from numpy.random import randint as randi
print("Enter size of desired list")
size_of_list = int(input())
orig_list = randi(0,high=999,size=size_of_list)
print(orig_list)
''' From CLRS Introduction to Algorithms:
quicksort:
divide:
partition array into 2 subarrays such that A[p..q-1] and A[q+1..r]
such that all elements in A[p..q-1] are smaller than A[q] and all
elements of A[q+1..r] are larger than A[q]
conquer:
sort the two subarrays by recursively calling quicksort
combine:
subarrays are sorted so A[p..r] should be sorted
'''
def quick_sort(sorting_list:list) -> list:
if len(sorting_list) <= 1: #base case
return sorting_list
else: #divide+conquer+combine all inline
return(quick_sort([element for element in sorting_list[1:] if element <=sorting_list[0]])
+ [sorting_list[0]] + quick_sort([element for element in sorting_list[1:] if element > sorting_list[0]])
)
sorted_list = quick_sort(orig_list)
print(sorted_list)
def add_known_complex(x, fs, easy=False, window_duration=3, group=None):
if easy:
M = np.round(fs * window_duration)
g = gaussian(M, std=50)
t = np.arange(0, g.shape[0], 1) / fs
z = np.sin(2 * 3.14 * 10 * t) * 2
w = g * z
for i in range(0, x.shape[1]):
x[0, i, 100:100 + M] = x[0, i, 100:100 + M] + w
else:
if group == None:
for i in range(0, x.shape[1]):
# Sine complex
M = np.round(fs * window_duration)
M = M + randi(-M, M) // 2
g = gaussian(M, std=50)
t = np.arange(0, g.shape[0], 1) / fs
amplitude = randn(0, .5) + 8
frequency = 10 + randn(0, 2)
z = np.sin(2 * np.pi * frequency * t) * amplitude
w = g * z
chan = randi(0, 2)
start_index = randi(0, x.shape[2] - M)
x[chan, i, start_index:start_index +
M] = x[chan, i, start_index:start_index + M] + w
# Saw-tooth complex
M = np.round(fs * window_duration)
M = M + randi(-M, M) // 2
g = gaussian(M, std=100)
t = np.arange(0, g.shape[0], 1) / fs
amplitude = randn(0, .5) + 6
frequency = 2 + randn(0, .1)
z = sawtooth(2 * np.pi * frequency * t) * amplitude
w = g * z
chan = randi(0, 2)
start_index = randi(0, x.shape[2] - M)
x[chan, i, start_index:start_index +
M] = x[chan, i, start_index:start_index + M] + w
else:
for i in range(0, x.shape[1]):
if group == 0:
# Sine complex
M = np.round(fs * window_duration)
M = int(M + randi(-M, M) // 2)
g = gaussian(M, std=50)
t = np.arange(0, g.shape[0], 1) / fs
amplitude = randn(0, .5) + 8
frequency = 10 + randn(0, 2)
z = np.sin(2 * np.pi * frequency * t) * amplitude
w = g * z
chan = randi(0, 2)
start_index = randi(0, x.shape[2] - M)
x[chan, i, start_index:start_index +
M] = x[chan, i, start_index:start_index + M] + w
if group == 1:
# Saw-tooth complex
M = np.round(fs * window_duration)
M = int(M + randi(-M, M) // 2)
g = gaussian(M, std=100)
t = np.arange(0, g.shape[0], 1) / fs
amplitude = randn(0, .5) + 6
frequency = 2 + randn(0, .1)
z = sawtooth(2 * np.pi * frequency * t) * amplitude
w = g * z
chan = randi(0, 2)
start_index = randi(0, x.shape[2] - M)
x[chan, i, start_index:start_index +
M] = x[chan, i, start_index:start_index + M] + w
return x
annById = defaultdict(list)
for ann in boxAnn['annotations']:
annById[ann['image_id']].append(ann)
i = 0
impath = os.path.join(basepath, 'images/train2014',
boxAnn['images'][i]['file_name'])
img = cv2.imread(impath, 1)
imgBox = img
for ann in annById[boxAnn['images'][i]['id']]:
imgBox = cv2.rectangle(imgBox,
(int(ann['bbox'][0]), int(ann['bbox'][1])),
(int(ann['bbox'][0] + ann['bbox'][2]),
int(ann['bbox'][1] + ann['bbox'][3])),
(randi(0, 256), randi(0, 256), randi(0, 256)),
3)
cv2.imshow('imBox', imgBox)
if cv2.waitKey(0) & 0xFF == ord('q'):
cv2.destroyAllWindows()
annKeys = set(annById.keys())
for i, img in enumerate(dataset['images']):
#dataset['images'][i]['bboxAnn'] = []
if img['split'] != 'train':
sz = (imgByimgId[img['cocoid']]['width'],
imgByimgId[img['cocoid']]['height'])
dataset['images'][i]['imgSize'] = sz
if img['cocoid'] in annKeys:
for ann in annById[img['cocoid']]:
def run_trials(k=3, nMC=10, num_processes=mp.cpu_count(), filename=None):
tic = time.time()
if (filename == None):
date_str = datetime.datetime.now().strftime("%b-%d-%y-%I:%M%p")
filename = "OUTPUT_n_%s_nMc_%s_%s" % (2**k, nMC, date_str)
np.random.seed(12181990)
n = 2**k
# to sweep over m and r and generate a whole bunch of trials of
# each and call solver on those while logging output
p = Pool(processes=num_processes)
for ensemble in xrange(len(common.ENSEMBLE_NAMES)):
for target in [2, 3]: #range(len(common.TARGET_NAMES)):
#40 iterations of this loop it seems.
real_target = (target == common.TARGET_TYPES.RPSD
or target == common.TARGET_TYPES.RSYM)
complex_measurement = (common.ENSEMBLE_NAMES[ensemble][0] == 'C')
if not (real_target and complex_measurement):
# Choose random ranks
if n < 32:
l = 1
else:
l = 6
rs = range(1, 6, 1) + randi(l, n / 4, 5).tolist() + randi(
n / 4 + 1, n / 2, 5).tolist() + randi(n / 2 + 1, n,
5).tolist()
#print rs
# # First, calculate upper bound on number of measurements
# is_entry = (ensemble == common.ENSEMBLE_TYPES.ENTRY )
# is_real_dirac = (common.ENSEMBLE_NAMES[ensemble][1:] == "DIRAC" and real_target)
# if is_entry or is_real_dirac:
# UB = n ** 2 / 2 + n/2
# else:
# UB = n ** 2
UB = n**2 / 2 + n / 2
ms_foreach_r = [[] for _ in rs]
for (ms, r) in zip(ms_foreach_r, rs):
rho = float(r) / n
# Compute lower bound on number of measurements in a given test
LB = max(np.rint((rho - rho**2 / 2) * n**2), 1)
ms.extend(randi(LB, UB, 10).tolist())
for (ms, r) in zip(ms_foreach_r, rs):
for m in ms:
for trial in xrange(nMC):
p.apply_async(compute_task,
args=(
n,
nMC,
filename,
m,
r,
target,
ensemble,
))
#Toggle when debugging.
#compute_task(n,nMC,filename,m,r,target,ensemble)
toc = time.time() - tic
print "Queued all jobs \t %f" % (toc)
sys.stdout.write('Waiting...')
sys.stdout.flush()
tic = time.time()
p.close()
p.join()
toc = time.time() - tic
print " done."
print "Spent %f seconds waiting for processes" % (toc)
# smin = 30
# smax = 40
simulatedPair_no = 10000 # set how many simulated pairs are negated
simulatedPair_no = 100
test_static = zeros(simulatedPair_no)
pvalue = zeros(simulatedPair_no)
alpha = 0.05
print(
' Exp-#, X1-Size X2-Size, Mean1 Mean2 Var1 Var2 t-statistic p-value\n'
)
for i in range(0, simulatedPair_no):
sample_size = randi(smin, smax, 2) # set the sample sizes
x1 = rand(sample_size[1 - 1]) # uniform distribution
x2 = rand(sample_size[2 - 1]) # uniform distribution
t, p = ttest_ev(x1, x2)
# [h,p2,ci,stats] = ttest2(x1, x2)
t, p = ttest_uev(x1, x2)
# [h,p2,ci,stats] = ttest2(x1, x2, 'Vartype','unequal')
test_static[i] = t
pvalue[i] = p
# for output purpose
o1 = len(x1)
o2 = len(x2)
mean1 = mean(x1)