Python* API Reference for Intel® Data Analytics Acceleration Library 2019 Update 4

svd_dense_distr.py

Deprecation Notice: With the introduction of daal4py, a package that supersedes PyDAAL, Intel is deprecating PyDAAL and will discontinue support starting with Intel® DAAL 2021 and Intel® Distribution for Python 2021. Until then Intel will continue to provide compatible pyDAAL pip and conda packages for newer releases of Intel DAAL and make it available in open source. However, Intel will not add the new features of Intel DAAL to pyDAAL. Intel recommends developers switch to and use daal4py.

Note: To find daal4py examples, refer to daal4py documentation or browse github repository.

1 # file: svd_dense_distr.py
2 #===============================================================================
3 # Copyright 2014-2019 Intel Corporation.
4 #
5 # This software and the related documents are Intel copyrighted materials, and
6 # your use of them is governed by the express license under which they were
7 # provided to you (License). Unless the License provides otherwise, you may not
8 # use, modify, copy, publish, distribute, disclose or transmit this software or
9 # the related documents without Intel's prior written permission.
10 #
11 # This software and the related documents are provided as is, with no express
12 # or implied warranties, other than those that are expressly stated in the
13 # License.
14 #===============================================================================
15 
16 ## <a name="DAAL-EXAMPLE-PY-SVD_DISTRIBUTED"></a>
17 ## \example svd_dense_distr.py
18 
19 import os
20 import sys
21 import numpy as np
22 
23 from daal import step1Local, step2Master, step3Local
24 from daal.algorithms import svd
25 from daal.data_management import FileDataSource, DataSourceIface
26 
27 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
28 if utils_folder not in sys.path:
29  sys.path.insert(0, utils_folder)
30 from utils import printNumericTable
31 
32 DAAL_PREFIX = os.path.join('..', 'data')
33 
34 # Input data set parameters
35 nBlocks = 4
36 
37 datasetFileNames = [
38  os.path.join(DAAL_PREFIX, 'distributed', 'svd_1.csv'),
39  os.path.join(DAAL_PREFIX, 'distributed', 'svd_2.csv'),
40  os.path.join(DAAL_PREFIX, 'distributed', 'svd_3.csv'),
41  os.path.join(DAAL_PREFIX, 'distributed', 'svd_4.csv')
42 ]
43 
44 dataFromStep1ForStep2 = [0] * nBlocks
45 dataFromStep1ForStep3 = [0] * nBlocks
46 dataFromStep2ForStep3 = [0] * nBlocks
47 Sigma = None
48 V = None
49 Ui = [0] * nBlocks
50 
51 
52 def computestep1Local(block):
53  global dataFromStep1ForStep2, dataFromStep1ForStep3
54 
55  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
56  dataSource = FileDataSource(
57  datasetFileNames[block],
58  DataSourceIface.doAllocateNumericTable,
59  DataSourceIface.doDictionaryFromContext
60  )
61 
62  # Retrieve the input data
63  dataSource.loadDataBlock()
64 
65  # Create an algorithm to compute SVD on the local node
66  algorithm = svd.Distributed(step1Local,fptype=np.float64)
67 
68  algorithm.input.set(svd.data, dataSource.getNumericTable())
69 
70  # Compute SVD and get OnlinePartialResult class from daal.algorithms.svd
71  pres = algorithm.compute()
72 
73  dataFromStep1ForStep2[block] = pres.get(svd.outputOfStep1ForStep2)
74  dataFromStep1ForStep3[block] = pres.get(svd.outputOfStep1ForStep3)
75 
76 
77 def computeOnMasterNode():
78  global Sigma, V, dataFromStep2ForStep3
79 
80  # Create an algorithm to compute SVD on the master node
81  algorithm = svd.Distributed(step2Master,fptype=np.float64)
82 
83  for i in range(nBlocks):
84  algorithm.input.add(svd.inputOfStep2FromStep1, i, dataFromStep1ForStep2[i])
85 
86  # Compute SVD and get DistributedPartialResult class from daal.algorithms.svd
87  pres = algorithm.compute()
88 
89  for i in range(nBlocks):
90  dataFromStep2ForStep3[i] = pres.getCollection(svd.outputOfStep2ForStep3, i)
91 
92  res = algorithm.finalizeCompute()
93 
94  Sigma = res.get(svd.singularValues)
95  V = res.get(svd.rightSingularMatrix)
96 
97 
98 def finalizeComputestep1Local(block):
99  global Ui
100 
101  # Create an algorithm to compute SVD on the master node
102  algorithm = svd.Distributed(step3Local,fptype=np.float64)
103 
104  algorithm.input.set(svd.inputOfStep3FromStep1, dataFromStep1ForStep3[block])
105  algorithm.input.set(svd.inputOfStep3FromStep2, dataFromStep2ForStep3[block])
106 
107  # Compute SVD
108  algorithm.compute()
109  res = algorithm.finalizeCompute()
110 
111  Ui[block] = res.get(svd.leftSingularMatrix)
112 
113 if __name__ == "__main__":
114 
115  for i in range(nBlocks):
116  computestep1Local(i)
117 
118  computeOnMasterNode()
119 
120  for i in range(nBlocks):
121  finalizeComputestep1Local(i)
122 
123  # Print the results
124  printNumericTable(Sigma, "Singular values:")
125  printNumericTable(V, "Right orthogonal matrix V:")
126  printNumericTable(Ui[0], "Part of left orthogonal matrix U from 1st node:", 10)

For more complete information about compiler optimizations, see our Optimization Notice.