Outer function does not work with vectors of float32 type

[DenisScherbakov]

Here is test code:

import dpnp
dtype = dpnp.float32
a = dpnp.array([0.1, 0.2, 0,3, 0.4, 0.5]).astype(dtype)
b = dpnp.array([0.1, 0.2, 0,3, 0.4, 0.5]).astype(dtype)
c = dpnp.outer(a,b)

print(c)

Error message is

File /localdisk/work/dscherba/miniconda3/lib/python3.9/site-packages/dpnp/dpnp_iface_linearalgebra.py:318, in outer(x1, x2, **kwargs)
    316 x2_desc = dpnp.get_dpnp_descriptor(x2)
    317 if x1_desc and x2_desc and not kwargs:
--> 318     return dpnp_outer(x1_desc, x2_desc).get_pyobj()
    320 return call_origin(numpy.outer, x1, x2, **kwargs)

File dpnp/dpnp_algo/dpnp_algo_linearalgebra.pyx:323, in dpnp.dpnp_algo.dpnp_algo.dpnp_outer()

File dpnp/dpnp_algo/dpnp_algo_linearalgebra.pyx:330, in dpnp.dpnp_algo.dpnp_algo.dpnp_outer()

AttributeError: 'dpctl.tensor._usmarray.usm_ndarray' object has no attribute 'flat'

[antonwolfy]

@DenisScherbakov, the issue is fixed in scope of PR #1205

[antonwolfy]

The fix is merged into gold/2021 by PR #1209.

[DenisScherbakov]

Currently, dpnp.outer() failing due to #1238 :

File "/home/gta/test2.py", line 13, in <module>
    c = outer(a,b)
  File "/home/gta/test2.py", line 9, in outer
    return dpnp.outer(a,b)
  File "/home/gta/ff2899c2ac822b2d/envs/intelpython_dpbench/lib/python3.9/site-packages/dpnp/dpnp_iface_linearalgebra.py", line 318, in outer
    return dpnp_outer(x1_desc, x2_desc).get_pyobj()
  File "dpnp/dpnp_algo/dpnp_algo_linearalgebra.pyx", line 369, in dpnp.dpnp_algo.dpnp_algo.dpnp_outer
  File "dpnp/dpnp_algo/dpnp_algo_linearalgebra.pyx", line 388, in dpnp.dpnp_algo.dpnp_algo.dpnp_outer
  File "/home/gta/ff2899c2ac822b2d/envs/intelpython_dpbench/lib/python3.9/site-packages/dpnp/dpnp_array.py", line 228, in __mul__
    return dpnp.multiply(self, other)
  File "/home/gta/ff2899c2ac822b2d/envs/intelpython_dpbench/lib/python3.9/site-packages/dpnp/dpnp_iface_mathematical.py", line 1125, in multiply
    return call_origin(numpy.multiply, x1, x2, dtype=dtype, out=out, where=where, **kwargs)
  File "dpnp/dpnp_utils/dpnp_algo_utils.pyx", line 120, in dpnp.dpnp_utils.dpnp_algo_utils.call_origin
NotImplementedError: Requested funtion=multiply with args=(<dpnp.dpnp_array.dpnp_array object at 0x7fd513309dc0>, <dpnp.dpnp_array.dpnp_array object at 0x7fd513309e20>) and kwargs={'dtype': None, 'out': None, 'where': True} isn't currently supported and would fall back on NumPy implementation. Define enviroment variable `DPNP_RAISE_EXCEPION_ON_NUMPY_FALLBACK` to `0` if the fall back is required to be supported without rasing an exception.

[DenisScherbakov]

Problem not observed with latest fix.