p50[:,2].reshape((N, 1)),
# a_0p50[:,3].reshape((N, 1)),
# a_0p50[:,4].reshape((N, 1))))
#np.savetxt("cached_kde.ada.hack4.csv", a, delimiter=" ")
# over-adaptation = bad
#a = np.hstack((a5[:,0].reshape((N, 1)),
# a5[:,1].reshape((N, 1)),
# a10[:,2].reshape((N, 1)),
# a5[:,3].reshape((N, 1)),
# a5[:,4].reshape((N, 1))))
parser.add_argument('--zero_abs_tol', type=float, default=1e-15,
help='whether to bootstrap the data. ')
parser.add_argument('cache_dir', type=str,
help='directory holding the cache files. ')
parser.add_argument('n_components', type=int,
help='number of components. ')
parser.add_argument('alphas', type=str, nargs='*',
help='sensitiviy parameters.')
args = parser.parse_args()
# assemble cached data based on sensitivity list
print '+ assembling cached records. '
start = time.time()
cache_fname = args.cache_dir + '/' + args.alphas[0] + args.cache_suffix
arr = genfromtxt(cache_fname)
if len(args.alphas) == 1:
pass
else:
if len(args.alphas) != args.n_components:
raise RuntimeError('alphas must have either 1 or the same number of '
'arguments as what is specified in n_components. ')
for i in range(1, args.n_components):
curr_cache_fname = args.cache_dir + '/' + args.alphas[i] + args.cache_suffix
curr_arr = genfromtxt(curr_cache_fname)
arr[:,i] = curr_arr[:,i]
N, D = arr.shape
def plot_kde2d_summary(
data2d_fname,kde2d_fname, kde1d_x_fname, kde1d_y_fname,
title=None,
xlabel=None, ylabel=None,
draw_scatter=True,
scatter_color='grey', scatter_marker='o',
scatter_undersample=1.0, scatter_s=20, scatter_alpha=0.5,
contour_nlevels=10,
data1d_x_bins=100,
data1d_y_bins=100,
figsize=(12,12), xlim=None, ylim=None):
# define axes dimensions
left, width = 0.1, 0.65
bottom, height = 0.1, 0.65
bottom_h = left_h = left + width + 0.02
rect_c = [left, bottom, width, height]
rect_x = [left, bottom_h, width, 0.18]
rect_y = [left_h, bottom, 0.2, height]
# create plot area
fig = plt.figure(1, figsize=figsize)
ax_c = plt.axes(rect_c)
ax_x = plt.axes(rect_x)
ax_y = plt.axes(rect_y)
nullfmt = NullFormatter()
ax_x.xaxis.set_major_formatter(nullfmt)
ax_y.yaxis.set_major_formatter(nullfmt)
# draw center contour plot
plot_contour2d(kde2d_fname, ax=ax_c,
nlevels=contour_nlevels, colorbar=False,
xlabel=xlabel, ylabel=ylabel)
# draw center scatter plot
if draw_scatter:
plot_data2d(data2d_fname, ax=ax_c,
color=scatter_color, marker=scatter_marker,
undersample=scatter_undersample, s=scatter_s, alpha=scatter_alpha)
# draw marginal densities
plot_data1d(kde1d_x_fname, ax=ax_x, xlabel=' ')
plot_data1d(kde1d_y_fname, ax=ax_y, xlabel=' ', orientation='horizontal')
# draw marginal histograms
arr = iopro.genfromtxt(data2d_fname)
plot_data1d_histogram(arr[:,0], weights=arr[:,2], bins=data1d_x_bins,
normed=True, ax=ax_x, xlabel=' ');
plot_data1d_histogram(arr[:,1], weights=arr[:,2], bins=data1d_y_bins,
normed=True, ax=ax_y, xlabel=' ', orientation='horizontal');
ax_c.set_xlim(xlim)
ax_c.set_ylim(ylim)
ax_x.set_xlim(ax_c.get_xlim())
ax_y.set_ylim(ax_c.get_ylim())
if title:
fig.suptitle(title, fontsize=20)
return fig, ax_c, ax_x, ax_y
writer = csv.writer(csv_file, delimiter=',', lineterminator='\n')
writer.writerow(['ImageId', 'Label'])
for i, p in enumerate(y_pred):
writer.writerow([i + 1, p])
#------------------------------------------
# MAIN ENTRY POINT
#------------------------------------------
if __name__ == "__main__":
# Misura il tempo per le operazioni principali
start = time.time()
# Fase 1: Training
# Read CSV from Numpy, Link:
# https://docs.scipy.org/doc/numpy/reference/generated/numpy.genfromtxt.html
import iopro
my_data = iopro.genfromtxt('Projects/MINST/train.csv',
delimiter=',',
skip_header=1)
print('Reading time:', time.time() - start)
start = time.time()
ann = ANN(len(my_data[0]), 10, 10)
ann.Learn(my_data)
print('Learning time:', time.time() - start, '- size:', len(my_data))
if __name__ == '__main__':
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('nbag', type=int,
help='Number of trials to bag/fit.')
parser.add_argument('sample_fname', type=str,
help='Path to the data sample cached kde score.')
parser.add_argument('output_fname', type=str,
help='Path to store output.')
parser.add_argument('--obj_scale', type=float, default=1e-8,
help='Scale factor to apply to objective function.')
args = parser.parse_args()
# Read in cached KDE evalutions of the data sample to fit
p_raw = genfromtxt(args.sample_fname)
N, D = p_raw.shape
# Open the file to write results
fout = open(args.output_fname, 'w')
for i in range(args.nbag):
sys.stdout.write('Bag iteration {0}:\n\n'.format(i+1))
sys.stdout.flush()
# Create a bagged sample.
p = p_raw[np.random.choice(N, N)]
# Specify the objective function and its derivatives
def F(x=None, z=None):
import sys
import math
import numpy as np
import cvxopt as cvx
from iopro import genfromtxt
#p = genfromtxt('cached_kde/cached_kde.ada.0p28125.csv')
p = genfromtxt('cached_kde/cached_kde.ada.hack10.csv')
N, D = p.shape
I = np.random.choice(N, N)
p = p[I]
s = 1e-8
# Truth proportions:
# 0.004597168626601867
# 0.009011011183443553
# 0.41817119150694954
# 0.07374540140202024
# 0.49265798277959727
def F(x=None, z=None):
if x is None:
return 0, cvx.matrix(0.2, (D, 1))
x_arr = np.array(x.trans()).reshape(-1)
arg = np.dot(p, x_arr)
if np.min(arg) <= 0.0:
def plot_kde2d_summary(data2d_fname,
kde2d_fname,
kde1d_x_fname,
kde1d_y_fname,
title=None,
xlabel=None,
ylabel=None,
draw_scatter=True,
scatter_color='grey',
scatter_marker='o',
scatter_undersample=1.0,
scatter_s=20,
scatter_alpha=0.5,
contour_nlevels=10,
data1d_x_bins=100,
data1d_y_bins=100,
figsize=(12, 12),
xlim=None,
ylim=None):
# define axes dimensions
left, width = 0.1, 0.65
bottom, height = 0.1, 0.65
bottom_h = left_h = left + width + 0.02
rect_c = [left, bottom, width, height]
rect_x = [left, bottom_h, width, 0.18]
rect_y = [left_h, bottom, 0.2, height]
# create plot area
fig = plt.figure(1, figsize=figsize)
ax_c = plt.axes(rect_c)
ax_x = plt.axes(rect_x)
ax_y = plt.axes(rect_y)
nullfmt = NullFormatter()
ax_x.xaxis.set_major_formatter(nullfmt)
ax_y.yaxis.set_major_formatter(nullfmt)
# draw center contour plot
plot_contour2d(kde2d_fname,
ax=ax_c,
nlevels=contour_nlevels,
colorbar=False,
xlabel=xlabel,
ylabel=ylabel)
# draw center scatter plot
if draw_scatter:
plot_data2d(data2d_fname,
ax=ax_c,
color=scatter_color,
marker=scatter_marker,
undersample=scatter_undersample,
s=scatter_s,
alpha=scatter_alpha)
# draw marginal densities
plot_data1d(kde1d_x_fname, ax=ax_x, xlabel=' ')
plot_data1d(kde1d_y_fname, ax=ax_y, xlabel=' ', orientation='horizontal')
# draw marginal histograms
arr = iopro.genfromtxt(data2d_fname)
plot_data1d_histogram(arr[:, 0],
weights=arr[:, 2],
bins=data1d_x_bins,
normed=True,
ax=ax_x,
xlabel=' ')
plot_data1d_histogram(arr[:, 1],
weights=arr[:, 2],
bins=data1d_y_bins,
normed=True,
ax=ax_y,
xlabel=' ',
orientation='horizontal')
ax_c.set_xlim(xlim)
ax_c.set_ylim(ylim)
ax_x.set_xlim(ax_c.get_xlim())
ax_y.set_ylim(ax_c.get_ylim())
if title:
fig.suptitle(title, fontsize=20)
return fig, ax_c, ax_x, ax_y
import sys
import math
import numpy as np
import cvxopt as cvx
from iopro import genfromtxt
#p = genfromtxt('cached_kde/cached_kde.ada.0p28125.csv')
p = genfromtxt('cached_kde/cached_kde.ada.hack10.csv')
N, D = p.shape
I = np.random.choice(N, N)
p = p[I]
s = 1e-8
# Truth proportions:
# 0.004597168626601867
# 0.009011011183443553
# 0.41817119150694954
# 0.07374540140202024
# 0.49265798277959727
def F(x=None, z=None):
if x is None:
return 0, cvx.matrix(0.2, (D, 1))
x_arr = np.array(x.trans()).reshape(-1)
arg = np.dot(p, x_arr)
if np.min(arg) <= 0.0:
return None
help='directory holding the cache files. ')
parser.add_argument('n_components',
type=int,
help='number of components. ')
parser.add_argument('alphas',
type=str,
nargs='*',
help='sensitiviy parameters.')
args = parser.parse_args()
# assemble cached data based on sensitivity list
print '+ assembling cached records. '
start = time.time()
cache_fname = args.cache_dir + '/' + args.alphas[0] + args.cache_suffix
arr = genfromtxt(cache_fname)
if len(args.alphas) == 1:
pass
else:
if len(args.alphas) != args.n_components:
raise RuntimeError(
'alphas must have either 1 or the same number of '
'arguments as what is specified in n_components. ')
for i in range(1, args.n_components):
curr_cache_fname = args.cache_dir + '/' + args.alphas[
i] + args.cache_suffix
curr_arr = genfromtxt(curr_cache_fname)
arr[:, i] = curr_arr[:, i]
def test_integer_delimiter(self):
"Test using an integer for delimiter"
data = " 1 2 3\n 4 5 67\n890123 4"
test = iopro.genfromtxt(StringIO(data), delimiter=3, dtype=int)
control = np.array([[1, 2, 3], [4, 5, 67], [890, 123, 4]])
assert_equal(test, control)
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('nbag', type=int, help='Number of trials to bag/fit.')
parser.add_argument('sample_fname',
type=str,
help='Path to the data sample cached kde score.')
parser.add_argument('output_fname', type=str, help='Path to store output.')
parser.add_argument('--obj_scale',
type=float,
default=1e-8,
help='Scale factor to apply to objective function.')
args = parser.parse_args()
# Read in cached KDE evalutions of the data sample to fit
p_raw = genfromtxt(args.sample_fname)
N, D = p_raw.shape
# Open the file to write results
fout = open(args.output_fname, 'w')
for i in range(args.nbag):
sys.stdout.write('Bag iteration {0}:\n\n'.format(i + 1))
sys.stdout.flush()
# Create a bagged sample.
p = p_raw[np.random.choice(N, N)]
# Solve
sol = perform_mle(p, args.obj_scale)
import iopro
import numpy as np
a_0p28125 = iopro.genfromtxt('cached_kde.ada.0p28125.csv')
a_0p50 = iopro.genfromtxt('cached_kde.ada.0p5.csv')
N, D = a_0p50.shape
#a = np.hstack((a0[:,0:2], a5[:,2:]))
#np.savetxt("cached_kde.ada.hack1.csv", a, delimiter=" ")
#a = np.hstack((a5[:,0:2], a0[:,2:]))
#np.savetxt("cached_kde.ada.hack2.csv", a, delimiter=" ")
#a = np.hstack((a5[:,0].reshape((N, 1)), a0[:,1].reshape((N,1)), a5[:,2:]))
#np.savetxt("cached_kde.ada.hack3.csv", a, delimiter=" ")
# !!
# mildly adapting c0?
#a = np.hstack((a_0p0[:,0].reshape((N, 1)),
# a_0p50[:,1].reshape((N, 1)),
# a_0p50[:,2].reshape((N, 1)),
# a_0p50[:,3].reshape((N, 1)),
# a_0p50[:,4].reshape((N, 1))))
#np.savetxt("cached_kde.ada.hack4.csv", a, delimiter=" ")
# over-adaptation = bad
#a = np.hstack((a5[:,0].reshape((N, 1)),
# a5[:,1].reshape((N, 1)),
# a10[:,2].reshape((N, 1)),
# a5[:,3].reshape((N, 1)),
# a5[:,4].reshape((N, 1))))
