Research-Stack/4-Infrastructure/shim/online_domain_eigen_pruning.json
2026-05-11 22:18:31 -05:00

290 lines
7.3 KiB
JSON

{
"schema": "online_domain_eigen_pruning_v1",
"claim_boundary": "Leading eigenvector is a ranking prior over adjacent source-backed domains, not proof of correctness.",
"domain_count": 7,
"weighted_domains": [
{
"domain": "arithmetic_entropy_coding",
"equation": "message -> interval with subinterval widths proportional to symbol probabilities",
"role": "baseline entropy coder and probability weighting model",
"source": "Youssef, Parallel Algorithms for Entropy-Coding Techniques, NIST, 1998",
"url": "https://www.nist.gov/publications/parallel-algorithms-entropy-coding-techniques",
"eigen_weight": 0.32818448514877907
},
{
"domain": "asymmetric_numeral_systems",
"equation": "state-machine entropy coding with symbol-state allocation f_s ~= p_s R",
"role": "finite-state entropy coding, table/LUT-adjacent for hardware",
"source": "Pieprzyk et al., The Compression Optimality of Asymmetric Numeral Systems, Entropy, 2023",
"url": "https://www.mdpi.com/1099-4300/25/4/672",
"eigen_weight": 0.2598079624847316
},
{
"domain": "minimum_description_length",
"equation": "argmin_M L(M) + L(D | M)",
"role": "model selection prior; choose the shortest lawful template before encoding",
"source": "Grunwald, Model Selection Based on Minimum Description Length, Journal of Mathematical Psychology, 2000",
"url": "https://www.sciencedirect.com/science/article/abs/pii/S0022249699912804",
"eigen_weight": 0.19256323124324387
},
{
"domain": "bounce_lightweight_integer_compression",
"equation": "compress k separate blocks of size N across SIMD lanes to preserve scalar ratio",
"role": "partitioned lane layout prior for avoiding wide-register ratio loss",
"source": "Bittner et al., BOUNCE: memory-efficient SIMD approach for lightweight integer compression, Distributed and Parallel Databases, 2023",
"url": "https://link.springer.com/article/10.1007/s10619-023-07426-0",
"eigen_weight": 0.10931260799338613
},
{
"domain": "simd_bp128_bitpacking",
"equation": "block_width = ceil(log2(max(block)+1)); pack N integers at block_width bits",
"role": "lane-width prior for GPU/FPGA integer surfaces",
"source": "Lemire and Boytsov, Decoding billions of integers per second through vectorization, Software: Practice and Experience, 2015",
"url": "https://arxiv.org/abs/1209.2137",
"eigen_weight": 0.057946842152125255
},
{
"domain": "delta_sigma_one_bit",
"equation": "b_t = Q(v_t + e_{t-1}); e_t = v_t + e_{t-1} - b_t",
"role": "1-bit residual-feedback transport prior, adjacent to PBACS",
"source": "Zierhofer, Adaptive Delta-Sigma Modulation for Enhanced Input Dynamic Range, EURASIP JASP, 2008",
"url": "https://link.springer.com/article/10.1155/2008/439203",
"eigen_weight": 0.03805214750155165
},
{
"domain": "normalized_compression_distance",
"equation": "NCD_Z(x,y) = (Z(xy) - min(Z(x), Z(y))) / max(Z(x), Z(y))",
"role": "compressor-backed similarity gate for choosing nearby templates",
"source": "Cilibrasi and Vitanyi, Clustering by Compression, IEEE Transactions on Information Theory, 2005",
"url": "https://ir.cwi.nl/pub/16389",
"eigen_weight": 0.014132723476182319
}
],
"top_terms": [
{
"term": "entropy",
"weight": 0.2063358047091217
},
{
"term": "model",
"weight": 0.13192504349539264
},
{
"term": "selection",
"weight": 0.0837713577517334
},
{
"term": "parallel",
"weight": 0.07848924644513405
},
{
"term": "algorithms",
"weight": 0.07515803351452537
},
{
"term": "arithmetic_entropy_coding",
"weight": 0.07515803351452537
},
{
"term": "baseline",
"weight": 0.07515803351452537
},
{
"term": "coder",
"weight": 0.07515803351452537
},
{
"term": "entropy-coding",
"weight": 0.07515803351452537
},
{
"term": "interval",
"weight": 0.07515803351452537
},
{
"term": "message",
"weight": 0.07515803351452537
},
{
"term": "nist",
"weight": 0.07515803351452537
},
{
"term": "probabilities",
"weight": 0.07515803351452537
},
{
"term": "probability",
"weight": 0.07515803351452537
},
{
"term": "proportional",
"weight": 0.07515803351452537
},
{
"term": "subinterval",
"weight": 0.07515803351452537
},
{
"term": "symbol",
"weight": 0.07515803351452537
},
{
"term": "techniques",
"weight": 0.07515803351452537
},
{
"term": "weighting",
"weight": 0.07515803351452537
},
{
"term": "widths",
"weight": 0.07515803351452537
},
{
"term": "youssef",
"weight": 0.07515803351452537
},
{
"term": "allocation",
"weight": 0.05780453403001633
},
{
"term": "asymmetric",
"weight": 0.05780453403001633
},
{
"term": "asymmetric_numeral_systems",
"weight": 0.05780453403001633
},
{
"term": "f_s",
"weight": 0.05780453403001633
},
{
"term": "finite-state",
"weight": 0.05780453403001633
},
{
"term": "hardware",
"weight": 0.05780453403001633
},
{
"term": "lut-adjacent",
"weight": 0.05780453403001633
},
{
"term": "numeral",
"weight": 0.05780453403001633
},
{
"term": "optimality",
"weight": 0.05780453403001633
},
{
"term": "p_s",
"weight": 0.05780453403001633
},
{
"term": "pieprzyk",
"weight": 0.05780453403001633
},
{
"term": "state-machine",
"weight": 0.05780453403001633
},
{
"term": "symbol-state",
"weight": 0.05780453403001633
},
{
"term": "table",
"weight": 0.05780453403001633
},
{
"term": "prior",
"weight": 0.04862003891674652
},
{
"term": "argmin_m",
"weight": 0.0418856788758667
},
{
"term": "before",
"weight": 0.0418856788758667
},
{
"term": "choose",
"weight": 0.0418856788758667
},
{
"term": "description",
"weight": 0.0418856788758667
}
],
"domain_similarity_matrix": [
[
0.9999999999999997,
0.06864763884530972,
0.0,
0.01513151045105087,
0.014647293359011769,
0.015228563066514943,
0.0
],
[
0.06864763884530972,
1.0,
0.10532564245617808,
0.0,
0.028001404572565464,
0.0,
0.0
],
[
0.0,
0.10532564245617808,
0.9999999999999998,
0.0,
0.0,
0.0,
0.0
],
[
0.01513151045105087,
0.0,
0.0,
1.0000000000000002,
0.034758527243063664,
0.01283417735563,
0.031889604696206345
],
[
0.014647293359011769,
0.028001404572565464,
0.0,
0.034758527243063664,
1.0000000000000002,
0.01242347625226291,
0.0
],
[
0.015228563066514943,
0.0,
0.0,
0.01283417735563,
0.01242347625226291,
1.0,
0.0
],
[
0.0,
0.0,
0.0,
0.031889604696206345,
0.0,
0.0,
1.0
]
]
}