Question

例如我们有矩阵（例如我们想要存储numpy数组）并将它存储在HDF5文件中，但是我们想通过在原始矩阵的末尾附加一些行来扩展矩阵（考虑到原始矩阵）可能非常大〜几十Gb而且无法加载到RAM中）

此外，我们希望能够从任何点读取矩阵中的几行（可能称为切片（？）），而无需在RAM中加载整个矩阵。

有人能举例说明如何在python中完成它吗？

更新

我认为另一种选择是numpy.memmap，但似乎没有追加。

This似乎也是一个选项，但它使用原始二进制数据操作，但我希望能够访问matrix.Also我不知道在这种情况下如何附加。

Answer 1

如果您要使用HDF5文件，那么我是否可以建议使用其中一个可用的库，例如Pytables。我在这里发布并简化了他们的教程：http://pytables.github.io/usersguide/tutorials.html

from tables import *

# Define a user record to characterize some kind of particles
class Particle(IsDescription):
    name      = StringCol(16)   # 16-character String
    idnumber  = Int64Col()      # Signed 64-bit integer
    ADCcount  = UInt16Col()     # Unsigned short integer
    TDCcount  = UInt8Col()      # unsigned byte
    grid_i    = Int32Col()      # integer
    grid_j    = Int32Col()      # integer
    pressure  = Float32Col()    # float  (single-precision)
    energy    = FloatCol()      # double (double-precision)

filename = "test.h5"
# Open a file in "w"rite mode
h5file = openFile(filename, mode = "w", title = "Test file")
# Create a new group under "/" (root)
group = h5file.createGroup("/", 'detector', 'Detector information')
# Create one table on it
table = h5file.createTable(group, 'readout', Particle, "Readout example")
# Fill the table with 10 particles
particle = table.row
for i in xrange(10):
    particle['name']  = 'Particle: %6d' % (i)
    particle['TDCcount'] = i % 256
    particle['ADCcount'] = (i * 256) % (1 << 16)
    particle['grid_i'] = i
    particle['grid_j'] = 10 - i
    particle['pressure'] = float(i*i)
    particle['energy'] = float(particle['pressure'] ** 4)
    particle['idnumber'] = i * (2 ** 34)
    # Insert a new particle record
    particle.append()
# Close (and flush) the file
h5file.close()

#now we will append some data to the table, after taking some slices 
f=tables.openFile(filename, mode="a")
f.root.detector
f.root.detector.readout
f.root.detector.readout[1::3]
f.root.detector.readout.attrs.TITLE
ro = f.root.detector.readout

#generators work
[row['energy'] for row in ro.where('pressure > 10')]


#append some data
table = f.root.detector.readout
particle = table.row
for i in xrange(10, 15):
  particle['name']  = 'Particle: %6d' % (i)
  particle['TDCcount'] = i % 256
  particle['ADCcount'] = (i * 256) % (1 << 16)
  particle['grid_i'] = i
  particle['grid_j'] = 10 - i
  particle['pressure'] = float(i*i)
  particle['energy'] = float(particle['pressure'] ** 4)
  particle['idnumber'] = i * (2 ** 34)
  particle.append()
table.flush()
f.close()

HDF5矩阵附加在python中

1 个答案: