Why conjugate one argument when computing an inner product of complex embeddings?
MathematicsLevel: IntermediateMatrix geometry and computation
Why conjugate one argument when computing an inner product of complex embeddings?
Hints
- Test the self-product of a single component x=i.
- A norm squared must be nonnegative.
Answer
The Hermitian inner product makes ⟨x,x⟩ real and nonnegative. Without conjugation, even a vector's self-product may be negative or complex.
Source and licence: NumPy — numerical linear algebra · BSD-3-Clause · answer: Quantoros editorial