{"key": "fresh8-2", "track": "B", "status": "completed; null result, no promotion", "mechanism": {"name": "Silent-step no-fill likelihood (router-implied retail absence)", "causal_basis": "A step with no own trade means two things. The base uses only one: the arb did not hit us, so p is inside our no-arb interval (the cut). It ignores the other: no retail order reached us. The chance of that depends on p through the replica's router. Each step the 30 bps rival is first arbed to clip(s2, pG, p/G), and a lognormal retail order fills us only if its size is above a router threshold set by our reserves and fees and by the rival's p-dependent spot. So on every silent step each lattice cell i is weighted by h_i^k.", "formula": "h_i = exp(-LAM*q_i); q_i = 0.5*P(Q>Qbuy_i) + 0.5*P(Q>p_i*Xsell_i); s2_i=clip(s2h,p_i*G,p_i/G); y2=sqrt(K*s2_i), x2=K/y2; rb=sqrt(x1*(1-fa)*y1)/sqrt(G*K), rs=sqrt(y1*(1-fb)*x1)/sqrt(G*K); Qbuy=(y1/rb-y2)/G; Xsell=(x1/rs-x2)/G; P(Q>q)=0.5*erfc((ln q-MU)/(1.2*sqrt2)), MU=ln20-0.72; LAM=0.8; G=0.997; K=1e6. x1,y1 = reserveX/Y of our last afterSwap, fa/fb = fees in force, s2h = the base's rival estimate. Applied after the cut, before the norm, only inside the base's silent-step loop. Not applied to the first-trade retail branch.", "distinct_from_exclusions": "It does not change the arb/retail classifier, threshold, likelihood of the trade itself, trade-size reading, lattice geometry/recenter, rival transition or label, or the quote choice/blend. It adds a likelihood factor for an event the base treats as uninformative beyond the arb cut: the absence of any retail fill over a silent step. No per-run fit; the constants are the replica generator's midpoints (order rate U(0.6,1.0) -> 0.8, size mean U(19,21) -> 20, log-sd 1.2). It is related to, but not the same as, the README's per-hypothesis router likelihood of the observed trade: it uses only steps with no trade.", "base": "thrpol:ThrLattice thr=.9 qm=qm_k1e-5.json, unchanged (thrpol.py from artifact c9d678ab..., shipped in the tar)"}, "settings_tried": [1, 2, 4, 8], "frozen_settings": {"k": 1.0, "thr": 0.9, "qm": "qm_k1e-5.json", "LAM": 0.8, "size_mean": 20.0, "size_logsd": 1.2}, "selection_rule": "Highest paired mean over unchanged base on 65000..65255; frozen before confirmation; no retuning after confirmation or holdout.", "tuning_vs_base_65000_65255": {"1": {"n": 256, "mean_diff": 0.013644388604753033, "se": 0.008050310207106661, "sum": 3.4929634828167764, "sumsq": 4.278292587574316, "worst5": [[65050, -0.3842], [65157, -0.3293], [65242, -0.3264], [65154, -0.2926], [65035, -0.2877]], "seeds": [65000, 65255]}, "2": {"n": 256, "mean_diff": 0.007813296989301666, "se": 0.010444082061234256, "sum": 2.0002040292612264, "sumsq": 7.136295522765004, "worst5": [[65254, -0.5114], [65115, -0.4556], [65096, -0.3931], [65112, -0.3699], [65163, -0.3544]], "seeds": [65000, 65255]}, "4": {"n": 256, "mean_diff": -0.027089814617428987, "se": 0.013814397348216368, "sum": -6.934992542061821, "sumsq": 12.645744499220623, "worst5": [[65166, -0.7112], [65080, -0.7106], [65163, -0.591], [65207, -0.5555], [65038, -0.5279]], "seeds": [65000, 65255]}, "8": {"n": 256, "mean_diff": -0.1466783644822165, "se": 0.020272246548899493, "sum": -37.54966130744742, "sumsq": 32.33545153093545, "worst5": [[65141, -1.2771], [65166, -1.2246], [65088, -1.1441], [65171, -1.0074], [65038, -0.9853]], "seeds": [65000, 65255]}}, "confirmation_vs_base_65256_65511": {"n": 256, "mean_diff": -0.02209438908094774, "se": 0.009046881161777234, "sum": -5.656163604722622, "sumsq": 5.467880194935513, "worst5": [[65296, -0.7221], [65322, -0.5794], [65293, -0.467], [65256, -0.3411], [65321, -0.3403]], "seeds": [65256, 65511]}, "holdout": {"seed_range": [10000, 12047], "n_per_arm": 2048, "passes_per_arm": 1, "runner": "unchanged run.py (run.Blind) in chunks: candidate 10000+768, 10768+1280; base 10000+256, 10256+896, 11152+448, 11600+448; joined per seed, statistics recomputed with sum/sumsq. A first base chunk 10256+1536 timed out at 480 s and wrote nothing; it was rerun as the 896/448/448 chunks. No seed missing.", "candidate_vs_fresh_base": {"n": 2048, "mean_diff": -0.0027610489841786157, "se": 0.003349988251142319, "sum": -5.654628319597805, "sumsq": 47.062875663088406, "worst5": [[10822, -1.3644], [11950, -1.3307], [11280, -1.3186], [10236, -0.5763], [10163, -0.5404]], "seeds": [10000, 12047]}, "candidate_vs_ref_lattice_qm": {"n": 2048, "mean_diff": 0.3201308708263126, "se": 0.010881720081813609, "sum": 655.6280234522882, "sumsq": 706.2994830029514, "worst5": [[11487, -1.8472], [10822, -1.4757], [11950, -1.3705], [10942, -1.3508], [10982, -1.3182]], "seeds": [10000, 12047]}, "fresh_base_vs_ref_lattice_qm": {"n": 2048, "mean_diff": 0.3228919198104912, "se": 0.01069159987528695, "sum": 661.282651771886, "sumsq": 692.7408990952625, "worst5": [[11487, -1.9115], [10942, -1.6077], [10982, -1.4458], [11380, -1.2365], [10990, -1.1505]], "seeds": [10000, 12047]}, "decomposition": "candidate - lattice_qm = 0.32013 = inherited threshold gain 0.32289 (fresh base - lattice_qm, matching the brief's +.322892 to 6 digits) + incremental -0.00276 se 0.00335 (not significant, n=2048)."}, "first_trade_misread_and_price_rms_by_arm_holdout": {"candidate": {"n": 2048, "mean_edge": 534.1410552449877, "first_trades": 14474076, "misread": 4182956, "misread_rate": 0.2889964098571819, "callbacks": 22640349, "price_rms_bps": 13.279331280400893}, "base_thr09": {"n": 2048, "mean_edge": 534.1438162939719, "first_trades": 14472181, "misread": 4180058, "misread_rate": 0.28883400504733875, "callbacks": 22636622, "price_rms_bps": 13.277152756808208}, "lattice_qm": {"n": 2048, "mean_edge": 533.8209243741613, "first_trades": 14315685, "misread": 2322695, "misread_rate": 0.162248261260289, "callbacks": 22263118, "price_rms_bps": 13.64283973652866}}, "checks": {"disabled_change_equivalence": {"k": 0, "seeds": "65000..65007", "all_row_fields_identical_to_base": true}, "hidden_input_scramble_and_replay": [{"seed": 65000, "callbacks": 12208, "replay_identical": true, "scramble_identical": true, "blind_edge_equal": true, "edge": 605.8994504400474}, {"seed": 65001, "callbacks": 9314, "replay_identical": true, "scramble_identical": true, "blind_edge_equal": true, "edge": 398.7987908754069}], "notes": "Replay: the afterSwap tuples recorded from a replica run, fed alone into a fresh NoFill(k=1), reproduce every returned (bid, ask) fee exactly. Scramble: the arb flag, p, s2 and j were randomized in the wrapper. The policy never receives them, and the fees and edge are bit-identical. Edge through run.Blind is equal as well."}, "approximation_errors": ["The rival spot per cell uses one clip of the static s2h on every silent step instead of a path-dependent rival state; retail orders that went only to the rival (moving s2) are ignored.", "Order rate, size mean and router are at generator midpoints; the 1e-4 minimum-fill and the price drift are ignored; rho, sigma and the lattice are unchanged from the base.", "Orders after our last trade in a step (also unfilled) are not used.", "Weight variation is small: on seed 65000, k=1, max/min cell weight ratio over a silent step: median 1.0003, p90 1.159, p99 1.285; 34% of silent steps above 1.05."], "gas_estimate": {"label": "ESTIMATE, not measured; no Solidity written", "incremental_slots": "2 (own reserves x1,y1 of the last afterSwap; one slot if packed); fees and s2h already stored", "incremental_ops": "per silent step (capped at 30), per active lattice cell (about 20 to 60): 1 exp for p_i, 1 to 2 sqrt, 2 log+erfc tail evaluations, 1 exp for the weight. Cells inside the rival's unclipped band share one cached value.", "gas": "roughly 1,500 to 3,000 gas per cell per silent step with fixed-point exp/log/erfc approximations. A gap of several silent steps would need well over 245,000 gas unless tabulated or capped, so it does not fit a lattice contract as written. Irrelevant given the null result."}, "limitations": ["Null result: +0.0136 se 0.0081 in tuning, -0.0221 se 0.0090 in confirmation, -0.0028 se 0.0034 on holdout. The mechanism adds nothing measurable over the base, and misread (28.90% vs 28.88%) and price RMS (13.279 vs 13.277 bps) do not improve.", "Why it is weak: the factor varies with p only where the rival gets band-clipped (p more than about 30 bps from s2h). The lattice mass is mostly within about 20 bps, so the likelihood is close to flat over the cells that matter.", "No numba/accelerated path; plain Python only (about 1.1 s per run).", "Replica only; no engine run, no board entry."], "recommendation": "Do not port. Not worth passing to Track A (gain << 0.3).", "code_files": ["nofill.py", "thrpol.py", "diag.py", "final.py"], "code_note": "tar (USTAR, zlib, base64): nofill.py (the mechanism, put beside run.py), thrpol.py (unchanged base), diag.py (weight-variation diagnostic), final.py (pairing/sum/sumsq, misread/RMS, replay+scramble checks, packaging). Reproduce: python3 -I run.py nofill:NoFill 10000 2048 out.json qm=qm_k1e-5.json thr=.9 k=1 (in chunks).", "code_sha256": "5b9b097f2d7a2f5e0f06c44ab24f71b8f3713d8e3acd4616ffc73f6aeae908ca", "code_zb64": "eNrtW+1z2zaT72f+Fbhm5kSmNC0pdvpYjjLTpNOkk6R5cZ6Z56nGp6FEyEJEkTRB2VJz+d/vtwvwRRKdpNdr70PFiSMKWCwWu4t9A5SkMxXHQbb55s97ungenpzwJ56dz16v2zst22z79w/xIbrf/AXPShdhjim/+Xs+33777YWKZVIc6UJmIkmPSB9ErBYyVvM0jUSaiCLNRDoTxVyKVTKdh8mVjPAtz9J48H6evwyLQk2lcNE0DM58cb0cL3ry6DT4oNPECxznlaRRSi8DIX4QPJNKxO1cTeciXeUiS9NYFHkYSQ0SChHiH6jQ4UaL4jbFXCq50hj8HiRMQi0FqCxkLlYaI9Ik3jBxM5XrYuDQa5hPRKQixsaIxa0q5oD3hU5FJuJwAxK0QgcRgHXTCJUA6U0Y01KkmK4KD3P+XAh1laQ5ZqJWLadpEg0wROSyCFXspHkEUnIZTufgy0oLialkjqHP01vDSqIPi1JaRDKTSUREi8wX03ClsdKNT+xMV1dzngJvoGNAy9aGW9yqQJnjztQaszzoikmmPcLIq6YFk1BSMY1V5uq+L7Jn+Dt+5vkiTCIRWmqFoRbi1Oo3Kd4KN05pdcsw9sW/IV9VAO2VusFysZTQYX1gFt/OZYIBcj2VEisILZ1EutTzNI7Am0JMNsxRNIGVQEKTz6R9mWzqpXS0cLMjww8ooCd0lhaBc5EK4qSYSsyrSr2LrY5hvbdSXc0LGfnEQ3kj8w2WQirMrPIxhyPwzMdKDEFs5h69/OGVuC+uxwq8oA+094774o37VjwWbyerDfiFTk9812jPAPcvDSJsJ4bSC8ZaDs99hnlmPo7FM8/JQ6WNFEIQk8vkCjq3EO4Co7pEe1Hqr1yH0yLekH49TRPYoKQAh3Jp2COzWE1DMChbTWKlSa2uZCLzsEhzsVRRlkJTteNazQsLKbrBP3wj0qUME9Hv+gKCPdKR6AV9zzeqGy6xSYDCcC1fJee0QWhzGUUqwFaREcpVEjg/J9mq0INS+LQpZ5D3xW2YiavwBosI81gB2q2EDWmR8ONQ223nG9ljt2PeqfRr8QupC7Uk0sFKMhLYCVdyILJNMU+TB+LoZ6Yi24BE0sHBL+lPpIqPNLSv+1g8SvAHBWQj8xg2Z7hldoQxRmIx7Dkwco5aZmleCEw4L98tm/si1OKdM8vTJauZsN3Pn/ril65px+aOy/ZnfdNmDGDZWttBx3kBcb8LXjikd0OSjPPqn+OXv+Cdpg8gF7ffDbqeOELnKVQTIjL/Oxc/GkB+f/uiHKKv88J9RkAvPFE996odKPOlYJh1X1z1xabvlRuHeT0Qa2oEthds/1QhotTaWrMBaS/pvuM4kZyJjAyFe+0NeCOBe2+sprF+gdcecZT61Exci0e0xu6ADMwqT0B6l/vsV7NCXoXMZ1PXrXhwTQxgznjYPq5ZOrHhpHfS7z04fdh/8P2D7tlpz/NA2BRKpYVRArfmtiWSyB6PFazXeOxi18588R4GIiuGv6RJCURPPTJohfcqSGoN1jkJo4v1nZuGjW2gpkoM6W3SMHi0SeptMIV9n4RTWIF691jYfx3/e2c6MhSzOA0LpilIgytZuJ1FxycOezvAcF8xG5bPPPeatrG2ncKNVAizjy04BXNLDpqdNjZQTEKDc5D5chUXKrPeFu6xMsuw629ceETDCPYz7Df+W2Tefy3Y5kAMTVrOrX4Ym7iQGe1NrUvFoucNFscLfXMupuX79FwsyvdFBbru+WLTK9vXuV8Kq4KAd2/upXUPmuZCVaGCDDoLPbRseqSKtPEs9+AcBEcJDUx6C9NmD9OEMK13MZEr2UUFy1fSjNeqOU5pq1LnfVibczFX5Xd4mf5dgoZaUWRGbrr0I2xnldnn5LUyGVWTTClcAVraH1VjAUDezVULiU6R/WbkRi3jdKwsh+cKvg9us6En9GRqDDxvRupyqxnmgnuswUAgVahkJbeHltwl1z0FbleBs790ianPn3q7+IhXMEAZ/Qc+bZNhYcxSlea17kPQw8axFuoLTEZu6ZwMJ6wmGL/ptw68Jr1ySQuOoWNk0cj4kpxawdeawdcMrgl8/TnwUkju9cTHWM/Zn943OBlyq1vGumWxmtaTsWIRa0ipHpXfaQTeW6lpZdAX+PM7ePOVfLlm7SSHYn3UhIK2ZgstZZdRpHP3G1p1tACQDQq3IUljiT9qfM574Tt+bzG76OjVFsF4OgyoTSmso/UrcepDMxsbxGyyhiPag3Uaystzjq3lpBjaOor/4E3ELYTwsUDo86Db3Za4mcqQvWXbvTuJL6caip9C6EO9ojQZc0hnSVV6DOOI5GKNP0pgoIgwwgj44vBq1+veqzI3BLW6kXDGKXJLcpGaOWbXj7ngVCgQZSBO3ax3sVFlQbkIeSCGXCI7Y0tVNPbAvdITTWC6KNGc1UmiQYPg+3WZOzaJMvTA9XJPmRMbzgUV/oh4VJRWvyBnv60o46Ri5TJcu4jIlypxIxoCd/UA/pzki++PoUxm89V2F4HdtpL8HvbvBAq7It2JcPxGZEM4dlUDpDRDhHxpSYDKbImY2IbQIic22SiIM7kt1q4KOPN0lTBvlwhDIVOd2rSHxHVkxUXChcmO5YyKAU0nlW8zZoeiu/bO9sbYFtBRczd/kW2WLXnNlEjNZistdyOmBpFbEA2eheVYqG8cp7eSc+TQrN7UJ2rt5N0yQJq45EajkFbvkZ2Z7RBv7hI+FBrU5JZXnXL1FFx6sB9d55vD83d7TBL7/1n/7SLH6+/Vfx/0DvXfv6j++xRZ4FGVGIb55Nj6TbLng63qm626wpNSC7TGrwvCPhcGs8LYoh1PS0NCduYObHouw4iKLmWhtq4uqkKLDPTIXMGqUV+Wp5NwomJVbBCKNYqPxSoxVWhbe8SbcC1Zpi6MsNALHA48ZKxgpldZRPUmUAD/k66oXHteV4+zORXm3K640chnTGBVJlFHVJssx/MCjbk2iTDVqg0XgrurWPv18t9TycJa/pdlLDvbmxRAkNfzp/sVrWyOdO5Z3xfvnr/2xcsfx6/4/4uq6lIT7W6haxRf/kh0mJYxalq7LjRtEP+vvyrcwmwhJ/gt8Y3RwmEZZzXDA9vUEhlM2WF2tzritFnAM0l7tp/7e5Svt0Ee79YbtlOPlMoa7/OdbJic+7hOv3cDyJb8jpCX0cZ+vpgu6nzAhk1BI/D22tJnqh6kiwHiaBkudvNv6vjMnHFmJVOzo4VlhMhojAmDG8Bf4FpNnrNfEbHbPlJU8FqmkYwRgWsgn849KleS6S+QXURkhhCTRpVFIgulkqt27mYQglkXZJ3xaULQ/YrUO7asMOySiS6x7EFSqW4obPWPbLEtALZJh2Afcz2ltYKwXPri5qacdZkuEfHpFsVoEmg5LumA6b5t/Y4nuk+2wrV0A2S5pCT9pCuPTqqXPdQRUIYT7W5J1e6KFo3bydtgS5uR8M5J3d7oNwsps6pwOBpcngtTruJKlW3fKWDtIfm1qjbWW6NFnKATgL+CK1ZU+TztsBX1Sk5Va6YMED4SVv3Mo/IGWVliGdlZLn83Aag8Qe17c7K2GYFgSCkZpoQ30W4TFMPsqF0NLXevoTvNw2ksMZJzgUGl1ARCBtsmpy045mEe1apK31hVz87OWpUViGFxeRAM269UtGsrW9y13fY3SmyQPjLkHjHqz+Bio/drGx59R2HwHkzGGvkefLrbwwSZIulYDa1Dle+ICsicDua4kC1sEPQ5a/JFsrar4K3dX12fNbWt+9XuzlT7fFZKFD1Zy57LKYxGLu+UQu13JDiiCgvo781R+WmeobRv85x15tTb98NLu4wvGLB7gnc98dvw/jbcCHtIRyf/SYRQTUZXctd3kFkyeEsPM6VNc9LlYvPu+g2kd/67aMtks6BtUezSUR4BUC1WcjF2ULcNyzZnT/UrmMcEc9wybq+Earo3yVRGewGT9aZGTINdIo2H3t8mUwqVwYOWc5DWWi9X5DHkUUXqFtHUd75D5l5c1L5hG5Q86++HDndS8vgPUFLFi9/tFpGyOXnxXiVfAnpUlfpqvuyvxYzsbgey1eYxL9udHA9TgFzSyz6wgrklr177X69kypKMWU+enIvb/jaI5cIeZH0whP09C2uPmi5WmXvrA5FP9PuNwwKzTSyCndokobHy4VNzJBzLbOZCefzGt3CvXm2mR4ZyqOH8kYeOgf/c6s8X6z+9/vd79Z9+//RQ//krHlsB0Bu7YR28BRntXMTcEnu361NnEOZXN6PepddadPCpxuHb2wOmvCARmzdux8RpGDnvKXbFi9tAiNikcwzA8VJFXO3oeLbm8KNr8AXmykej1PDZWwoTOaPDoEYCsHPutoU1aMG1dc4wsmgQOh1b3PT+tVEX2f56FIKqy6avzVsqD7oIwoxu5HBCkJMLpaQ/91pP7MjK/+i+98XHzvWyMxC7nN0qJCHK4mBrIOiSZmeBF/jHT5TEBuw8RpfOuwCidB+eYh+yVIMnMaInl/I0BtGQpsR3DMC4hCQqExdfnCxH6OtCDUacSqNtRA3XFBUkVLHwvEtfIE0k3l0T79xu0PNFcIq/7+m/M/o7Iyi9WiLYJsAbAsTM4NYNnXEG3VNiSeId7P7/yTNTSfgnl/+/XP+v+0r7f9I/3P/+y+0/2QjjBXyya1GKDGce6nmsEGup1BfgFKyl9MVv3ETH+Q9PfJFq54JMQ217zsXrZkP/ss2tXHh82TBPb5GDZXJqjThZ3Y+fnDK35Zvi1I2QNkZq2fE7DWtPEGuCINoDdi4pjCfZq1GHMHcuByIfrUcdqrV3Lulyx9rZOsTlC4+hymHVQqzJItd8ylDZO41ENfTEf9INZ5drvJQLjNxwpC9HHcosO5ewcZPGV1gxY+v49ivjM8a/NJoRJZJLmgPWbm2W4ptiL3ddV133xX43D1ySKbT0bl3PcTH6CDPdBxz+uMBkjLAxng0ORGoKsz0kycswGVMZd8ip7VBLtsRDveRPfT3U14jv01wXp8PRCBpjLP0NmXXjE/FqlmvYFnmjweklsYKWPxzxx6hbNoyOepeXRgsoDgWHDfNnduX5qMPV+45BnhPmMEBksZLIsokJSteg+EKFljuAay5ZcC3vRJs04ZI7AVu4SHINPctK0oOhxWJVpA0NJFIO49WacxQ9nNF9EV7SEJ/VlzFd96YWjJv51fVSPUxIXvDCUznOl0jVMz1spmAkeC1B9VOsjTfdrPPx9afj+Xg67nJ84JdfezYSOxdPSlhsu+BDikCkHDQZf1SfDNxuNHTiYZaXjVkuPh3ncnZsTxHH18sy0qPfVPBux8uoQ8d46aoYU9o9vsEKQmb9sNyeT33xxKN7oy2wNe6dES93RxDWu0c8oRGGHFJJ7mLZ0kSKjv+GrKpPPWP/xgioumem7Qnaaqym7aVnsRWrRCUG38fFoJyuIQc6zxwvPi4sUz1f7HbSfLbTWJYFR1GIoZByww7/w7OMnKYJNAnCV4BtLK6BkEAQGrbNxV3NuZxFd1tlFt1KQYrtnqIhWeKgDiexjMbyeqVual4aa2BtxKJLv1BBDD5TMo70WNGPQZBaxEM0ohcmlS6MFvRZmdMFXYa+x1lIuDG34qZ5uMRkX05QfHsU+xWJykVLenIu3tKt/ouW+PqcLpCTgE0w/vZVe8BtE5x3cmrM3S39+GP8oVlc2r7ODu+bJJLuadlFDrls5yHeD7iHikL0Ca9B1RsO5endXCIz75qvUFsE1JAnBGgcffCOP9zvvWp6mpymvh961dV+O13AfV7L2S8onbQc+ppiMOiHon5ouG+qjIGwwTYA0UnUBYY21xOPOEvIqt8RUCGVIa7CldYUTATdHv1ER5vfUXSxJ87OtqpfdS5jFwFB3lbcs23bJ9J2sdXqWpbWKGYR48vUrRXUsxIpgVLvvLph50znC5OAsYpjf/DWtnkYfZQ3rHO65Q7V2c6OXfvjBaCUDMBpHOGB6ve80v845fkabxtz525KP+2Iaj9iLtvRARv9imlG1ytExqf9TnkDt3Xqc3Ft9YKWRYnp6Lrm3f2pMVlTdhM9YlZJTamTkZirKKJLIPZ3R0QfVipuc+JYTvdBwa2J5jsq2bHuH3/w7Y8I+f4GEykS+nETrUvSb9gMz/qf45nP244419/lXH+XcxPSoJogKKNwzf2O8jdz1e+squMG+WAXbaWK7TLcnhSaUWpMZTyHBlHt+ymAMFwlW87ibRhSFseQ2A71q61IA8LN+8ZYlFBsU4khQygU+TqilyMaMuYYQauiPp8XOpS90tHZ2ctJ2OpjDc5kRWGdSoMnm0Lqn1/DfvC9XptUBBy301s6+TAEMAKeNJLDzm3HJ83Bxh+WoP+8eP/Du/FPr9+9+uG9xxeJZoMqYZglHICWuceDwWVtb+C/qcpspoJD6OSTjhdQVEVaW9CpZznH+zD/OZmlbmrzFnKICeSKcR7DBvxjKBvKA69pXBZqSa1d8w0roC8p0t+6DDULwiiiSdxC+U2OMBrPKajIDw7Q8RhHia5X+vVIjvU87J8+ZLba5CwwTW4x8YK5XEfqSmq7DctBv00envAQk7MF+CoT6nEplQum6TIDsAYKGE7PCyLJnc15iWDNOEb7PDGbe7bL+UuH87JotczoN4K+5TyHCqbyUNalbinOMCWkaoh2KVK6MaGOb0pBwBKoQi4ROJP34GvwjSV+IldJWj3sIvV4COt5eagUHZ7Dc3gOz+E5PIfn8Byew3N4Ds/hOTyH5+/9/A/rFNxQ"}