create a python sequence of 1-D array_like in C++ - c++

Question

I'm trying to embed Python in my C++ code.
Now what I'm trying to do is use the function scipy.stats.anderson_ksamp that takes as input a "sequence of 1-D array_like".
the thing is that in Python one would call the function in this simple way:
from scipy import stats
# two example np-arrays
samp1 = np.array([1,2,3,4,5,6,7,8,9,0])
samp2 = np.array([0,5,6,7,3,9,2,7,1,7])
ander_out1, ander_out2, ander_out3 = stats.anderson_ksamp([samp1, samp2])
print("Anderson statistic: %s, critical values: %s, significance level: %s"% (ander_out1, ander_out2, ander_out3))
In C++ instead I call this function in this way:
#include <Python.h>
#include <numpy/arrayobject.h>
std::vector<int> sample1 = {1,2,3,4,5,6,7,8,9,0};
std::vector<int> sample2 = {0,5,6,7,3,9,2,7,1,7};
Py_Initialize();
import_array();
npy_intp dim1[] = {(npy_intp)sample1.size()};
PyObject* np_array1 = PyArray_SimpleNewFromData(1, dim1, NPY_INT, &sample1[0]);
npy_intp dim2[] = {(npy_intp)sample2.size()};
PyObject* np_array2 = PyArray_SimpleNewFromData(1, dim2, NPY_INT, &sample2[0]);
pModule2 = PyImport_ImportModule("scipy.stats");
if (pModule2 != NULL) {
pValue3 = PyObject_CallMethod(pModule2, "anderson_ksamp", "O", /*HERE GOES THE 1-D SEQUENCE*/);
if (pValue3 != NULL) {
Py_ssize_t pValue3Size = PyTuple_Size(pValue3); // size of the tuple
printf("Anderson tuple: (");
for (Py_ssize_t j = 0; j < pValue3Size; j++) {
pElem3 = PyTuple_GetItem(pValue3, j); // getting item j of tuple
printf("%f", PyFloat_AsDouble(pElem3)); // printing item j
if (j!=pValue3Size-1) {
printf(", ");
}
}
printf(")\n");
} else {
printf("Call to anderson_ksamp failed.\n");
PyErr_Print();
}
}
What I miss is how to "merge" the two vectors sample1 and sample2 in order to make a Python object that resembles the Python statement [samp1, samp2].
I tried to create a 2-D vector in C++
std::vector<std::vector<int>> sample3 = {{1,2,3,4,5,6,7,8,9,0}, {0,5,6,7,3,9,2,7,1,7}};
and then use
npy_intp dim3[] = {10, 2};
PyObject* both_np_arrays = PyArray_SimpleNewFromData(2, dim3, NPY_INT, &sample3);
in order to create the object I need. But the object I get is something like this:
array([[ 8003984, 1],
[ 8004032, 1],
[ 8004032, 1],
[50571184, 1],
[50411920, 1],
[ 0, 0],
[ 0, 0],
[ 0, 0],
[ 0, 0],
[ 0, 0]], dtype=int32)
That is definitely not what I want.
I've also tried to do it this way:
int sample3[10][2] = {{1,0}, {2,5}, {3,6}, {4,7}, {5,3}, {6,9}, {7,2}, {8,7}, {9,1}, {0,7}};
npy_intp dim3[] = {10, 2};
PyObject* both_np_arrays = PyArray_SimpleNewFromData(2, dim3, NPY_INT, &sample3);
but the result is this:
array([[1, 0],
[2, 5],
[3, 6],
[4, 7],
[5, 3],
[6, 9],
[7, 2],
[8, 7],
[9, 1],
[0, 7]], dtype=int32)
so, again, not what I need.
I think that to use the function properly I'd need to obtain something like this:
array([[1, 2, 3, 4, 5, 6, 7, 8, 9, 0],
[0, 5, 6, 7, 3, 9, 2, 7, 1, 7]], dtype=int32)
How can I do it?