Mathematical Derivation of iRoPE's Improvements to Standard Attention