radix4_fft is not radix4

[Stenzel]

looking at the code of radix4_fft I fail to see how this is radix4.

[muditbhargava66]

You are absolutely correct.

Upon reviewing algorithms/core/radix4.c, I confirmed that the current implementation is indeed a Radix-2 algorithm disguised as Radix-4.

Lines 108-110 explicitly state:

// Use radix-2 butterflies for all stages (reliable approach)
    // This ensures correctness while still providing the radix-4 interface

The loop iterates log2n times (Radix-2) instead of log4n times (Radix-4), and it uses standard Radix-2 butterflies.

This appears to be a placeholder implementation that was never updated to the actual Radix-4 algorithm, despite the documentation and file naming claiming otherwise. This explains why it doesn't show the expected performance characteristics of a true Radix-4 implementation (25% fewer multiplications).

I will mark this as a bug/enhancement to implement the actual Radix-4 algorithm.