scipy.integrate.quad ctypes function error "quadpack.error: quad: first argument is a ctypes function pointer with incorrect signature"

Question

I am trying to use scipy.integrate.nquad with a ctypes function. I exactly followed the instruction on Faster integration using Ctypes.

ctypes integration can be done in a few simple steps:

1. Write an integrand function in C with the function signature double f(int n, double args[n]), where args is an array containing the arguments of the function f.

//testlib.c

double f(int n, double args[n])    
{
   return args[0] - args[1] * args[2]; //corresponds to x0 - x1 * x2
}

2. Now compile this file to a shared/dynamic library (a quick search will help with this as it is OS-dependent). The user must link any math libraries, etc. used. On Linux this looks like:

$ gcc -shared -o testlib.so -fPIC testlib.c

The output library will be referred to as testlib.so, but it may have a different file extension. A library has now been created that can be loaded into Python with ctypes.

3. Load shared library into Python using ctypes and set restypes and argtypes - this allows Scipy to interpret the function correctly:

>>> import ctypes

>>> from scipy import integrate

>>> lib = ctypes.CDLL('/**/testlib.so') # Use absolute path to testlib

>>> func = lib.f # Assign specific function to name func (for simplicity)

>>> func.restype = ctypes.c_double

>>> func.argtypes = (ctypes.c_int, ctypes.c_double)

Note that the argtypes will always be (ctypes.c_int, ctypes.c_double) regardless of the number of parameters, and restype will always be ctypes.c_double.

4. Now integrate the library function as normally, here using nquad:

>>> integrate.nquad(func, [[0,10],[-10,0],[-1,1]])

(1000.0, 1.1102230246251565e-11)

However, at the final step, I didn't get the result of the integral, but the following errors instead:

>>> integrate.nquad(func,[[0,1.0],[-2.0,3.0],[1.0,2.0]])

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 618, in nquad
      return _NQuad(func, ranges, opts).integrate(*args)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 670, in integrate

    value, abserr = quad(f, low, high, args=args, **opt)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 254, in quad

    retval = _quad(func,a,b,args,full_output,epsabs,epsrel,limit,points)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 319, in _quad

    return _quadpack._qagse(func,a,b,args,full_output,epsabs,epsrel,limit)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 670, in integrate

    value, abserr = quad(f, low, high, args=args, **opt)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 254, in quad

    retval = _quad(func,a,b,args,full_output,epsabs,epsrel,limit,points)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 319, in _quad

    return _quadpack._qagse(func,a,b,args,full_output,epsabs,epsrel,limit)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 670, in integrate

    value, abserr = quad(f, low, high, args=args, **opt)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 254, in quad

    retval = _quad(func,a,b,args,full_output,epsabs,epsrel,limit,points)

  File "/home/bfn097/apps/scipy/0.13.3_mkl-11.1.2_gcc-4.4.7/lib64/python/scipy/integrate/quadpack.py", line 319, in _quad

    return _quadpack._qagse(func,a,b,args,full_output,epsabs,epsrel,limit)

quadpack.error: quad: first argument is a ctypes function pointer with incorrect signature

I am using gcc-4.4.7,python 2.6.6, numpy-1.7.1, scipy-0.13.3

Answer 1

The type of the argument should be: ctypes.POINTER(ctypes.c_double)

But have you considered using cffi? Besides being faster than ctypes, you also don't have to hand-write the argument stuff, just copy the C declarations and let cffi parse them.