Repository navigation
Releases
Releases · emil513t/obsidian-math-autocomplete
0.7.0
Proofs that something fails, from your own values
Write the values in a ```vars block and pick the ★ route. The plugin finds the witness and writes the book's argument with it. There are 11 engines, each tested on the book's exercises:
- E1 Negation:
\negatewrites ¬P rule by rule ((1.3), Ex 1.20, Def 1.10) and gives Remark 5.34 word for word. It also fills\contradictionand\contrapositive. - E2 + E3 One concrete case: an element in one set but not the other; the truth-table row where two propositions differ; function tables (f(1) = 7 = f(4)); x² never hits −1; x³ at 0 is not an extremum; (a, b) = (2, 1) for Ex 1.72.
- E4 Contradicting inequalities: "adding c1, c2 and c3 gives 0 < 0", in
\fm,\svmNOT separable (Ex 5.9, 4.42) and\kktno feasible point. - E5 No solution:
\gausspicks its route and gives the row 0 = c, plus the one-line combination. - E6 Matrix witnesses: the bad vector for "not positive definite", Av = 0, inverses with parameters.
- E7 Not bounded / open / closed, no limit, not differentiable: n = ⌈|R|⌉ + 1 for ℕ, (0, ε/2) ∉ the x-axis, 1 − min(ε/2, 1) for [0, 1), jumps with ε = ¼ gap, corners.
- E8 Not convex: an explicit pair breaking Def 4.26 (also with e^x); the segment where f is linear.
- E9 Not optimal: the KKT equations at the point with no λ ≥ 0 (Ex 9.52, 9.56, 9.57), or a better feasible point.
- E10 No rational root: Example 1.83's proof for √2, ∛2 (Ex 5.82), √3, √8, …
- E11 Find the mistake: "divides by a − b, which the first line makes 0" (Ex 2.2); the fake induction of Ex 1.89.
Also new
- README is now a full instruction book, and every command, route and possible ending is listed. Run
npm run readmeto regenerate the list. - Python output is readable: exact values when short (with the decimal added),
~2.09455instead of cubic-formula blobs, intervals as(-inf, a) U (b, inf), matrices one row per line, and summaries instead of long loops. - Fixes:
\logis the natural logarithm, as in the book. Derivatives of logs used to get a factor ln 10.- The
\fouriermotzkinworked solution printed0 ≤ y ≤ 0fory > 0, y ≤ 0; it now says there is no solution. - The first ⇔ of a chain was skipped.