Question

如何在Python CULA程序中加入Python的solve（）？我正在使用

LA = libculaC.solve()

结果：

Traceback (most recent call last):
      File "culaTest.py", line 96, in <module>
        LA = libculaC.solve(0)
      File "/usr/lib/python2.7/ctypes/__init__.py", line 378, in __getattr__
        func = self.__getitem__(name)
      File "/usr/lib/python2.7/ctypes/__init__.py", line 383, in __getitem__
        func = self._FuncPtr((name_or_ordinal, self))
    AttributeError: /usr/local/cula/lib64/libcula_lapack.so: undefined symbol:     
    solve

liculaC和ctypes的任何组合都会给我类似的错误。我该如何使用此功能？我是否需要使用C函数（scanf）或其他东西。感谢。

Answer 1

这花了一段时间，但这是我到目前为止所做的。必须使用ctypes并且必须转换为列主副标准行主要。使用矩阵而不是数组，并始终保持数据类型的一致性。

import ctypes
from scipy import *
from scipy.linalg import *
import numpy as np
import sys
import csv
print     "___________________________________________________________________________________________________"

libculaC=ctypes.CDLL('libcula_lapack.so',mode=ctypes.RTLD_GLOBAL)
libculaC.culaGetStatusString.restype=ctypes.c_char_p

info=libculaC.culaInitialize()

#Row major-normal form, but must be converted-('4 1;2 5')
Anp = np.matrix('4.0 2.0;1.0 5.0') #Column major
print "This is Anp: "
print Anp
print '___________END Anp______________'

#use ctypes to convert from Py to C
#2x2 matrix
Anp = Anp.astype(numpy.float32)  #astype is array type for ctype
c_float_p = ctypes.POINTER(ctypes.c_float)
A1_p = Anp.ctypes.data_as(c_float_p)
# 2x1 matrix
B1 = np.matrix('5.0 ;7.0')
print "This is B1"
print B1
print '__________________B1 END______________________'
B1 = B1.astype(numpy.float32)
B1_p = B1.ctypes.data_as(c_float_p)

X=np.empty([2])
X=X.astype(numpy.float32)
X_p =X.ctypes.data_as(c_float_p)
print "This is X"
print X
print '__________________X END______________________'


info = 0
libculaC.culaSgesv(2,1,A1_p,2,X_p,B1_p,2)  #libculaC.culaSgesv

a = np.fromiter(B1_p, dtype=np.float32, count=2)
a = np.reshape(a,(-1,2))
print "The solution returning from Sgesv: "
print a
print "-----------------------Program End----------------------------"

libculaC.culaShutdown()

输出：这是Anp：

[[4. 2.] [1. 5。]]

___________ END Anp ______________

这是B1

[[5.] [7。]]

__________________ B1结束______________________

这是X

[5. 7。]

__________________ X END ______________________

从Sgesv返回的解决方案：

[[1. 1。]]

-----------------------程序结束---------------------- ------

CULA - 在CULA中使用Python solve（）

1 个答案: