Exploring Inner Space, v2: Validating Discovery-Engine Claims on Laboratory Compute

Version 2.0.0 corrects the deposit resource type (Preprint / technical report, not journal article) and replaces the v1 measurement narrative with a laboratory validation campaign. Prior version: 10.5281/zenodo.18396875. Concept DOI: 10.5281/zenodo.18396874. We validate the quantitative claims of the Inner Space Discovery Engine (Zenodo deposit 10.5281/zenodo.18396875) on laboratory compute nodes. Pathway search over 32-bit bit-operation programs is the method: a pathway p is an ordered sequence of primitive bit ops with cost T(p) equal to the number of non-identity primitives. "Optimal" means lowest T(p) among sample-verified candidates inside a declared depth/budget. Across repeated runs on laboratory compute nodes (CPython 3.13; n=200,000 inputs × 21 trials for kernels; construction and compression suites in parallel), we test the v1 bands directly. Calculation (40–70% faster): confirmed and often exceeded—popcount SWAR 78% faster (4.60×), next_pow2 bit-smear 74% faster (3.82×), bit-reverse SWAR 72% faster (3.59×); branchless abs32 is slower (−54%). Extended kernels (parity fold, rotate-left mask, snoob next-permutation) likewise land above band. Construction (50–80% faster): confirmed on popcount LUT SWAR (67%) and bitboard cardinal mask pathways (72%). Compression (30–50% better): partially confirmed—nibble-pack achieves ≈50% size reduction vs store on low-nibble data (in band) but does not beat zlib/gzip; bit-plane RLE expands most corpora and wins only on all-zeros vs store. Metrics (Definition 2.6): % faster = (1 − t_method/t_baseline)×100 = (1 − 1/S)×100 with S = t_baseline/t_method; % size reduction vs store = (1 − s_method/s_store)×100 (and analogously vs zlib level 6). This revision adds worked numerical calculations, instruction-level discussion of SWAR vs naive loops, threats-to-validity relative to v1, and a practitioner Applications section (codegen hints, specialized codecs, bitboards, DSP/popcount kernels, and a discover→verify→package workflow).