calvo_machine_learn#

  • Series: lecture-dp

  • File: lectures/calvo_machine_learn.md

  • Audit date: 2026-05-28

  • Categories audited: writing, math, code, jax, figures, references, links, admonitions

  • Overall score: 5.3 / 10

  • Priority: HIGH

Score breakdown#

Category

Score

One-line note

Writing

7/10

H2/H3 mostly sentence case; many long paragraphs.

Math

3/10

^T transpose throughout; \mathbf{I}/\mathbf{v}; \vec everywhere.

Code

5/10

pip install not at top (line 402 — three blocks for quantecon, optax, statsmodels); optax is a binary-package install with no notes.

JAX

out of scope

uses JAX but JAX rules out of scope per spec.

Figures

8/10

No matplotlib titles, no figsize, no Title-Case labels.

References

4/10

Eight narrative-author {cite} patterns (“Calvo {cite}…”, “Chang {cite}…”, etc.) — should be {cite:t}.

Links

5/10

Some markdown links to other QuantEcon series; {doc} used in places.

Admonitions

N/A

No exercises / proofs.

Issues#

Critical#

None found.

High severity#

  • [qe-math-002] — ^T (uppercase T) used as transpose throughout the gradient-descent quadratic-form derivations (e.g. g^T \vec{\mu}, \vec{\mu}^T M \vec{\mu}, B^T \vec{\beta}). Lines: 870, 872, 875, 877, 880, 889-891. Count: 13+ occurrences. Should be \top.

  • [qe-math-004] — \mathbf{I} used for the identity matrix (lines 875, 877, 880); \mathbf decoration prohibited.

  • [qe-code-003] — Three !pip install calls placed mid-lecture (line 402-404) rather than at the top. optax and statsmodels are also installed there.

  • [qe-code-006] — optax (a binary/JAX-companion package) installed without installation notes about Anaconda incompatibility.

Medium severity#

  • [qe-math-004] — \vec \mu, \vec \theta, \vec \beta, \vec v used throughout in narrative and math. Count: dozens.

  • [qe-ref-001] — Author-name narrative citations using parenthetical {cite} instead of textual {cite:t}. Count: 8 occurrences (“Calvo ”, “Chang ”, “Cagan ”, “Sargent and Wallace’s ”, etc.).

  • [qe-writing-001] — Many multi-sentence paragraphs (e.g. 39-54, 56-67).

  • [qe-writing-005] — Double-backtick quotes machine learning, gradient descent, dynamic programming squared used in narrative.

Low severity#

  • [qe-math-013 (proposed)] — Mixed eq-label style: some $$ … $$ (label) (line 181, 864) without eq: prefix.

Strengths#

  • Lecture title in correct Title Case.

  • H2/H3 mostly sentence case (“Introduction”, “The model”, “Model components”, “Parameters and variables”, “Timing protocol”, “A gradient descent algorithm”, “Implementation”).

  • Matrices use bmatrix (8 matrix blocks).

  • Definitions bolded (“state”, “Bellman equations”, “time inconsistency”, “machine learning”, “square summable”).

  • Equation labels and {eq} references used cleanly.

  • “IID”/Greek-letter conventions consistent.

  • No \tag, no align-inside-$$ issues.

  • No matplotlib title / figsize / spine violations.