[vEnhance]

Closes #69.

Proofread src/sol-juliett.typ (solutions for Part Juliett). Found and fixed two issues:

• exer-flux-sealing: clarified an ambiguous sentence that said "We orient the normal vector pointing upward" without naming which surface; the orientation choice being described is for $cal(S)_"lid"$, not $cal(S)$ (whose orientation is already fixed by the problem statement). Specified the surface explicitly.
• exer-gravity-div2: the divergence theorem identity had the two surface-flux terms subtracted in the wrong order. Given that $cal(S)1$'s normal points away from $cal(T)$ (matching the outward-from-$cal(T)$ orientation) while $cal(S)2$'s normal points toward $cal(T)$ (opposite the outward-from-$cal(T)$ orientation), correctly applying the divergence theorem gives iint_(S1) - iint_(S2) = iiint_T div, not iint_(S2) - iint_(S1) as written. I verified the corrected sign by direct computation with a radial test field ($bf(F) = chevron.l x,y,z chevron.r$, $div bf(F) = 3$) on concentric spheres of radius 1 and 2: $32pi - 4pi = 28pi$ matches $iiint_T 3 dif V = 28pi$, confirming iint_(S1) - iint_(S2) is the correct order. Since this exercise has $div bf(G) = 0$, the final boxed conclusion ($oiint(S_1) = oiint(S_2)$) is unaffected either way, but the intermediate identity as stated was incorrect in general.

No other mathematical errors were found; the remaining solutions (flux through $x^3+y^4=e^z$, divergence-to-Green, Stokes-to-Green) were independently re-derived and check out correctly.

Test plan

• Installed typst 0.15.0 plus the @local/evan:1.0.0 package (from vEnhance/dotfiles) and the required @preview deps (gentle-clues, linguify), generated placeholder SVGs standing in for the Asymptote figures (asy isn't available in this environment), then ran typst compile lamv.typ end-to-end — compiles cleanly with no errors.
• Ran codespell on the file — no issues.

---

Generated by Claude Code