MATH40007 Over Anchoring Diagnosis and Correction

English passage 1 (1% down the source page) MATH40007 Correction report 2026 sets difficulty; seven years of past papers set the scope

English passage 2 (6% down the source page) MATH40007 Diagnosis of over-anchoring and correction of historical coverage

English passage 3 (12% down the source page) Core conclusion: the old six papers were not entirely useless, but their structure treated the 2026 main examination as both a difficulty benchmark and a content template. Question positions, entry points and method sequences consequently remained fixed. The revision retains only one paper for 2026 structural reinforcement; the other five use historical ability families from 2020–2025. Each paper remains calibrated to the 2026 overall difficulty allocation of A32/B20/C12/D16.

English passage 4 (19% down the source page) 1 Where exactly the problem lies

English passage 5 (23% down the source page) The old materials contain three structural errors :

English passage 6 (26% down the source page) 1. Fixed four-slot structure. The old plan fixed Q1 as linear algebra/networks, Q2 as conductance/random walks, Q3 as spectral dynamical systems and Q4 as continuous media. These functions closely match the four 2026 questions.

English passage 7 (29% down the source page) 2. Treating labels as coverage. For example, mentioning eigenvalues in two questions was treated as coverage. Directly calculating matrix eigenvalues and constructing a generalised eigenvalue problem from an unfamiliar physical model, then projecting initial conditions, are different transferable abilities.

English passage 8 (33% down the source page) 3. The seven-year analysis did not constrain generation. Historical tree recurrences, Green matrices, complete spectral expansions of graphs, generalised mass matrices, slit-endpoint singularities and hitting problems with different absorbing sets appeared in reports but received insufficient direct practice.

English passage 9 (37% down the source page) The old papers explicitly said based on the 2026 paper structure, and their coverage table used the same four question positions. It is therefore unsurprising that Tony could recognise the 2026 format fluently but struggled to identify starting points in official papers such as 2025.

English passage 10 (43% down the source page) 2 Revised mandatory rules

English passage 11 (47% down the source page) 2026 main examination: 2 hours, 80 marks, four questions, overall cognitive load, derivation length, computation, abstraction and A32/B20/C12/D16. 2020–2026 papers: knowledge scope, historical question types, recognition signals, first key steps, course method sequences and combinations. Effective coverage: A = direct question-type coverage; B = method coverage with transfer to unfamiliar formulations. Ineffective coverage: C = matching knowledge-point names only; D = entirely absent. C no longer counts towards effective coverage.

English passage 12 (57% down the source page) The 2026 paper determines difficulty; seven years of past papers determine content, question forms and recognition of the first step. The equal-difficulty principle remains. Requirements for identical question types, knowledge points and fixed positions are removed.

English passage 13 (63% down the source page) 3 Roles of the six new papers

English passage 14 (68% down the source page) Number of papers role Main historical abilities

English passage 15 (70% down the source page) Paper 1 — 2026 structural reinforcement. The only paper retaining a near-2026 structure: four fundamental subspaces, small circuits, network reduction, forced diffusion and electrothermal rods. Paper 2 — 2025 bridge. Tridiagonal/cyclic spectra, Thomson minimisation, random walks with different absorbing sets, two-mass oscillations and conductors with piecewise sources. Paper 3 — 2023–2024 abilities. Parameterised circuits, tree recurrences, complete graph spectral expansions, multiple-source complex potentials and boundary-current diagrams. Paper 4 — 2020–2022 abilities. Large-network response, Green matrices and reciprocity, lifting block-matrix spectra and circular-electrode limits. Paper 5 — Seven-year gap closure. Wheel-graph nullspaces, hitting after graph modification, generalised modes with nonuniform masses and slit-endpoint singularities. Paper 6 — Mixed-transfer pressure. Varied question order: first-quadrant complex potentials, initial-velocity modes for three masses, random walks on unfamiliar graphs and Green matrices for chorded graphs.

English passage 16 (84% down the source page) At least five papers have main ability combinations different from 2026. The new papers no longer require each question number to correspond to a fixed module; they require only a reconciled whole-paper difficulty budget.

English passage 17 (88% down the source page) 4 How old files are handled

English passage 18 (92% down the source page) Old question booklet MATH40007_All_6_Questions_FINAL_V6.pdf: label OBSOLETE - 2026 OVER-ANCHORED, remove from the active allowlist and retain in quarantine for traceability.

English passage 1 (1% down the source page) MATH40007 Correction report 2026 sets difficulty; seven years of past papers set the scope

English passage 2 (6% down the source page) Old combined solutions: also quarantine them and stop using them as answers to new questions.

English passage 3 (8% down the source page) Old coverage reports: preserve as historical records, but stop displaying the old conclusion that most methods across seven years have been covered.

English passage 4 (11% down the source page) Tony’s first-paper marking report: may be retained as learning feedback; it does not validate the old six-paper structure.

English passage 5 (15% down the source page) 5 Difficulty calibration

English passage 6 (19% down the source page) Each paper satisfies :

English passage 7 (23% down the source page) Calibration checks not only marks but also normal completion time, method steps, algebraic load, modelling and diagram interpretation, proof or explanation demands, and unfamiliar transfer. It is a practice design based on the official 2026 marking system. Only the institution’s exam setters and external assessors can certify official equivalence.

English passage 8 (28% down the source page) 6 Effective coverage results and limitations

English passage 9 (32% down the source page) The new workbook maps 28 historical year/whole-question ability units individually to the new papers. Several units had only C-level label coverage in the old system. After correction, all 28 have A- or B-level practice mappings. The count of 28 uses whole 20-mark ability families; the workbook also records each question’s internal subskill sequences.

English passage 10 (37% down the source page) Must honestly state :

English passage 11 (40% down the source page) Practice papers cannot enumerate every algebraic variation.

English passage 12 (43% down the source page) An A/B mapping means a question type or method sequence has been practised; it does not mean Tony has mastered it.

English passage 13 (45% down the source page) The 2020–2026 official papers should remain the final transfer test; practice papers alone are insufficient.

English passage 14 (47% down the source page) If course materials conflict with a solution method, prioritise the course notes, problem sheets and standard methods in official solutions.

English passage 15 (52% down the source page) 7 Suggested order of use

English passage 16 (56% down the source page) Complete Paper 2 first to address unfamiliarity with 2025, then Paper 4 for Green matrices and block spectra, followed by Papers 3 and 5. Use Paper 1 only to maintain 2026 fluency and complete Paper 6 last under timed conditions. The recommended daily allocation is 70% historical official questions or historical-ability papers and 30% 2026 reinforcement, until Tony can identify the module, recognition signal and first key step in unfamiliar questions.

English passage 17 (63% down the source page) 8 Final deliverables

English passage 18 (67% down the source page) The package contains six revised question papers, course-method solutions, the seven-year effective-coverage workbook, this diagnostic report, a file QA report and Codex import notes. The final ZIP excludes old V1–V6 question booklets.