notebook.community

Edit and run 





In [4]:



    


import tensorflow as tf 

N = 100
matrix1 = tf.random_normal((N,N)) 
matrix2 = tf.random_normal((N,N))
product = tf.matmul(matrix1, matrix2)

sess = tf.Session()
sess.run(product)



    


















    
Out[4]:








array([[ 15.77658367,  -2.07524729,  -2.59064078, ..., -10.11330986,
         16.45804787,  -5.94125605],
       [ -2.97194099,  -1.14456856,   1.36590683, ...,  -1.27831125,
         -0.13424999,   6.05601835],
       [ -0.9950223 ,  -8.15255451,  12.7596817 , ...,   1.71796381,
         13.33861732,  -7.33154106],
       ..., 
       [ -5.44885159,  -0.34687257,   3.1139648 , ...,  18.74840355,
         -1.53867686, -15.98628235],
       [ 22.92699432, -14.36166096,  11.64274025, ...,  -9.20965481,
          0.19976738,  13.28671074],
       [ -8.31401062,   3.67289186,  -6.58151674, ...,   8.36485386,
          4.19419813,   9.93695164]], dtype=float32)

















In [1]:



    


import tensorflow as tf
from tensorflow.python.client import timeline
N = 100

run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()

sess = tf.Session()

matr1 = tf.random_normal((N,N), name = "matrix1")
matr2 = tf.random_normal((N,N),name = "matrix2")
		
prod = tf.matmul(matr1, matr2,name = "product")

sess.run(prod,options=run_options,run_metadata = run_metadata)



logs_path = 'logs'
#writer = tf.summary.FileWriter(logs_path, graph=tf.get_default_graph())


writer = tf.summary.FileWriter(logdir=logs_path,graph=sess.graph)

time_path = 'logs/timeline.json'
tl = timeline.Timeline(run_metadata.step_stats)
ctf = tl.generate_chrome_trace_format()
with open(time_path, 'w') as f:
    f.write(ctf)

writer.add_run_metadata(run_metadata,"mySess")
writer.close()
sess.close()



    











In [5]:



    


import tensorflow as tf 

constScal = tf.constant(20.0)
varScal = tf.Variable(3.0)
varScal = tf.Variable.assign(varScal,constScal+varScal)

sess = tf.Session()
sess.run(tf.global_variables_initializer())

sess.run(varScal)



    


















    
Out[5]:








23.0

















In [1]:



    


import tensorflow as tf
from tensorflow.python.client import timeline
N = 100

run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()

sess = tf.Session()

constScal = tf.constant(20.0, name  = 'constant')
varScal = tf.Variable(3.0, name = 'variable')
varScal = tf.Variable.assign(varScal,constScal+varScal)
#varScal = tf.assign(varScal,constScal+varScal)

sess.run(tf.global_variables_initializer())
sess.run(varScal,options=run_options,run_metadata = run_metadata)



logs_path = 'logs'
#writer = tf.summary.FileWriter(logs_path, graph=tf.get_default_graph())


writer = tf.summary.FileWriter(logdir=logs_path,graph=sess.graph)

time_path = 'logs/timeline.json'
tl = timeline.Timeline(run_metadata.step_stats)
ctf = tl.generate_chrome_trace_format()
with open(time_path, 'w') as f:
    f.write(ctf)

writer.add_run_metadata(run_metadata,"mySess")
writer.close()
sess.close()



    











In [8]:



    


import tensorflow as tf
import numpy

N = 100
matrix = numpy.random.rand(N,N)

placeholder = tf.placeholder(tf.float32)
product = tf.matmul(placeholder, placeholder)

sess = tf.Session()

sess.run(product, feed_dict={placeholder: matrix})



    


















    
Out[8]:








array([[ 21.05047989,  24.69828224,  25.1553936 , ...,  20.99284744,
         22.27706718,  22.63986206],
       [ 21.37840462,  25.7079277 ,  26.34488106, ...,  23.93360519,
         24.47508812,  24.95463371],
       [ 21.96219635,  25.62042427,  26.43340111, ...,  23.33528519,
         24.9746151 ,  23.71904373],
       ..., 
       [ 20.27998543,  23.27656555,  24.41584587, ...,  21.35247421,
         22.94180489,  23.17783356],
       [ 22.97197342,  28.81153488,  27.904356  , ...,  23.02546501,
         27.00261497,  25.77598572],
       [ 18.6430378 ,  22.21147346,  22.73134613, ...,  20.82949448,
         21.8946476 ,  22.23680305]], dtype=float32)

















In [3]:



    


import tensorflow as tf
import numpy
from tensorflow.python.client import timeline
N = 100

run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()

sess = tf.Session()

N = 100
matrix = numpy.random.rand(N,N)

placeholder = tf.placeholder(tf.float32, name = "plchold")
product = tf.matmul(placeholder, placeholder, name = 'product')

sess.run(product, feed_dict={placeholder: matrix},options=run_options,run_metadata = run_metadata)



logs_path = 'logs'
#writer = tf.summary.FileWriter(logs_path, graph=tf.get_default_graph())


writer = tf.summary.FileWriter(logdir=logs_path,graph=sess.graph)

time_path = 'logs/timeline.json'
tl = timeline.Timeline(run_metadata.step_stats)
ctf = tl.generate_chrome_trace_format()
with open(time_path, 'w') as f:
    f.write(ctf)

writer.add_run_metadata(run_metadata,"mySess")
writer.close()
sess.close()



    











In [2]:



    


import tensorflow as tf
import numpy
from tensorflow.python.client import timeline

run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()




enhancement = 10
securbelt = 5000
numColumns = 100
numRows = 20
epoch = -60


frequencies = numpy.random.rand(numColumns)
times = numpy.arange(numColumns) + epoch
weights = numpy.ones(numColumns)
spindowns = numpy.arange(numRows)

numColumns = numColumns + securbelt

freqTF = tf.constant(frequencies, dtype=tf.float32, name= 'freqs')
timesTF = tf.constant(times, dtype=tf.float32, name= 'times')
weightsTF = tf.constant(weights, dtype=tf.float32, name= 'weights')
fdotTF = tf.constant(spindowns, dtype=tf.float32, name= 'sd')


def frequencyHough(frequencies,times, weigths, spindowns):

	def rowTransform(ithSD):
		sdTimed = tf.multiply(spindowns[ithSD], times, name= 'sdXt')

		transform = tf.round(frequencies-sdTimed+securbelt/2, name = 'rounding')
		transform = tf.cast(transform, dtype=tf.int32)

		values = tf.unsorted_segment_sum(weights, transform, numColumns, name = 'binning')
		values = tf.cast(values, dtype=tf.float32)
		return values

	houghLeft = tf.map_fn(rowTransform, tf.range(0, numRows), dtype=tf.float32, parallel_iterations=8, name= 'leftMap')

	houghRight = tf.subtract( houghLeft[:,enhancement:numColumns],houghLeft[:,0:numColumns - enhancement], name='rightMap')
	houghDiff = tf.concat([houghLeft[:,0:enhancement],houghRight],1, name='diffMap')

	houghMap = tf.cumsum(houghDiff, axis = 1, name='hough')
	return houghMap 



image = frequencyHough(freqTF,timesTF, weightsTF, fdotTF)
sess = tf.Session()
sess.run(image,options=run_options,run_metadata = run_metadata)



logs_path = 'logs'
#writer = tf.summary.FileWriter(logs_path, graph=tf.get_default_graph())


writer = tf.summary.FileWriter(logdir=logs_path,graph=sess.graph)

time_path = 'logs/timeline.json'
tl = timeline.Timeline(run_metadata.step_stats)
ctf = tl.generate_chrome_trace_format()
with open(time_path, 'w') as f:
    f.write(ctf)

writer.add_run_metadata(run_metadata,"mySess")
writer.close()
sess.close()



    











In [31]:



    


from matplotlib import pyplot
%matplotlib notebook
pyplot.imshow(alal, aspect = 200)



    


















    






Warning: Cannot change to a different GUI toolkit: notebook. Using qt instead.










    
Out[31]:








<matplotlib.image.AxesImage at 0x7f9a2c033780>

















In [ ]:

Content source: iurilarosa/thesis

Similar notebooks:

notebook.community | gallery | about