copyfile(h5_og_path, h5_path)
M = 125
Ns = int(5e8)
with h5py.File(h5_path, mode='r+') as h5_f:
h5_grp = h5_f['Measurement_000/Channel_000']
f = usid.hdf_utils.get_attr(h5_grp, 'excitation_frequency_[Hz]')
V0 = usid.hdf_utils.get_attr(h5_grp, 'excitation_amplitude_[V]')
h5_resh = h5_f[
'Measurement_000/Channel_000/Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
# initialize the Process with the desired parameters
abi = AdaptiveBayesianInference(h5_resh, f=f, V0=V0, Ns=Ns, M=M)
# record the time it takes to compute one random pixel
startTime = time.time()
pixResultBoi = abi.test()
totalTime = time.time() - startTime
# delete the copy of the dataset we created
os.remove(h5_path)
# record the time in a file pixel by pixel so we don't have to wait for all
# the pixels to finish (in case something crashes haha)
outputFile = open("timings.txt", "a")
outputFile.write("{}\n".format(totalTime))
outputFile.close()
# This makes graphs for the subset results to see if they worked.
import h5py
import io
import time
from bayesian_inference import AdaptiveBayesianInference
from matplotlib import pyplot as plt
import pyUSID as pyUSID
import numpy as np
import pyUSID as usid
h5_path = r"C:\Users\Administrator\Documents\29TSummer2019\subsetFile1565658168.7634015.h5"
with h5py.File(h5_path, mode="r+") as h5_f:
h5_main = h5_f["subsetBoi/Measured Current"]
h5_results = h5_f["subsetBoi/Measured Current-Adaptive_Bayesian_000"]
abi = AdaptiveBayesianInference(h5_main)
for i in range(h5_main.shape[0]):
figBoi = abi.plotPixel(i, h5_results_grp=h5_results)
figBoi.set_size_inches(30, 10)
figBoi.show()
input("Press <Enter> for next figure...")
"nA",
None,
None,
dtype=np.float64,
h5_pos_inds=h5_pos_inds,
h5_pos_vals=h5_pos_vals,
h5_spec_inds=h5_spec_inds,
h5_spec_vals=h5_spec_vals)
h5_subset[()] = h5_resh[()][pixelInds, :]
#print("check if main returns: {}".format(usid.hdf_utils.check_if_main(h5_subset)))
# remove the temporarily created full copy
os.remove(h5_path)
# create new Process class and run compute
abi = AdaptiveBayesianInference(h5_subset, f=f, V0=V0, Ns=Ns, M=M)
print("starting compute")
startTime = time.time()
h5_bayes_group = abi.compute()
totalTime = time.time() - startTime
print("compute ended")
# close the subset file
sub_f.close()
os.remove(sub_f_path)
# record the timing results
outputFile = open("parallelTimings.txt", "a")
outputFile.write("{}\n".format(totalTime))
outputFile.close()
h5_grp = h5_f['Measurement_000/Channel_000']
f = usid.hdf_utils.get_attr(h5_grp, 'excitation_frequency_[Hz]')
V0 = usid.hdf_utils.get_attr(h5_grp, 'excitation_amplitude_[V]')
h5_resh = h5_f[
'Measurement_000/Channel_000/Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
for i in range(len(NsVals)):
for j in range(len(mVals)):
print("Starting run with Ns = {} and M = {}".format(
NsVals[i], mVals[j]))
# initialize the Process with the desired parameters
abi = AdaptiveBayesianInference(h5_resh,
f=f,
V0=V0,
Ns=NsVals[i],
M=mVals[j])
# record the time it takes to compute one random pixel
startTime = time.time()
pix_ind, figBoi = abi.test(pix_ind=43763)
totalTime = time.time() - startTime
# store the time for graphing later
timingResults[i][j] = totalTime
# record the time in a file pixel by pixel so we don't have to wait for all
# the pixels to finish (in case something crashes haha)
outputFile = open("parameterTests/timings.txt", "a")
outputFile.write(
import h5py
#from mpi4py import MPI
from bayesian_inference import AdaptiveBayesianInference
h5_path = 'pzt_nanocap_6_split_bayesian_compensation_R_correction.h5'
with h5py.File(h5_path, mode='r+') as h5_f:
h5_grp = h5_f['Measurement_000/Channel_000']
h5_resh = h5_grp[
'Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
abi = AdaptiveBayesianInference(h5_resh, verbose=True)
h5_bayes_grp = abi.compute()
import h5py
import os
import matplotlib.pyplot as plt
#from mpi4py import MPI
from bayesian_inference import AdaptiveBayesianInference
h5_path = 'pzt_nanocap_6_split_bayesian_compensation_R_correction.h5'
with h5py.File(h5_path, mode='r+') as h5_f:
h5_grp = h5_f['Measurement_000/Channel_000']
h5_resh = h5_grp[
'Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
abi = AdaptiveBayesianInference(h5_resh, Ns=int(10), verbose=True)
fig1, fig2 = abi.test()
#plt.savefig("result0.png")
#plt.savefig("result1.png")
with h5py.File(h5_path, mode='r+') as h5_f:
h5_grp = h5_f['Measurement_000/Channel_000']
f = usid.hdf_utils.get_attr(h5_grp, 'excitation_frequency_[Hz]')
V0 = usid.hdf_utils.get_attr(h5_grp, 'excitation_amplitude_[V]')
h5_resh = h5_f['Measurement_000/Channel_000/Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
for i in range(len(timingNames)):
print("Starting running stuff in {}".format(timingNames[i]))
os.chdir(timingNames[i])
from bayesian_inference import AdaptiveBayesianInference
# initialize the Process with the desired parameters
abi = AdaptiveBayesianInference(h5_resh, f=f, V0=V0, Ns=timingNs[i], M=timingM[i])
for j in range(10):
# record the time it takes to compute one random pixel
startTime = time.time()
pixResultBoi = abi.test()
totalTime = time.time() - startTime
# record the time in a file pixel by pixel so we don't have to wait for all
# the pixels to finish (in case something crashes haha)
outputFile = open("VMtimings.txt", "a")
outputFile.write("{}\n".format(totalTime))
outputFile.close()
del AdaptiveBayesianInference
print("Finished running stuff in {}".format(timingNames[i]))
from bayesian_inference import AdaptiveBayesianInference
from matplotlib import pyplot as plt
import pyUSID as usid
import numpy as np
h5_path = r"C:\Users\Administrator\Dropbox\GIv Bayesian June 2019\pzt_nanocap_6_split_bayesian_compensation_R_correction.h5"
with h5py.File(h5_path, mode='r+') as h5_f:
h5_grp = h5_f['Measurement_000/Channel_000']
f = usid.hdf_utils.get_attr(h5_grp, 'excitation_frequency_[Hz]')
V0 = usid.hdf_utils.get_attr(h5_grp, 'excitation_amplitude_[V]')
h5_resh = h5_f['Measurement_000/Channel_000/Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
abi = AdaptiveBayesianInference(h5_resh, f=f, V0=V0, Ns=int(1e8))
figFor, figRev = abi.test(pix_ind=27935)
figFor.set_size_inches(25, 10)
figRev.set_size_inches(25, 10)
figFor.savefig("27935forward1e8.png")
figRev.savefig("27935reverse1e8.png")
#figFor.show()
#figRev.show()
#breakpoint()
h5_path = r"C:\Users\Administrator\Dropbox\GIv Bayesian June 2019\pzt_nanocap_6_split_bayesian_compensation_R_correction (Alvin Tan's conflicted copy 2019-06-25).h5"
with h5py.File(h5_path, mode='r+') as h5_f:
h5_grp = h5_f['Measurement_000/Channel_000']
f = usid.hdf_utils.get_attr(h5_grp, 'excitation_frequency_[Hz]')
V0 = usid.hdf_utils.get_attr(h5_grp, 'excitation_amplitude_[V]')
h5_resh = h5_f[
'Measurement_000/Channel_000/Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
Ns = int(1e7)
M = 24
abi = AdaptiveBayesianInference(h5_resh, f=f, V0=V0, Ns=Ns, M=M)
pos_in_batch = [0, 138, 3994, 27935]
start_time = time.time()
# lol bad coding practices for the win
abi._create_results_datasets()
abi._unit_computation(pos_in_batch=pos_in_batch)
abi._write_results_chunk(pos_in_batch=pos_in_batch)
print(
"Processing {} pixels with Ns = {} and M = {} took {} seconds".format(
len(pos_in_batch), Ns, M,
time.time() - start_time))
h5_resh = h5_grp[
'Raw_Data-FFT_Filtering_000/Filtered_Data-Reshape_000/Reshaped_Data']
# Each 'X' axes has 256 images, so [0,1] have 512 images
myDict = {}
#myDict['X'] = [0,1]
'''
# Pseudorandomly select two rows to process
randNum1 = random.randrange(266)
randNum2 = random.randrange(266)
myDict['X'] = [randNum1, randNum2]
print("Running compute on rows {} and {}".format(randNum1, randNum2))
'''
# Creates object from original data
abi = AdaptiveBayesianInference(h5_resh)
'''
#slice data
subset = abi.h5_main.slice_to_dataset(slice_dict=myDict)
#create object from sliced data
abi_subset = AdaptiveBayesianInference(subset, f=f, V0=V0, Ns=Ns, M=M)
'''
#compute
startTime = time.time()
#h5_bayes_group = abi_subset.test()
h5_bayes_group = abi.compute()
totalTime = time.time() - startTime
#copy result group to new h5 file
with h5py.File('bayesian_results{}.h5'.format(time.time()), 'a') as f:
f.create_group("results0")