fix: proofread Probability part

[vEnhance]

Proofreading pass over the Probability part (issue #315): tex/measure/randvar.tex, tex/measure/large-laws.tex, and tex/measure/martingale.tex. (randvar.tex needed no changes.)

Most fixes are minor (typos/grammar). A few are small calculation fixes, and two are corrections to mathematical arguments that I'd appreciate a second look at (flagged below).

Calculation / math fixes

• martingale.tex (casino example): the winnings on the third bet were computed as $1000 - $20 - $2 = $978, but betting $100 wins $100, so it should be $100 - $20 - $2 = $78.
• martingale.tex (ballot example): the third illustrative martingale $X_1 = \frac94 (A_1-B_1)^2$ gives \EE[X_1] = 3, not the intended 1. The coefficient should be \frac34 (since \EE[(A_1-B_1)^2] = \frac43).
• martingale.tex (product martingale): \EE(Y_n \mid \dots) = \EE[Y_n]\cdot Y_{n-1} should be \EE[X_n]\cdot Y_{n-1} — the conditional expectation factors out the known Y_{n-1}, leaving \EE[X_n] = 1.
• large-laws.tex (Kolmogorov inequality proof): the cross term in the expansion of (S_i + (S_n-S_i))^2 was missing its factor of 2. (The term is 0, so the conclusion is unaffected, but the expansion was incorrect.)
• large-laws.tex ("Quantifier hell" solution): the a.s.-convergence conclusion read \lim_n |X-X_n| = 1; it should be = 0 (the surrounding argument bounds it by 1/m for every m).
• large-laws.tex: added the missing \,d\mu to \EE[X_i] = \int_\Omega X_i(\omega).

⚠️ Argument corrections — please review

• martingale.tex, "Stopped supermartingales are still supermartingales" proof: the increment decomposition was written as X_{n-1} + \mathbf 1_{\tau = n-1}(X_n - X_{n-1}) bounded by X_{n-1}. As written this identity does not hold (e.g. when \tau = n-1 or \tau \ge n). I corrected it to the standard decomposition X_{\tau\wedge(n-1)} + \mathbf 1_{\tau \ge n}(X_n - X_{n-1}) \le X_{\tau\wedge(n-1)}, which is what the supermartingale conclusion requires. Worth a sanity check.
• martingale.tex, optional stopping "Finite bets" hypothesis: changed the conditioning from \SF_n to \SF_{n-1} in \EE(|X_n - X_{n-1}| \mid \SF_n). Conditioning on \SF_n makes the increment already measurable (so the conditional expectation is trivial); the standard bounded-increments hypothesis conditions on \SF_{n-1}.

Typos / grammar

• "converges in probability to $X$ if if for every" → removed duplicate "if".
• "most distribution … has finite variance" → "most distributions … have".
• "expected number of steps starting to reach 0 starting from" → removed redundant "starting".
• casino game: "bet any amount of on a fair coin flip" → "bet any amount on".
• "remains at slightly uncertain" → "remains slightly uncertain".
• example X \colon \Omega \to \{0,1,2\} was inconsistent with the events X=1,2,3 defined just below → \{1,2,3\}.
• "This is a sub-$\sigma$-algebra on $\SF$" → "on $\Omega$".
• "rolling two dices" → "two dice"; "would takes the value" → "would take"; "in any world whether $X+Y=3$" → "where".
• "this naive attempts will give" → "these naive attempts".
• "Frivolous/inflamamtory example" → "inflammatory".
• "take it to be the our currently value" → "our current value".
• "Almost sure convorgence is not topologizable" → "convergence".

Closes the Probability checkbox in #315.

🤖 Generated with Claude Code

---

Generated by Claude Code