Diagnostic Intake Key¶
Use this key to grade diagnostic-intake.md Mode A answers. It is not a placement test for the whole foundations course. Its only purpose is to decide how to start the first live foundation session.
Scoring:
strong: answer is correct enough to compress or skip the corresponding material.partial: answer has the right direction but needs cleanup during the lesson.weak: answer is missing, incorrect, or too vague to use as evidence.
Do not mark Lesson 01 as skipped by diagnostic unless questions 1-5 are all strong.
Question 1¶
Prompt: In GF(2), why is \(1+1=0\), and why does that matter for parity checks?
Strong answer:
- GF(2) arithmetic is modulo 2, so two ones cancel.
- Parity checks count whether the number of selected
1entries is even or odd. - A parity-check equation is satisfied when the selected sum is
0mod 2.
Partial answer:
- Says "mod 2" but does not connect it to parity checks.
Weak answer:
- Uses ordinary integer arithmetic;
- says only "because binary" without explaining cancellation;
- cannot connect to parity.
Routing:
- If weak, teach Lesson 01 from the GF(2) arithmetic section.
Question 2¶
Prompt: For a parity-check row \(h\), what does \(h c^T=0\) mean?
Strong answer:
- It means the dot product over GF(2) between the check row and codeword is zero.
- Equivalently, the coordinates selected by \(h\) have even parity.
- For a parity-check matrix \(H\), a codeword satisfies all rows, so \(Hc^T=0\).
Partial answer:
- Correctly says "it is a parity check" but does not mention GF(2) or selected coordinates.
Weak answer:
- Treats it as a real-valued orthogonality statement without mod 2 context;
- cannot identify what the row checks.
Routing:
- If weak, teach Lesson 01 slowly and include explicit dot-product examples.
Question 3¶
Prompt: In the \([7,4,3]\) Hamming code, what information does a single-bit-error syndrome give you?
Strong answer:
- For a standard Hamming parity-check matrix whose columns are the nonzero binary triples, a single-bit error at position \(i\) gives syndrome equal to column \(i\).
- The syndrome identifies the error location.
Partial answer:
- Says "it tells you the error" but does not explain the column-label mechanism.
Weak answer:
- Confuses syndrome with codeword;
- says it only detects but cannot locate a single-bit error;
- cannot state what the syndrome indexes.
Routing:
- If weak, start Lesson 01 from the Hamming worked example rather than skipping.
Question 4¶
Prompt: What is the difference between a generator matrix and a parity-check matrix?
Strong answer:
- A generator matrix \(G\) spans the code by linear combinations of message rows.
- A parity-check matrix \(H\) defines the code as the kernel:
- In the full-rank classical setting, the row space of \(H\) is the dual code \(C^\perp\).
Partial answer:
- Says "one generates, one checks" but cannot state span versus kernel.
Weak answer:
- Treats \(G\) and \(H\) as interchangeable;
- cannot describe either span or kernel.
Routing:
- If questions 1-3 are strong but question 4 is weak, compress Lesson 01 and move quickly into Lesson 02.
Question 5¶
Prompt: In one sentence, why does the Conjecture 3 course begin with classical linear codes rather than QLDPC codes directly?
Strong answer:
- QLDPC and CSS constructions are built from binary parity-check matrices, Tanner graphs, duality, and code parameters, so the classical layer provides the algebraic and graph language needed for the Conjecture 3 barrier.
Partial answer:
- Says classical codes are prerequisites but does not name the specific objects that carry into QLDPC.
Weak answer:
- Gives a generic "basics first" answer with no research-map connection;
- says classical codes are unrelated and only historical background.
Routing:
- If weak while questions 1-4 are strong, use the research-orientation part of Lesson 01 but compress the calculations.
Question 6¶
Prompt: Which part feels least automatic: GF(2) arithmetic, syndrome calculation, matrix row spaces, or research-map motivation?
Strong answer:
- Identifies a concrete weak area and gives a usable reason or example.
Partial answer:
- Selects a weak area but gives no detail.
Weak answer:
- Does not answer, or says everything is fine while earlier answers show a gap.
Routing:
- Use this answer to choose emphasis. It should not by itself pass or fail Lesson 01.
Mode A Routing Summary¶
Use this routing after grading questions 1-6.
| Evidence pattern | Route |
|---|---|
| Any of questions 1-3 weak | Start Lesson 01 from the beginning |
| Questions 1-3 strong, question 4 weak or partial | Compress Lesson 01 calculations, then emphasize Lesson 02 |
| Questions 1-4 strong, question 5 weak | Teach the research-orientation part of Lesson 01, then continue to Lesson 02 |
| Questions 1-5 strong | Propose marking Lesson 01 as skipped by diagnostic, but update progress only after learner confirmation |
Recording Rule¶
After grading, propose a concise update:
- Mode A result;
- question-level strengths and gaps;
- recommended next lesson;
- whether Lesson 01 should remain
not started, becomein progress, or be markedskipped by diagnostic.
Record the result in session-log.md and the foundation progress tracker only after learner confirmation, following session-protocol.md.