Add V4 toy MLWE logical resource pass and charts

README.md

@@ -26,11 +26,15 @@ configs:
   data_files:
   - split: observations
     path: data/toy_ring_lwe_counts.csv
+- config_name: toy_mlwe_oracle_resources
+  data_files:
+  - split: observations
+    path: data/toy_mlwe_oracle_resources.csv
 ---
 
 # PrePQC Exploitation: public benchmark observations
 
-Four **separate** small experiment records from the [PrePQC Exploitation source repository](https://github.com/Maverick0351a/prepqc-exploitation), run on 2026-10-08. Each row is an aggregate of the stated job, operation, or fixed-seed trials; rows are not raw per-shot/per-case records or independent replications. These tables support reproduction and plotting; they are **not** a training set, a leaderboard, evidence of quantum speedup, or evidence that production ML-KEM was broken. The subsets have different inputs and success criteria and must not be pooled into a single score. The IBM and Garnet jobs used different providers and devices.
+Five **separate** small experiment records from the [PrePQC Exploitation source repository](https://github.com/Maverick0351a/prepqc-exploitation), run on 2026-10-08 and 2026-10-09. Each row is an aggregate of the stated job, operation, fixed-seed trials, or offline circuit case; rows are not raw per-shot/per-case records or independent replications. These tables support reproduction and plotting; they are **not** a training set, a leaderboard, evidence of quantum speedup, or evidence that production ML-KEM was broken. The subsets have different inputs and success criteria and must not be pooled into a single score. The IBM and Garnet jobs used different providers and devices; the toy oracle resource rows used neither device.
 
 | Subset | Rows | What each row means | Primary local result |
 | --- | ---: | --- | --- |
@@ -38,6 +42,7 @@ Four **separate** small experiment records from the [PrePQC Exploitation source
 | `garnet_basis_counts` | 16 | One outcome in one Open Quantum IQM Garnet job | [Garnet result record](https://github.com/Maverick0351a/prepqc-exploitation/blob/5485d9e/docs/results/openquantum-garnet-basis-2026-10-08.json) |
 | `mlkem_acvp_counts` | 5 | One OpenSSL/NIST sample-vector operation category | [ACVP result record](https://github.com/Maverick0351a/prepqc-exploitation/blob/main/docs/results/openssl-acvp-2026-10-08.json) |
 | `toy_ring_lwe_counts` | 3 | One fixed-seed metric from the tiny synthetic ring | [Toy method and result](https://github.com/Maverick0351a/prepqc-exploitation/blob/main/docs/TOY_RING_LWE.md) |
+| `toy_mlwe_oracle_resources` | 12 | One logical circuit resource ledger for a toy public-equation case/version | [V4 oracle protocol and result](https://github.com/Maverick0351a/prepqc-exploitation/blob/80116d8af606b6c69eff5725a7726aa6da4a53a3/docs/TOY_MLWE_ORACLE_V4_RESULTS_2026-10-09.md) |
 
 ## IBM QPU: observed and ideal outcome counts
 
@@ -63,6 +68,14 @@ Installed OpenSSL 3.5.6 matched 240/240 sample cases from the [NIST ACVP-Server
 
 For seed `20261008`, a search over all 729 candidate secrets accepted 20/20 constructed real challenges, rejected 20/20 paired uniform controls, and uniquely recovered the constructed secret in 20/20 real challenges. The ring is `Z_17[x]/(x^6+1)` with three public pairs per trial. The [source code](https://github.com/Maverick0351a/prepqc-exploitation/blob/main/tools/toy_ring_lwe.py) uses full negacyclic multiplication. This deliberately tiny synthetic task is not ML-KEM and does not establish a practical attack.
 
+## Tiny public-equation MLWE oracle: offline resource gate
+
+![Grouped bar chart comparing literal, recycled, and symbolic toy-oracle logical qubits](plots/toy_mlwe_oracle_qubits.png)
+
+![Grouped bar chart comparing literal, recycled, and symbolic toy-oracle two-qubit-gate proxies](plots/toy_mlwe_oracle_gate_proxy.png)
+
+The rank-two `Z_5[x]/(x^2+1)` toy has four binary secret bits and 16 candidates. The [preserved V1–V3 record](https://github.com/Maverick0351a/prepqc-exploitation/blob/17cd45fd31e755b792853824c8f3f88cef3e8e49/docs/TOY_MLWE_ORACLE_RESULTS_2026-10-09.md) includes the failed first mutation control, the amended control, and two resource failures. A [frozen V4 symbolic-coefficient circuit](https://github.com/Maverick0351a/prepqc-exploitation/blob/80116d8af606b6c69eff5725a7726aa6da4a53a3/docs/TOY_MLWE_ORACLE_V4_RESULTS_2026-10-09.md) passed all 64 final phase/cleanup checks and met the original **project pilot ceilings** of 64 qubits and 1,000 `CX + 6*CCX` proxy units: 26 qubits and 220–226 proxy units across four cases. V4 derives each coefficient check from public modular arithmetic without a global solution table, but its monomial basis grows as `2^n` and does not imply a scalable MLWE method. The proxy is not a transpiled circuit or calibrated hardware estimate. **Zero V4 QPU jobs were submitted**; backend compilation, noise, and matched-control gates remain. This toy does not imply a production ML-KEM break.
+
 ## Files and provenance
 
 - `data/*.csv`: numeric tables with units and identifiers; no keys, raw NIST vector bytes, IBM credentials, or CIC-PQC_OAV rows.

data/toy_mlwe_oracle_resources.csv

@@ -0,0 +1,13 @@
+date_local,case,oracle_version,solution_count,candidate_count,phase_checks_passed,ancilla_cleanup_failures,logical_qubits,x_gates,cx_gates,ccx_gates,z_gates,asap_logical_depth,two_qubit_gate_proxy_cx_plus_6ccx,pilot_resource_triage_pass,qpu_jobs_submitted
+2026-10-09,unique,literal_v1,1,16,16,0,295,112,588,232,1,207,1980,false,0
+2026-10-09,unique,recycled_v3,1,16,16,0,68,224,1188,454,1,847,3912,false,0
+2026-10-09,unique,symbolic_v4,1,16,16,0,26,10,34,32,1,27,226,true,0
+2026-10-09,no_solution,literal_v1,0,16,16,0,296,114,590,232,1,207,1982,false,0
+2026-10-09,no_solution,recycled_v3,0,16,16,0,70,228,1192,454,1,847,3916,false,0
+2026-10-09,no_solution,symbolic_v4,0,16,16,0,26,10,28,32,1,37,220,true,0
+2026-10-09,multiple_solution,literal_v1,2,16,16,0,297,116,592,232,1,207,1984,false,0
+2026-10-09,multiple_solution,recycled_v3,2,16,16,0,70,232,1196,454,1,847,3920,false,0
+2026-10-09,multiple_solution,symbolic_v4,2,16,16,0,26,12,30,32,1,37,222,true,0
+2026-10-09,mutated,literal_v2,0,16,16,0,294,110,586,232,1,207,1978,false,0
+2026-10-09,mutated,recycled_v3,0,16,16,0,68,220,1184,454,1,839,3908,false,0
+2026-10-09,mutated,symbolic_v4,0,16,16,0,26,10,32,32,1,33,224,true,0

manifest.json

@@ -7,16 +7,27 @@
     "circuits/grover2_mark_11_x_basis.qasm": "be3869202fb30d02117d06c013fbe2c82fcd0485d7da752c2b2b27c550dc5654",
     "circuits/grover2_no_mark_control.qasm": "bcb49e9936fdd475f69f3d32585e0574cdcddded5e93e8382604bc0aa08fe497",
     "circuits/grover2_no_mark_x_basis.qasm": "1167ddb55ba7468ccebe84714c4fc40726f3500a81ff1c735d8aac11ebf33086",
+    "docs/TOY_MLWE_ORACLE_AMENDMENT_V2.md": "afdf6cbea73f544c1094e0cfbeb973117779378c3f45a8f587759a6512b6e511",
+    "docs/TOY_MLWE_ORACLE_AMENDMENT_V3.md": "7eaa4cd5c9ae772d44ddbc9684c09d377a92563581f988be872d4e8af5d767ed",
+    "docs/TOY_MLWE_ORACLE_AMENDMENT_V4.md": "366b88e7ddcd2ceccb22faa2fcb53dfe3574f26d097c2508d8e82667c47a8dec",
+    "docs/TOY_MLWE_ORACLE_PROTOCOL.md": "13bf9b892d0e9be83db5002844df02c78a46cd9be264cf38fe23fb3541a7e7eb",
     "docs/results/ibm-qpu-2026-10-08.json": "57162ed0f9cfa8879e65de0fd90f2b484d389dccf2259aca50c05ecd25a8f67c",
     "docs/results/openquantum-garnet-basis-2026-10-08.json": "abe6a0e21742174bbc40b24cdbfdcf553191a375035604a07b8eda8dde1b0c2f",
     "docs/results/openssl-acvp-2026-10-08.json": "ad84740c632ad1ef80b29fa8d2b4b6f4f05f668259db02950f64b2894e7d687f",
+    "docs/results/toy-mlwe-oracle-v3-2026-10-09.json": "3c7936445c3c5e2371e2fcdea6a2a8d479aa3bc48e7d4f54e3260010132077c8",
+    "docs/results/toy-mlwe-oracle-v4-2026-10-09.json": "cbda086ded7c52a74b4cc435f2fcf94c5560110cd5d32cae9e574dabde38b458",
     "tools/qpu_twin.py": "12ceb02de96c0257fbc89aba5b23ea39147bf669c97b4b5feffe216b0205de25",
+    "tools/toy_mlwe_oracle.py": "adb8f04af82fb072b6f380e7bb6fe27edb8c041291828f5528882935ea4fa636",
+    "tools/toy_mlwe_oracle_v2.py": "7663de77b6e7c13be29fea724f9c1cd58034b51c33a4da09c04350f6c005409f",
+    "tools/toy_mlwe_oracle_v3.py": "dd2a2d69b16b4df130cd2e566cf86e7fe14e551ba44bf1a5323876f8b430071f",
+    "tools/toy_mlwe_oracle_v4.py": "9d4f08f84ba5240b67f35843d3b635f2c061bc36916e6bd60cc266830e7addc8",
     "tools/toy_ring_lwe.py": "eb7dd9e251f44fb7b9615de2943ff23811321a365c55a12bd6c213b2e92433d6"
   },
   "table_sha256": {
     "garnet_basis_counts.csv": "d3f8a44704c610a74d8fcbb59e8775e189016883c3388e6c7a9ce3cebac27f87",
     "mlkem_acvp_counts.csv": "cf4d30ce670af7c57cf0ad131f3f8b76a1c28017b52abd85b04c5b6e60bf8dbd",
     "qpu_counts.csv": "6462727962bcf105f2d9b00c70bcab98852766f202db0403f923f33faf203e91",
+    "toy_mlwe_oracle_resources.csv": "1e188ab1f9584bc5521d1328f9aefc9406348cfa082487f0a6ac5383f5d349df",
     "toy_ring_lwe_counts.csv": "d31f389f5092ef1aea35d64c25da7e045fadeb2fe80dc9e2e752e1dcaa9a38b1"
   }
 }