How can I detect and select the largest supported complex dtype in NumPy?

You can check the “true” precision of a complex dtype by inspecting its itemsize. In NumPy’s modern type names, a type such as np.complex256 should have an itemsize of 32 bytes (i.e. 256 bits). If the dtype is a dummy (substituted fallback), its itemsize will be lower than expected. For example, you can write a helper function like this:

PYTHONCopied
import numpy as np

def supports_dtype(dtype, expected_bits):
    # np.dtype(dtype).itemsize returns the size in bytes.
    return np.dtype(dtype).itemsize * 8 == expected_bits

# Here are the candidate dtypes with their expected total bit sizes.
complex_candidates = [
    (np.complex512, 512),
    (np.complex256, 256),
    (np.complex192, 192),
    (np.complex128, 128)
]

def get_largest_supported_complex():
    for dt, bits in complex_candidates:
        if supports_dtype(dt, bits):
            return dt
    return None

# Example usage:
supported = get_largest_supported_complex()
if supported is not None:
    print("Largest supported complex type:", supported)
else:
    print("No supported complex dtype found!")

This approach directly checks whether the dtype’s itemsize (multiplied by 8 to convert to bits) matches the advertised precision. If it doesn’t, then that complex dtype is essentially a dummy type. This avoids the less clean approach of trying to instantiate an array and catching exceptions.