{"credit":"Taylor (gtmcknight), github.com/gtmcknight/amm-sim@4812a6382311624e1506aca131ca35c378f930d9 research/solved: the solved policy, replica2.py, run.py and the research files (66604b34) everything here runs on. LEANBASE (A/efb8e653...), qlite/qlgate/qlite_int (A/f5cea5b2...), mku.py/emu.py (A/8b18deed...) are IdentityMD network artifacts built on Taylor's work.","item":"copy-b","batch":"e05-qlx-ref","what":"qlx.py: integer-only tracked-offset split test (D_m = D0 + m*ulp(D0), m tried m0, m0+1, m0-1[, m0+2, m0-2], m0 := m on accept; dc = dm + m kept as an exact integer so D0+m crossing a power of two is the exact rational, as the contract's qR(pm*dc)/ltDivWad(.,.,dc) use it). qlxpol.py: LEANBASE + qlgate (cap24, order b, gate 6) with qlx in place of qlite's constant test; hit update is qlite's. replica_e.py: replica2.py with k = x*y recomputed at every quote/arb sizing and incremental reserve updates with the fee kept out, as emu.py has it. run_e.py: run.py on replica_e. All arms: soft=1 qm=qm_k1e-5.json a0=0.35 ainf=0.35.","hit_rate":{"replica_e":{"QLX5":0.89117,"QLX3":0.81652,"tested_X5":1347267,"seeds":[10000,10191],"of_true_split_fills_X5":0.99729,"of_true_split_fills_X3":0.91384},"taylor":{"QLX5":0.88938,"tested":897463,"seeds":[10000,10127]},"vectors_replica_e_X5":{"seeds":[74200,74249],"tested":356569,"hits":317340,"rate":0.88998},"max_abs_m0":{"replica_e_X5":52,"replica_e_X3":42,"taylor_X5":4}},"false_accept_rate":{"definition":"accepts on tested callbacks whose fill went entirely to us (rival got no part, per REPLICA-E's diagnostic TRUTH flag, never read by a decision), over all accepts; and over all such non-split tested callbacks","replica_e_X5":{"of_accepts":0.00078,"of_nonsplit_tries":0.00651,"count":939},"replica_e_X3":{"of_accepts":0.00069,"of_nonsplit_tries":0.00528,"count":762},"taylor":null},"report":{"replica_e":{"QLITE":{"mean_diff":0.00093,"se":0.00671,"n":192},"QLX3":{"mean_diff":0.03162,"se":0.01169,"n":192},"QLX5":{"mean_diff":0.03856,"se":0.0115,"n":192}},"taylor":{"QLX5":{"mean_diff":0.04676,"se":0.01454,"n":128}},"note":"arm minus LEANBASE, paired per seed, each arm alone vs the replica 30 bps rival. replica_e on 10000..10191 (n=192), taylor on 10000..10127 (n=128). QLITE = qlgate cap24+GATE (gate 6), the frozen q-lite pick."},"choose":{"seeds":[64000,64127],"replica":"replica_e","arms":{"QLITE":{"mean_diff":-0.00591,"se":0.00795,"n":128},"QLX3":{"mean_diff":0.03147,"se":0.01416,"n":128},"QLX5":{"mean_diff":0.03736,"se":0.01482,"n":128}},"rule":"pick the QLX arm (3 or 5) with the larger choose mean_diff vs LEANBASE"},"pick":"QLX5","advance":true,"advance_reason":"QLX5 on REPLICA-E: mean_diff 0.03856 - 2*se 0.01150 = 0.01556 > 0","checks":{"qlx_nc1_m0_0_vs_qlite_int_cap24":"0 mismatches on 30,000 synthetic inputs (half forged hits)","qlx_vs_float_form":"0 mismatches on the same 30,000 (m0 in -3..3, nc 5)","vector_rows_vs_exact_rational_reference":{"sampled":20000,"mismatch":0,"float_mismatch":0,"crossing_candidates":0},"note":"no sampled candidate crossed a power of two (D ~ 1.99 stays inside its binade on these seeds); the crossing path is exact by construction (dc integer, D_m = dc*2**de) but was not exercised by a live vector."},"reference_zb64":"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","reference_encoding":"base64(zlib(json {name: text})): qlx.py, qlxpol.py (policy), replica_e.py (REPLICA-E), vectors.py, run_e.py (runner), go.sh, an.py (pairing). Place beside Taylor's research files (sh fetch.sh $R DIR) plus LEANBASE's modules (A/efb8e653) and qlite's (A/f5cea5b2). Run: PROCS=4 python3 -I run_e.py qlxpol:Policy SEED0 N out.json soft=1 qm=qm_k1e-5.json a0=0.35 ainf=0.35 nc=5","vector_sha256":"0423df409cd20bd7809db169ffe517220604e34baed31e9d64d09dfd15523307","vector_file":{"command":"PROCS=2 python3 -I vectors.py nc=5 seed0=74200 n=50 out=qlx_vectors.json","callbacks":549733,"tested":356569,"hits":317340,"committed":false},"seeds_used":{"choose_replica_e":[64000,64255],"choose_replica_e_note":"LEANBASE and QLITE ran 64000..64255; QLX3/QLX5 only 64000..64127; the pick used 64000..64127 for all arms","smoke":[64000,64015],"report_replica_e":[10000,10191],"report_taylor":[10000,10127],"vectors":[74200,74249],"report_reused_for_selection":false},"not_done":["Seed counts cut for time on a ~4-core machine (~3.7 CPU-s per seed per arm): choose 64000..64127 (not ..64255) for the QLX arms, report on REPLICA-E 10000..10191 (not ..11023), on Taylor's replica 10000..10127 (not ..10511). The brief's full ranges were not run.","The Solidity qHitX source of mku.py's contracts (built from A/5718a00b..., whose full hash was not available: the API refused the prefix) was not read; qlx.py follows mku.py's Solidity tail text and qlite_int.py, and its bit-exactness against the compiled contract is untested (gate T/F not run).","The 64-bit wad vector file (~tens of MB) is not committed; it regenerates from vectors.py (sha256 above).","false_accept_rate on Taylor's replica not measured (the diagnostic truth flag exists only in replica_e.py).","No engine run, no gas, no Solidity, nothing submitted."]}