% !TeX program = lualatex
% =====================================================================
%  analysis.tex
%  Analysis in one chapter: big operators with an index (sum, product),
%  limits — both the lim(...) operator and the arrow form written under
%  a long arrow — trigonometry with upright function names, derivatives
%  up to a differential equation and the heat equation, the vector
%  operators (grad, div, curl, Laplacian, directional derivative), the
%  whole integral family (primitive, definite, multiple, contour,
%  principal value, average, surface, volume), the Landau notations and
%  the integral transforms.
% =====================================================================
\documentclass[
  margins=8,
  font=Latin Modern Roman,
  size=12,
  linespread=1.4,
  lang=en
]{scholatex}
\begin{document}

let title = <Red b 18pt c>
let h1    = <Navy b section>

<title>scholatex — analysis

% =====================================================================
<h1>Operators with an index
% =====================================================================

A big operator carries its index in (...); its body follows freely
and is set in display style so fractions stay full size.

<box line:Indigo fill:Lavender radius:3 title:{Sum and product}>{
	$sum(i=1, n) i = n(n+1)/2$ <3tab> $prod(k=1, n) k$
}

% =====================================================================
<h1>Limits
% =====================================================================

A limit's (...) holds the approach, written with ->; the target
sits under the word, as it should:

<box line:Indigo fill:Lavender radius:3>{
	$lim(x->0) f(x)$ <3tab> $lim(x->+inf) 1/x$
}

For the running phrase "$u_n$ tends to $l$" set under a long arrow, use
arrow(...): the condition is written underneath, and to or ->
read the same inside it.

<box line:Indigo fill:Lavender radius:3 title:{A sequence converging}>{
	$u_n arrow(n to +inf) l$ <3tab> $1/n arrow(n to +inf) 0$
}

% =====================================================================
<h1>Trigonometry
% =====================================================================

Function names — sin, cos, tan, ln, exp and
the rest — are set upright automatically, and a name glued to (...)
takes its argument as one atom, so fractions behave.

<box line:Crimson fill:MistyRose radius:3 title:{The fundamental identity}>{
	$sin(x)^2 + cos(x)^2 = 1$
}

<box line:Crimson fill:MistyRose radius:3 title:{Addition formula}>{
	$cos(a+b) = cos(a)cos(b) - sin(a)sin(b)$
}

<box line:Crimson fill:MistyRose radius:3>{
	$tan(x) = sin(x)/cos(x)$ <3tab> $lim(x->0) sin(x)/x = 1$
}

% =====================================================================
<h1>Derivatives and differential equations
% =====================================================================

A derivative is written as the fraction it is. The differential d is
set upright (ISO 80000-2), matching the d of the integrals — but only
when both sides of the fraction carry it, so a variable named d is left
alone (d/2 stays a plain fraction).

<box line:DarkOrange fill:OldLace radius:3 title:{Leibniz notation}>{
	$dy/dx$ <3tab> $(d^2 y)/(dx^2)$ <3tab> $d/dx (x^2) = 2x$
}

A first-order differential equation reads in one line:

<box line:DarkOrange fill:OldLace radius:3>{
	$dy/dx + y = 0$
}

Partial derivatives use partial ($partial$); parenthesise each side so
the fraction groups correctly. The heat equation, for instance:

<box line:DarkOrange fill:OldLace radius:3 title:{The heat equation}>{
	$(partial u)/(partial t) = (partial^2 u)/(partial x^2)$
}

% =====================================================================
<h1>Differential operators
% =====================================================================

The first-order vector operators read as named operators: $grad(f)$ the
gradient, $div(F)$ the divergence, $curl(F)$ the curl. The Laplacian is
written $lap(f)$; it prefixes its operand without function parentheses, so
$lap(f)$ stays bare, while a compound operand keeps its grouping, as in
$lap(x^2 + y^2)$.

<box line:DarkOrange fill:OldLace radius:3 title:{The Laplacian as divergence of the gradient}>{
	$lap(f) = div(grad(f))$
}

The derivative of $f$ along a direction $u$ is $dirderiv(f, u)$, the gradient
indexed by the direction.

<box line:DarkOrange fill:OldLace radius:3 title:{Directional derivative}>{
	$dirderiv(f, u) = grad(f) cdot u$
}

% =====================================================================
<h1>Integrals: body and differential
% =====================================================================

An integral closes on a differential. Its head (...) names the
variable; everything up to the end of the formula is the integrand, and the
differential $dx$ is appended automatically.

<box line:DarkGreen fill:Honeydew radius:3 title:{Primitive and definite integral}>{
	$int(x) f(x)$ <3tab> $int(x=a, b) f(x)$
}

The head's variable is the differential, so a change of letter is just a
change in the head: $int(t=0, 1) t^2$. To keep a term outside the
integral, close the integrand in parentheses — these differ:

<box line:DarkGreen fill:Honeydew radius:3>{
	$(int(x=a, b) f(x)) + 1$ <3tab> $int(x=a, b) (f(x) + 1)$
}

% =====================================================================
<h1>Multiple integrals
% =====================================================================

Separate several domains with ; inside the head. The count of
domains chooses the single, double or triple integral sign; the
differentials come out in reverse order, the Fubini convention.

<box line:DarkGreen fill:Honeydew radius:3>{
	$int(x=a, b ; y=c, d) f(x,y)$

	$int(x=a, b ; y=c, d ; z=e, g) f(x,y,z)$
}

A single named domain is a region integral over that set, with the area
element giving the surface form: $int(D) f$.

% =====================================================================
<h1>Contour, principal value, average
% =====================================================================

Three named integral operators round out the family: a contour integral
$contourint(C) f(z)$, a Cauchy principal value $pvint(x=a, b) f(x)$, and
the average (normalised) integral $meanint(x=a, b) f(x)$.

<box line:DarkGreen fill:Honeydew radius:3 title:{The integral family}>{
	$contourint(C) f(z)$ <3tab> $pvint(x=a, b) f(x)$ <3tab> $meanint(x=a, b) f(x)$
}

% =====================================================================
<nextpage h1>Surface and volume integrals
% =====================================================================

The closed surface integral is $surfint(S)$ and the volume integral is
$volint(V)$, both built on the integral signs unicode-math provides
natively. A flux reads $flux(F, S)$.

<box line:DarkGreen fill:Honeydew radius:3 title:{The divergence theorem}>{
	$flux(F, S) = volint(V) div(F)$
}

% =====================================================================
<h1>Landau notation
% =====================================================================

The asymptotic comparisons are the explicit words $bigO(...)$ and
$litO(...)$, so the bare letters $o$ and $O$ stay free as variables.

<box line:Indigo fill:Lavender radius:3 title:{A first-order expansion}>{
	$exp(x) = 1 + x + litO(x)$ as $x to 0$

	$sum(k=1, n) k = n^2/2 + bigO(n)$
}

% =====================================================================
<h1>Integral transforms
% =====================================================================

The transforms name themselves: $laplace(f)$ and $fourier(f)$, with inverses
$ilaplace(f)$ and $ifourier(f)$.

<box line:Indigo fill:Lavender radius:3 title:{A transform pair}>{
	$laplace(f)$ on one side, $ilaplace(laplace(f)) = f$ on the other.
}

\end{document}