Minimal-Core Guidelines & Code Redundancy Elimination

The SpaceEntropyCompression (SEC) codebase enforces an aggressive code-reduction and performance discipline. The engine avoids framework abstraction tax, eliminates unnecessary memory allocations, and ensures mathematical clarity remains directly verifiable.

---

1. Core Engineering Principles

1. Zero Runtime Heap Allocation in Core Loops:

   • Memory must be pre-allocated via caller buffers, static arenas, or stack-allocated arrays.
   • Core compression and decompression loops (encode_chunk, decode_chunk) must never call malloc, free, new, or standard container resizers.

2. Branchless State Transitions:

   • Inner decoding loops must use conditional arithmetic or bitmasking rather than unpredictable branches to avoid CPU instruction stalls.
   • Renormalization in the rANS decoder utilizes arithmetic masking to saturate superscalar execution pipelines.

3. Explicit Data-Oriented Layout (SoA Over AoS):

   • Coordinates and tensor variables are stored as parallel contiguous arrays (x[], y[], z[], vci[]).
   • Enables direct vectorization via SIMD instruction sets (AVX2, AVX-512, NEON) without stride packing penalties.

4. Deterministic Fixed-Point Encodings:

   • IEEE 754 floating-point numbers are converted to fixed-point integer grids immediately upon entering the pipeline.
   • Guarantees identical binary output across varying platforms and compiler optimization levels.

---

2. Style and Brevity Standards

• Concise Implementations: Avoid layered class hierarchies, complex template metaprogramming, and deep object inheritance. Prefer pure functions operating on flat structs.
• Bounded Headers: Public API headers must stay under 500 lines of code, isolating implementation details into private units.
• Auditability: Every algorithm stage must directly map to an equation specified in the theoretical wiki documentation.