bdiv bug again

[John-zzh]

#540 is resolved, but the subsequent orbital_pair_cart2sph has a following up error, probably wrong array shape.

Traceback (most recent call last):
  File "/root/jojo/ris_tests/extest42/chkfiles_def2SVP/pbe0/55_DPP1/bdiv_bug.py", line 87, in <module>
    eri3c_batch = int3c2e_opt.orbital_pair_cart2sph(eri3c_batch, inplace=True)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/jojo/software/gpu4pyscf_ris/gpu4pyscf/df/int3c2e_bdiv.py", line 506, in orbital_pair_cart2sph
    t = contract('mpqr,pj->mjqr', t, cj)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/jojo/software/gpu4pyscf_ris/gpu4pyscf/lib/cutensor.py", line 163, in contract
    return contraction(pattern, a, b, alpha, beta, out=out)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/jojo/software/gpu4pyscf_ris/gpu4pyscf/lib/cutensor.py", line 91, in contraction
    desc_a = cutensor.create_tensor_descriptor(a)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "cupyx/cutensor.pyx", line 235, in cupyx.cutensor.create_tensor_descriptor
  File "cupyx/cutensor.pyx", line 255, in cupyx.cutensor.create_tensor_descriptor
  File "cupyx/cutensor.pyx", line 79, in cupyx.cutensor.TensorDescriptor.__init__
  File "cupy_backends/cuda/libs/cutensor.pyx", line 226, in cupy_backends.cuda.libs.cutensor.createTensorDescriptor
  File "cupy_backends/cuda/libs/cutensor.pyx", line 189, in cupy_backends.cuda.libs.cutensor.check_status
cupy_backends.cuda.libs.cutensor.CuTensorError: CUTENSOR_STATUS_NOT_SUPPORTED

from pyscf.df import incore
import pyscf
from gpu4pyscf.df import int3c2e_bdiv


mol = pyscf.M(
    atom='''
C     7.2664723   -0.0499943   -0.0065814 
C     7.1293317    1.3272633    0.2300085 
C     5.7902954    1.7503053    0.2723592 
C     4.8577950    0.7219655    0.0716040 
S     5.6819288   -0.7938244   -0.1722845 
H     7.9830786    1.9934581    0.3648773 
H     5.4983874    2.7886551    0.4429769 
C    -0.9120223    1.0426618   -0.0564051 
C    -1.8046659   -0.0450478   -0.0020236 
N    -1.6944313    2.1954095   -0.1378876 
C    -3.1470577    0.4382575   -0.0229097 
C    -3.1048622    1.8875924   -0.1162697 
C    -1.8495179   -1.4947275    0.0924656 
O    -0.9922691   -2.3766087    0.1408380 
N    -3.2619128   -1.8011177    0.1123776 
C    -4.0402808   -0.6478351    0.0321611 
O    -3.9645573    2.7674408   -0.1625691 
C    -1.2994101    3.5801035   -0.3457870 
H    -0.8941850    4.0362632    0.5745842 
H    -0.5592670    3.6609730   -1.1577256 
H    -2.2215514    4.1162944   -0.6188625 
C    -3.6623799   -3.1846460    0.3180075 
H    -4.0624797   -3.6394978   -0.6052655 
H    -2.7445188   -3.7243672    0.5982990 
H    -4.4092758   -3.2621573    1.1240239 
C    -8.3153036   -0.3890379   -0.0513520 
C    -7.5356331    0.7462399    0.2188833 
C    -7.7103592   -1.6258423   -0.3173492 
C    -6.1453369    0.6364155    0.2376686 
H    -8.0131390    1.7113545    0.4073224 
C    -6.3176166   -1.7292837   -0.2899240 
H    -8.3238887   -2.4998883   -0.5507637 
C    -5.5006445   -0.6069967    0.0010727 
H    -5.5454868    1.5354030    0.4104793 
H    -5.8758409   -2.6957884   -0.5290363 
C     0.5444130    1.0066291   -0.0167939 
C     1.3594225    2.1383186    0.2518919 
C     1.2033855   -0.2373239   -0.2205366 
C     2.7466528    2.0400635    0.2853088 
H     0.9120585    3.1071706    0.4678860 
C     2.5895867   -0.3264924   -0.1935757 
H     0.6115191   -1.1441206   -0.3779559 
C     3.4034596    0.8079664    0.0500586 
H     3.3311768    2.9370434    0.5069021 
H     3.0581271   -1.3025041   -0.3627858 
Br  -10.2111239   -0.2479833   -0.0751307 
C     8.4168303   -0.8906528   -0.1266851 
C     9.7525028   -0.5500912   -0.0365112 
C    10.2011789    0.7832768    0.2064403 
C    10.8248189   -1.5688820   -0.1897493 
N    10.5237407    1.8920973    0.4070390 
O    12.0162313   -1.3320076   -0.1205693 
O    10.3312507   -2.8183344   -0.4204713 
H    11.1168476   -3.3970633   -0.5033726 
H     8.2265581   -1.9524117   -0.3163640 
''',                     
    basis='def2-svp',                        
    verbose=4,
    charge=0                      
    )


auxmol = mol.copy()
auxmol.basis = {'H': [[0, [0.1999828038466018, 1.0]]], 'O': [[0, [0.2587932305664396, 1.0]]], 'C': [[0, [0.1320292535005648, 1.0]]], 'S': [[0, [0.07333458614245612, 1.0]]], 'Br': [[0, [0.06164023088322297, 1.0]]], 'N': [[0, [0.1900871004703463, 1.0]]]}

auxmol.build()


dat = int3c2e_bdiv.aux_e2(mol, auxmol)
ref = incore.aux_e2(mol, auxmol)


print('abs(dat.get()-ref).max()', abs(dat.get()-ref).max())
assert abs(dat.get()-ref).max() < 1e-10


int3c2e_opt = int3c2e_bdiv.Int3c2eOpt(mol, auxmol).build()

eri3c_batch = next(int3c2e_opt.int3c2e_bdiv_generator())
eri3c_batch = int3c2e_opt.orbital_pair_cart2sph(eri3c_batch, inplace=True)