Python numpy.require() Examples

def gw_comp_veff(self, vext, comega=1j*0.0):
    """
    This computes an effective field (scalar potential) given the external
    scalar potential as follows:
        (1-v\chi_{0})V_{eff} = V_{ext} = X_{a}^{n}V_{\mu}^{ab}X_{b}^{m} * 
                                         v\chi_{0}v * X_{a}^{n}V_{nu}^{ab}X_{b}^{m}
    
    returns V_{eff} as list for all n states(self.nn[s]).
    """
    
    from scipy.sparse.linalg import LinearOperator
    self.comega_current = comega
    veff_op = LinearOperator((self.nprod,self.nprod),
                             matvec=self.gw_vext2veffmatvec,
                             dtype=self.dtypeComplex)

    from scipy.sparse.linalg import lgmres
    resgm, info = lgmres(veff_op,
                         np.require(vext, dtype=self.dtypeComplex, requirements='C'),
                         atol=self.gw_iter_tol, maxiter=self.maxiter)
    if info != 0:
      print("LGMRES has not achieved convergence: exitCode = {}".format(info))
    return resgm 

def get_ac_vertex_array(self, dtype=np.float64):
    """ Returns the product vertex coefficients as 3d array (dense table) """
    atom2so = self.sv.atom2s
    nfap = self.c2s[-1]
    n = self.sv.atom2s[-1]
    pab2v = np.require( zeros((nfap,n,n), dtype=dtype), requirements='CW')
    for atom,[sd,fd,pt,spp] in enumerate(zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp)):
      if pt!=1: continue
      s,f = atom2so[atom:atom+2]
      pab2v[sd:fd,s:f,s:f] = self.prod_log.sp2vertex[spp]

    for sd,fd,pt,spp in zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp):
      if pt!=2: continue
      inf= self.bp2info[spp]
      lab = einsum('dl,dab->lab', inf.cc, inf.vrtx)
      a,b = inf.atoms
      sa,fa,sb,fb = atom2so[a],atom2so[a+1],atom2so[b],atom2so[b+1]
      for c,ls,lf in zip(inf.cc2a, inf.cc2s, inf.cc2s[1:]):
        pab2v[self.c2s[c]:self.c2s[c+1],sa:fa,sb:fb] = lab[ls:lf,:,:]
        pab2v[self.c2s[c]:self.c2s[c+1],sb:fb,sa:fa] = einsum('pab->pba', lab[ls:lf,:,:])
    return pab2v 

def get_dp_vertex_array(self, dtype=np.float64):
    """ Returns the product vertex coefficients as 3d array for dominant products """
    atom2so = self.sv.atom2s
    nfdp = self.dpc2s[-1]
    n = self.sv.atom2s[-1]
    pab2v = np.require(np.zeros((nfdp,n,n), dtype=dtype), requirements='CW')
    for atom,[sd,fd,pt,spp] in enumerate(zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp)):
      if pt!=1: continue
      s,f = atom2so[atom:atom+2]
      pab2v[sd:fd,s:f,s:f] = self.prod_log.sp2vertex[spp]

    for sd,fd,pt,spp in zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp):
      if pt!=2: continue
      inf= self.bp2info[spp]
      a,b = inf.atoms
      sa,fa,sb,fb = atom2so[a],atom2so[a+1],atom2so[b],atom2so[b+1]
      pab2v[sd:fd,sa:fa,sb:fb] = inf.vrtx
      pab2v[sd:fd,sb:fb,sa:fa] = einsum('pab->pba', inf.vrtx)
    return pab2v 

def get_ac_vertex_array(self, dtype=np.float64):
    """ Returns the product vertex coefficients as 3d array (dense table) """
    atom2so = self.sv.atom2s
    nfap = self.c2s[-1]
    n = self.sv.atom2s[-1]
    pab2v = np.require( np.zeros((nfap,n,n), dtype=dtype), requirements='CW')
    for atom,[sd,fd,pt,spp] in enumerate(zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp)):
      if pt!=1: continue
      s,f = atom2so[atom:atom+2]
      pab2v[sd:fd,s:f,s:f] = self.prod_log.sp2vertex[spp]

    for sd,fd,pt,spp in zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp):
      if pt!=2: continue
      inf= self.bp2info[spp]
      lab = einsum('dl,dab->lab', inf.cc, inf.vrtx)
      a,b = inf.atoms
      sa,fa,sb,fb = atom2so[a],atom2so[a+1],atom2so[b],atom2so[b+1]
      for c,ls,lf in zip(inf.cc2a, inf.cc2s, inf.cc2s[1:]):
        pab2v[self.c2s[c]:self.c2s[c+1],sa:fa,sb:fb] = lab[ls:lf,:,:]
        pab2v[self.c2s[c]:self.c2s[c+1],sb:fb,sa:fa] = einsum('pab->pba', lab[ls:lf,:,:])
    return pab2v 

def assign(self, coords, mask=None, output=None):
        try:
            passed_coord_dtype = coords.dtype
        except AttributeError:
            coords = numpy.require(coords, dtype=coord_dtype)
        else:
            if passed_coord_dtype != coord_dtype:
                coords = numpy.require(coords, dtype=coord_dtype)

        if coords.ndim != 2:
            raise TypeError('coords must be 2-dimensional')
        if mask is None:
            mask = numpy.ones((len(coords),), dtype=numpy.bool_)
        elif len(mask) != len(coords):
            raise TypeError('mask [shape {}] has different length than coords [shape {}]'.format(mask.shape, coords.shape))

        if output is None:
            output = numpy.empty((len(coords),), dtype=index_dtype)
        elif len(output) != len(coords):
            raise TypeError('output has different length than coords')

        apply_down(self.func, self.args, self.kwargs, coords, mask, output)

        return output 

def assign(self, coords, mask=None, output=None):
        try:
            passed_coord_dtype = coords.dtype
        except AttributeError:
            coords = numpy.require(coords, dtype=coord_dtype)
        else:
            if passed_coord_dtype != coord_dtype:
                coords = numpy.require(coords, dtype=coord_dtype)

        if coords.ndim != 2:
            raise TypeError('coords must be 2-dimensional')
        if mask is None:
            mask = numpy.ones((len(coords),), dtype=numpy.bool_)
        elif len(mask) != len(coords):
            raise TypeError('mask [shape {}] has different length than coords [shape {}]'.format(mask.shape, coords.shape))

        if output is None:
            output = numpy.empty((len(coords),), dtype=index_dtype)
        elif len(output) != len(coords):
            raise TypeError('output has different length than coords')

        self.func(coords, mask, output, *self.args, **self.kwargs)

        return output 

def dens_libnao(crds, nspin):
  """  Compute the electronic density using library call """
  assert crds.ndim==2  
  assert crds.shape[-1]==3
  
  nc = crds.shape[0]
  crds_cp = require(crds, dtype=c_double, requirements='C')
  dens = require( zeros((nc, nspin)), dtype=c_double, requirements='CW')

  libnao.dens_libnao(
    crds_cp.ctypes.data_as(POINTER(c_double)),
    c_int64(nc),
    dens.ctypes.data_as(POINTER(c_double)),
    c_int64(nspin))

  return dens 

def test_sumidxtab_core(self):
        nsq = 8
        ksub = 256
        cur_num = 10

        for iround in range(10):
            raw_D = np.random.random((nsq, ksub))
            raw_blk = np.random.random_integers(0, ksub-1, (cur_num, nsq))
            D = np.require(raw_D, np.float32, "C")
            blk = np.require(raw_blk, np.uint8, "C")

            self.assertLessEqual(np.abs(raw_D - D).sum(),  1e-4)
            self.assertEqual(np.abs(raw_blk - blk).sum(),  0)

            py_res = self.sumidxtab_core(D, blk)
            c_res = cext.sumidxtab_core(D, blk)
            self.assertLessEqual(np.abs(py_res - c_res).sum(),  1e-4) 

def init_mo_from_pyscf(self, **kw):
    """ Initializing from a previous pySCF mean-field calc. """
    from pyscf.nao.m_fermi_energy import fermi_energy as comput_fermi_energy
    from pyscf.nao.m_color import color as tc
    self.telec = kw['telec'] if 'telec' in kw else 0.0000317 # 10K
    self.mf = mf = kw['mf']
    self.xc_code = mf.xc if hasattr(mf, 'xc') else 'HF'
    self.k2xyzw = np.array([[0.0,0.0,0.0,1.0]])
    
    self.mo_energy = np.asarray(mf.mo_energy)
    self.nspin = self.mo_energy.ndim
    assert self.nspin in [1,2]
    nspin,n=self.nspin,self.norbs
    self.mo_energy = require( self.mo_energy.reshape((1, nspin, n)), requirements='CW')
    self.mo_occ = require( mf.mo_occ.reshape((1,nspin,n)), requirements='CW')    
    self.mo_coeff =  require(zeros((1,nspin,n,n,1), dtype=self.dtype), requirements='CW')
    conv = conv_yzx2xyz_c(kw['gto'])
    aaux = np.asarray(mf.mo_coeff).reshape((nspin,n,n))
    for s in range(nspin):
      self.mo_coeff[0,s,:,:,0] = conv.conv_yzx2xyz_1d(aaux[s], conv.m_xyz2m_yzx).T

    self.nelec = kw['nelec'] if 'nelec' in kw else np.array([int(s2o.sum()) for s2o in self.mo_occ[0]])
    fermi = comput_fermi_energy(self.mo_energy, sum(self.nelec), self.telec)
    self.fermi_energy = kw['fermi_energy'] if 'fermi_energy' in kw else fermi 

def init_mo_coeff_fireball(self, **kw):
    """ Constructor a mean-field class from the preceeding FIREBALL calculation """
    from pyscf.nao.m_fermi_dirac import fermi_dirac_occupations
    from pyscf.nao.m_fireball_get_eigen_dat import fireball_get_eigen_dat
    from pyscf.nao.m_fireball_hsx import fireball_hsx
    self.telec = kw['telec'] if 'telec' in kw else self.telec
    self.fermi_energy = kw['fermi_energy'] if 'fermi_energy' in kw else self.fermi_energy
    self.mo_energy = require(fireball_get_eigen_dat(self.cd), dtype=self.dtype, requirements='CW')
    ksn2fd = fermi_dirac_occupations(self.telec, self.mo_energy, self.fermi_energy)
    self.mo_occ = (3-self.nspin)*ksn2fd
    if abs(self.nelectron-self.mo_occ.sum())>1e-6: raise RuntimeError("mo_occ wrong?" )
    #print(__name__, ' self.nspecies ', self.nspecies)
    #print(self.sp_mu2j)
    
    self.hsx = fireball_hsx(self, **kw)
    #print(self.telec)
    #print(self.mo_energy)
    #print(self.fermi_energy)
    #print(__name__, ' sum(self.mo_occ)', sum(self.mo_occ))
    #print(self.mo_occ) 

def assign(self, coords, mask=None, output=None):
        try:
            passed_coord_dtype = coords.dtype
        except AttributeError:
            coords = numpy.require(coords, dtype=coord_dtype)
        else:
            if passed_coord_dtype != coord_dtype:
                coords = numpy.require(coords, dtype=coord_dtype)

        if coords.ndim != 2:
            raise TypeError('coords must be 2-dimensional')

        if mask is None:
            mask = numpy.ones((len(coords),), dtype=numpy.bool_)
        elif len(mask) != len(coords):
            raise TypeError('mask [shape {}] has different length than coords [shape {}]'.format(mask.shape, coords.shape))

        if output is None:
            output = numpy.empty((len(coords),), dtype=index_dtype)
        elif len(output) != len(coords):
            raise TypeError('output has different length than coords')

        rectilinear_assign(coords, mask, output, self.boundaries, self._boundlens)

        return output 

def aos_libnao(coords, norbs):
  assert len(coords.shape) == 2
  assert coords.shape[1] == 3
  assert norbs>0

  ncoords = coords.shape[0]
  co2val = np.require( np.zeros((ncoords,norbs)), dtype=c_double, requirements='CW')
  crd_copy = np.require(coords, dtype=c_double, requirements='C')

  libnao.aos_libnao(
    c_int64(ncoords),
    crd_copy.ctypes.data_as(POINTER(c_double)),
    c_int64(norbs),
    co2val.ctypes.data_as(POINTER(c_double)),
    c_int64(norbs))

  return co2val 

def assign(self, coords, mask=None, output=None):
        if output is None:
            output = numpy.zeros((len(coords),), dtype=index_dtype)

        if mask is None:
            mask = numpy.ones((len(coords),), dtype=numpy.bool_)
        else:
            mask = numpy.require(mask, dtype=numpy.bool_)

        coord_subset = coords[mask]
        fnvals = numpy.empty((len(coord_subset), len(self.functions)), dtype=index_dtype)
        for ifn, fn in enumerate(self.functions):
            rsl = numpy.apply_along_axis(fn,0,coord_subset)
            if rsl.ndim > 1:
                # this should work like a squeeze, unless the function returned something truly
                # stupid (e.g., a 3d array with at least two dimensions greater than 1), in which
                # case a broadcast error will occur
                fnvals[:,ifn] = rsl.flat
            else:
                fnvals[:,ifn] = rsl
        amask = numpy.require(fnvals.argmax(axis=1), dtype=index_dtype)
        output[mask] = amask
        return output 

def get_dp_vertex_array(self, dtype=np.float64):
    """ Returns the product vertex coefficients as 3d array for dominant products """
    atom2so = self.sv.atom2s
    nfdp = self.dpc2s[-1]
    n = self.sv.atom2s[-1]
    pab2v = np.require(zeros((nfdp,n,n), dtype=dtype), requirements='CW')
    for atom,[sd,fd,pt,spp] in enumerate(zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp)):
      if pt!=1: continue
      s,f = atom2so[atom:atom+2]
      pab2v[sd:fd,s:f,s:f] = self.prod_log.sp2vertex[spp]

    for sd,fd,pt,spp in zip(self.dpc2s,self.dpc2s[1:],self.dpc2t,self.dpc2sp):
      if pt!=2: continue
      inf= self.bp2info[spp]
      a,b = inf.atoms
      sa,fa,sb,fb = atom2so[a],atom2so[a+1],atom2so[b],atom2so[b+1]
      pab2v[sd:fd,sa:fa,sb:fb] = inf.vrtx
      pab2v[sd:fd,sb:fb,sa:fa] = einsum('pab->pba', inf.vrtx)
    return pab2v 

def rsphar_exp_vec(rvs,lmax):
  """
    Computes (all) real spherical harmonics up to the angular momentum lmax
    Args:
      rvs : Cartesian coordinates defining correct theta and phi angles for spherical harmonic
      lmax : Integer, maximal angular momentum
    Result:
      1-d numpy array of float64 elements with all spherical harmonics stored in order 0,0; 1,-1; 1,0; 1,+1 ... lmax,lmax, althogether 0 : (lmax+1)**2 elements.
  """  
  assert rvs.shape[0]==3
  r_cp = np.require(rvs,  dtype=np.float64, requirements='C')
  nc = rvs[0,...].size
  res = np.require( np.zeros(((lmax+1)**2,nc)), dtype=np.float64, requirements='CW')
  libnao.rsphar_exp_vec(r_cp.ctypes.data_as(POINTER(c_double)), c_int64(nc), c_int64(lmax), res.ctypes.data_as(POINTER(c_double)))
  
  #for irv,rvec in enumerate(rvs): rsphar(rvec,lmax,res[irv,:])
  return res 

def rsphar_vec(rvs,lmax):
  """
    Computes (all) real spherical harmonics up to the angular momentum lmax
    Args:
      rvs : Cartesian coordinates defining correct theta and phi angles for spherical harmonic
      lmax : Integer, maximal angular momentum
    Result:
      1-d numpy array of float64 elements with all spherical harmonics stored in order 0,0; 1,-1; 1,0; 1,+1 ... lmax,lmax, althogether 0 : (lmax+1)**2 elements.
  """
  assert rvs.shape[-1]==3
  r_cp = np.require(rvs,  dtype=float, requirements='C')
  nc = len(rvs)
  res = np.require( np.zeros((nc, (lmax+1)**2)), dtype=float, requirements='CW')
  libnao.rsphar_vec(r_cp.ctypes.data_as(POINTER(c_double)), c_int64(nc), c_int64(lmax), res.ctypes.data_as(POINTER(c_double)))
  
  #for irv,rvec in enumerate(rvs): rsphar(rvec,lmax,res[irv,:])
  return res

#
#
# 

def rsphar(r,lmax,res):
  """
    Computes (all) real spherical harmonics up to the angular momentum lmax
    Args:
      r : Cartesian coordinates defining correct theta and phi angles for spherical harmonic
      lmax : Integer, maximal angular momentum
    Result:
      1-d numpy array of float64 elements with all spherical harmonics stored in order 0,0; 1,-1; 1,0; 1,+1 ... lmax,lmax, althogether 0 : (lmax+1)**2 elements.
  """
  assert r.shape[-1]==3
  
  r_cp = np.require(r,  dtype=float, requirements='C')
  res = np.require(res,  dtype=float, requirements='CW')
  
  libnao.rsphar(r_cp.ctypes.data_as(POINTER(c_double)), c_int64(lmax), res.ctypes.data_as(POINTER(c_double)))
  return 0

#
#
# 

def build(self, pardic=None):
        # training data
        vals = pardic['vals']
        # the number of subquantizers
        nsubq = pardic['nsubq']
        # the number bits of each subquantizer
        nsubqbits = pardic.get('nsubqbits', 8)
        # the number of items in one block
        blksize = pardic.get('blksize', 16384)

        # vector dimension
        dim = vals.shape[1]
        # dimension of the subvectors to quantize
        dsub = dim / nsubq
        # number of centroids per subquantizer
        ksub = 2 ** nsubqbits

        """
        Initializing indexer data
        """
        ecdat = dict()
        ecdat['nsubq'] = nsubq
        ecdat['ksub'] = ksub
        ecdat['dsub'] = dsub
        ecdat['blksize'] = blksize
        ecdat['centroids'] = [None for q in range(nsubq)]

        logging.info("Building codebooks in subspaces - BEGIN")
        for q in range(nsubq):
            logging.info("\tsubspace %d/%d" % (q, nsubq))
            vs = np.require(vals[:, q*dsub:(q+1)*dsub],
                            requirements='C', dtype=np.float32)
            ecdat['centroids'][q] = kmeans(vs, ksub, niter=100)
        logging.info("Building codebooks in subspaces - DONE")

        self.ecdat = ecdat 

def construct_histogram(self):
        '''Construct a histogram using bins previously constructed with ``construct_bins()``.
        The time series of histogram values is stored in ``histograms``.
        Each histogram in the time series is normalized.'''
        
        self.scan_data_shape()
        
        iter_count = self.iter_stop - self.iter_start
        histograms_ds = self.output_file.create_dataset('histograms', dtype=numpy.float64,
                                                        shape=((iter_count,) + tuple(len(bounds)-1 for bounds in self.binbounds)),
                                                        compression=9 if self.compress_output else None)
        binbounds = [numpy.require(boundset, self.dset_dtype, 'C') for boundset in self.binbounds]

        
        
        self.progress.indicator.new_operation('Constructing histograms',self.iter_stop-self.iter_start)
        task_gen = ((_remote_bin_iter, (iiter, n_iter, self.duration_dsspec, self.wt_dsspec, 0, binbounds,
                                        self.ignore_out_of_range), {}) 
                    for (iiter,n_iter) in enumerate(range(self.iter_start, self.iter_stop)))
        #futures = set()
        #for iiter, n_iter in enumerate(xrange(self.iter_start, self.iter_stop)):
        #    initpoint = 1 if iiter > 0 else 0
        #    futures.add(self.work_manager.submit(_remote_bin_iter,
        #                                            args=(iiter, n_iter, self.dsspec, self.wt_dsspec, initpoint, binbounds)))
        
        #for future in self.work_manager.as_completed(futures):
            #future = self.work_manager.wait_any(futures)
        #for future in self.work_manager.submit_as_completed(task_gen, self.queue_size):
        log.debug('max queue length: {!r}'.format(self.max_queue_len))
        for future in self.work_manager.submit_as_completed(task_gen, self.max_queue_len):
            iiter, n_iter, iter_hist = future.get_result(discard=True)
            self.progress.indicator.progress += 1

            # store histogram
            histograms_ds[iiter] = iter_hist
            del iter_hist, future 

def ascontiguousarray(a, dtype=None):
    """
    Return a contiguous array (ndim >= 1) in memory (C order).

    Parameters
    ----------
    a : array_like
        Input array.
    dtype : str or dtype object, optional
        Data-type of returned array.

    Returns
    -------
    out : ndarray
        Contiguous array of same shape and content as `a`, with type `dtype`
        if specified.

    See Also
    --------
    asfortranarray : Convert input to an ndarray with column-major
                     memory order.
    require : Return an ndarray that satisfies requirements.
    ndarray.flags : Information about the memory layout of the array.

    Examples
    --------
    >>> x = np.arange(6).reshape(2,3)
    >>> np.ascontiguousarray(x, dtype=np.float32)
    array([[ 0.,  1.,  2.],
           [ 3.,  4.,  5.]], dtype=float32)
    >>> x.flags['C_CONTIGUOUS']
    True

    Note: This function returns an array with at least one-dimension (1-d) 
    so it will not preserve 0-d arrays.  

    """
    return array(a, dtype, copy=False, order='C', ndmin=1) 

def hamming_core(cls, D, blk):
        return np.require([sum([D[j, blk[i, j]] for j in range(D.shape[0])])
                           for i in range(blk.shape[0])], np.float32) 

def encode(self, vals):
        X = vals
        if X.ndim == 1:
            X = X.reshape((1, -1))

        Nsamples, Ndim = X.shape
        nbits = self.ecdat['nbits']
        mn = self.ecdat['mn']
        mx = self.ecdat['mx']
        pc = self.ecdat['pc']
        modes = self.ecdat['modes']

        X = X.dot(pc)
        X = X - mn.reshape((1, -1))
        omega0 = 0.5 / (mx - mn)
        omegas = modes * omega0.reshape((1, -1))

        U = np.zeros((Nsamples, nbits))
        for i in range(nbits):
            omegai = omegas[i, :]
            ys = X * omegai + 0.25
            ys -= np.floor(ys)
            yi = np.sum(ys < 0.5, 1)
            U[:, i] = yi

        b = np.require(U % 2 == 0, dtype=np.int)
        B = self.compactbit(b)
        return B 

def test_non_array_input(self):
        a = np.require([1, 2, 3, 4], 'i4', ['C', 'A', 'O'])
        assert_(a.flags['O'])
        assert_(a.flags['C'])
        assert_(a.flags['A'])
        assert_(a.dtype == 'i4')
        assert_equal(a, [1, 2, 3, 4]) 

def test_unknown_requirement(self):
        a = self.generate_all_false('f8')
        assert_raises(KeyError, np.require, a, None, 'Q') 

def set_and_check_flag(self, flag, dtype, arr):
        if dtype is None:
            dtype = arr.dtype
        b = np.require(arr, dtype, [flag])
        assert_(b.flags[flag])
        assert_(b.dtype == dtype)

        # a further call to np.require ought to return the same array
        # unless OWNDATA is specified.
        c = np.require(b, None, [flag])
        if flag[0] != 'O':
            assert_(c is b)
        else:
            assert_(c.flags[flag]) 

def asfortranarray(a, dtype=None):
    """
    Return an array (ndim >= 1) laid out in Fortran order in memory.

    Parameters
    ----------
    a : array_like
        Input array.
    dtype : str or dtype object, optional
        By default, the data-type is inferred from the input data.

    Returns
    -------
    out : ndarray
        The input `a` in Fortran, or column-major, order.

    See Also
    --------
    ascontiguousarray : Convert input to a contiguous (C order) array.
    asanyarray : Convert input to an ndarray with either row or
        column-major memory order.
    require : Return an ndarray that satisfies requirements.
    ndarray.flags : Information about the memory layout of the array.

    Examples
    --------
    >>> x = np.arange(6).reshape(2,3)
    >>> y = np.asfortranarray(x)
    >>> x.flags['F_CONTIGUOUS']
    False
    >>> y.flags['F_CONTIGUOUS']
    True

    Note: This function returns an array with at least one-dimension (1-d) 
    so it will not preserve 0-d arrays.  

    """
    return array(a, dtype, copy=False, order='F', ndmin=1) 

def __init__(self, data, reducefix=0):
        assert(data.dtype == _np.dtype('d'))
        if reducefix == 0:
            self.base = data
        else:
            # because serialization of numpy array flags is borked (around Numpy v1.16), we need to copy data
            # (so self.base *owns* it's data) and manually convey the writeable flag.
            self.base = _np.require(data.copy(), requirements=['OWNDATA', 'C_CONTIGUOUS'])
            self.base.flags.writeable = True if reducefix == 1 else False 

def test_ensure_array(self):
        class ArraySubclass(np.ndarray):
            pass

        a = ArraySubclass((2, 2))
        b = np.require(a, None, ['E'])
        assert_(type(b) is np.ndarray) 

def sumidxtab_core(cls, D, blk):
        return np.require([sum([D[j, blk[i, j]] for j in range(D.shape[0])])
                           for i in range(blk.shape[0])], np.float32) 

def _get_parent_ids(n_iter, iter_group):
    seg_index = iter_group['seg_index']
    try:
        return seg_index['parent_id'][:]
    except ValueError:
        # field not found
        offsets = seg_index['parents_offset'][:]
        all_parents = iter_group['parents'][...]
        return numpy.require(all_parents.take(offsets),dtype=numpy.int64)
    else:
        return seg_index['parent_id']