Commutativity conditions for the natural quaternion convolution in 1D

Determine non-trivial necessary and sufficient conditions on quaternion-valued functions f and g on the real line such that the natural quaternion convolution ostar, defined by Akila and Roopkumar for one-dimensional quaternion Fourier analysis, is commutative; specifically, characterize when f ostar g = g ostar f holds beyond the known trivial cases (both functions complex-valued or one function real-valued and even).

Background

In the one-dimensional quaternion Fourier transform setting, Akila and Roopkumar introduced a natural convolution operation (denoted ostar) tailored to quaternion-valued signals and proved a convolution theorem for it.

Due to the non-commutativity of quaternion multiplication, this convolution is generally non-commutative. The authors note that commutativity can be guaranteed in simple cases, such as when both functions are complex-valued or when one of the functions is a real-valued even function.

The survey explicitly highlights the unresolved task of identifying a precise and comprehensive characterization—necessary and sufficient conditions—under which the ostar convolution becomes commutative, thereby advancing the algebraic understanding of quaternion signal processing.

References

Finding a suitable non-trivial necessary and sufficient condition on f and g to get f\star g=g\star f is still and interesting open problem.

A mathematical survey on Fourier type integral transform and their offshoots: windowed Fourier transform, wavelet transform and Stockwell transform  (2402.06645 - Gupta et al., 2024) in Section “One dimensional quaternion Fourier transform” within “Quaternion Fourier Transform (QFT)”