summaryrefslogtreecommitdiff
path: root/Master/texmf-dist/doc/optex/optex-math.tex
diff options
context:
space:
mode:
Diffstat (limited to 'Master/texmf-dist/doc/optex/optex-math.tex')
-rw-r--r--Master/texmf-dist/doc/optex/optex-math.tex1662
1 files changed, 1662 insertions, 0 deletions
diff --git a/Master/texmf-dist/doc/optex/optex-math.tex b/Master/texmf-dist/doc/optex/optex-math.tex
new file mode 100644
index 00000000000..789dd5a1877
--- /dev/null
+++ b/Master/texmf-dist/doc/optex/optex-math.tex
@@ -0,0 +1,1662 @@
+%% This is part of the OpTeX project, see http://petr.olsak.net/optex
+
+% Run optex optex-math (two times) to generate this document
+% or look at PDF here: http://petr.olsak.net/ftp/olsak/optex/optex-math.pdf
+
+\fontfam[lmfonts]
+
+\font\ttlib=[LiberationMono-Regular]
+\def\ttspec{\let\_ttfont=\ttlib}
+
+\report
+
+\def\new #1 {\mnote{\Red$\blacktriangleleft$\,\sans\setfontsize{at9pt}\rm#1}}
+\fixmnotes\right
+\enquotes
+\verbchar`
+\catcode`\<=13
+\def<#1>{$\langle\hbox{\it#1\/}\rangle$}
+\everyintt={\catcode`\<=13 }
+\def\ss#1{$\vrule height3pt#1\vrule height3pt$}
+
+\addto\_secfont\Blue \addto\_seccfont\Blue
+\_def\_printsec#1{\_par
+ \_abovetitle{\_penalty-400}\_bigskip
+ {{\_secfont \_noindent \_raggedright \llap{\_printrefnum[@\_quad]}#1}\_nbpar}\_insertmark{#1}%
+ \_nobreak \_belowtitle{\_medskip}%
+ \_firstnoindent
+}
+\_def\_printsecc#1{\_par
+ \_abovetitle{\_penalty-200}\_medskip
+ {{\_seccfont \_noindent \_raggedright \llap{\_printrefnum[@\_quad]}#1}\_nbpar}%
+ \_nobreak \_belowtitle{\_medskip}%
+ \_firstnoindent
+}
+
+\def\i #1 {\ii .#1 \iis .#1 {{\code{\\#1}}}}
+\def\x`{\bgroup\_setverb\xx}
+\bgroup \lccode\string`\.=\string`\` \lowercase{\egroup \def\xx #1#2.{\i #2 \egroup `#1#2.}}
+
+
+\hyperlinks\Green\Green
+
+\insertoutline{MATH}
+\outlines{0}
+
+
+
+\tit Typesetting Math with \OpTeX/
+
+\hfill Version 03, January 2021
+
+\author Petr Olšák
+
+
+This document is a brief summary of typesetting math. It describes \TeX/,
+Plain \TeX/ and \OpTeX/ features concerned to math. The first two types of
+features are documented in \TeX/book in chapters 16, 17, and 18, but it is
+summarized here in short again in order to give a complete guide about math
+typesetting for \OpTeX/ users.
+
+\new {}
+The \OpTeX/ features which differs from standard \TeX/ or Plain \TeX/ are
+documented with the red triangle at the margin (like in this paragraph).
+Reader can simply distinguish between \"standard" features (given by
+\TeX/ or Plain \TeX/) and new \OpTeX/ features.
+
+There are more types of extensions: e\TeX, lua\TeX/, Unicode math and
+\OpTeX/ macros. The appropriate label (e\TeX, Lua\TeX/, Unicode, \OpTeX/)
+is appended to the red triangle to inform you about the extension type.
+Nevertheless, \OpTeX/ user doesn't have to worry about it, all extensions
+are available if Unicode Math font is loaded (e.g., by the command
+`\fonfam[lmfonts]`). See section 1.3.3 in \OpTeX/ documentation about
+loading Unicode math fonts.
+
+{\iindent=2em
+\bigskip
+\maketoc }
+\vfil\break
+
+\sec Basics structure of math formulas
+
+\secc General rules and terminology
+
+The \ii in-line/math in-line math (in the paragraph) is created by `$<math list>$`. The
+\ii display/math display math (a standalone line between paragraphs) is created by `$$<math list>$$`.
+More than one line can be here if an appropriate macro is used. In-line math is
+processed in a \TeX/ group in \ii in-line/math/mode {\em in-line math mode}. The display math is
+processed in a \TeX/ group in \ii display/math/mode {\em display math mode}. Spaces are
+ignored in math modes, so `$x+y$` and `$x + y$` gives the same result: $x+y$.
+
+The \ii math/list <math list> is a sequence of \ii math/atom,atom {\em math atoms} and
+\ii other/material {\em other material}.
+The math atoms are \ii single/math/object {\em single math objects} or
+\ii composed/math/atom composed math atoms.
+
+\begitems \hfuzz=.6pt
+* The single math object is a single character to be printed in math mode
+ like `x`, `+`, `\int`.
+* The math atom is constructed in general by `{<math list 1>}^{<math list 2>}_{<math list 3>}`.
+ It consists from \ii nucleus {\em nucleus} <math list 1>, \ii exponent exponent <math list 2>
+ and \ii subscript subscript <math list 3>. Each part of the atom should be empty.
+ If <math list 2> or <math list 3> is empty, we need not to write brackets and
+ the prefix `^` or `_`.
+ If the <math list 1> or <math list 2> or <math list 3>
+ consist only from a single math object then we need not use brackets.
+ For example
+ `x^2` is a math atom with `x` in the nucleus, `2` in the exponent, and with empty subscript.
+ Or `a_{i,j}` is a math atom with `a` in the nucleus, empty exponent, and `i,j` in the subscript.%
+ \fnote{In \OpTeX/, the character `_` can be interpreted as a part of
+ the control sequence name, not as the subscript constructor. But in common cases,
+ constructions of math atoms are interpreted exactly as in plain \TeX. See sections
+ 2.2.2 and 2.14 of \OpTeX/ documentation for more details. If you want to
+ be sure that `_` is just a subscript constructor in \OpTeX/ then you can set \code{\\catcode`\\_=8}
+ but after this, you cannot use control sequences with `_` character.}
+ The constructors for exponent `^` and subscript `_` can be used in arbitrary order
+ after the nucleus, for example, `z_1^{x+y}` is the same math atom as
+ `z^{x+y}_1`. The single math objects not followed by `^` nor `_` are
+ considered as math atoms with this object in the nucleus and with empty
+ exponent and subscript (this is a very common case).
+ \TeX/ assigns the \ii class {\em class} for each math atom, see section~\ref[class].
+* Other material can be \TeX/ box or glue (space) or `\kern` or `\vrule` etc.
+\enditems
+
+Example: The `Z = \int``_\Omega x^{2y} + z\, dx` generates
+$Z = \int_\Omega x^{2y} + z\, dx$ and it is <math list> which consists from:
+
+\begitems
+* `Z` is math atom with empty exponent and subscript, class: Ord,
+* `=` is math atom with empty exponent and subscript, class: Rel,
+* `\int``_\Omega` is math atom with empty exponent and with subscript `\Omega`, class: Op,
+* `x^{2y}` is math atom with exponent `2y` and empty subscript, class: Ord,
+ \begitems \let\_bullet=\circ
+ * `2` is math atom with empty exponent and subscript, class: Ord,
+ * `y` is math atom with empty exponent and subscript, class: Ord,
+ \enditems
+* `+` is math atom with empty exponent and subscript, class: Bin,
+* `z` is math atom with empty exponent and subscript, class: Ord,
+* `\,` is another material, the glue (space) in this case,
+* `d` is math atom with empty exponent and subscript, class: Ord,
+* `x` is math atom with empty exponent and subscript, class: Ord.
+\enditems
+
+\secc[class] Classes of math atoms
+
+\TeX/ assigns \ii class {\em a class} for each math atom.\fnote
+{Using terminology of \TeX/book, each single math object has its {\em class} but the
+ math atom has its {\em kind} derived from this class. I use only one word
+ for both meanings in this document.}
+This data type is used when
+\TeX/ decides about \ii horizontal/spacing horizontal spaces between atoms in the output. (Note
+that spaces in the input are ignored.) For example,
+`$xy$` prints two atoms without space between them but `$x+y$` is printed with
+small spaces around the `+` binary operator. Compare: $xy$ and $x+y$.
+
+The class is assigned depending on the nucleus of the atom. If the nucleus is
+not a single math object, i.e. it is constructed by `{<math list>}` with braces
+then the atom has its class Ord. If the nucleus is a single math object constructed
+without braces then the class of the atom depends on this single math
+object. Each single math object must be declared in \TeX/ with its default
+class. The following table lists the classes with typical examples.
+The full set of all math objects used in math typesetting
+is listed in section~\ref[objects] with their default classes.
+
+\bigskip
+\noindent\hfil\table{llll}{
+ & \ii class Class & Meaning & Example \crl
+ 0 & \iid Ord & ordinary object & variables, digits, $x, {\bbchar R}, \Gamma, 0, 1$ \cr
+ 1 & \iid Op & big opertator & $\sum, \int, \bigcup$ \cr
+ 2 & \iid Bin & binary operator & $+, \times, -, \pm, \cup$ \cr
+ 3 & \iid Rel & reations & $=, \ne, \leq, \supseteq, \succsim$ \cr
+ 4 & \iid Open & opening bracket & $\{, (, [, \langle$ \cr
+ 5 & \iid Close & closing bracket & $\}, ), ], \rangle$ \cr
+ 6 & \iid Punct & punctuation & comma \cr
+ & \iid Inner & left-right & \code{\\left...\\right} outputs, see section~\ref[delims]
+}
+\bigskip
+
+There are \ii horizontal/spacing,spacing three space types used
+by the algorithm for horizontal spacing in the math formulas.
+
+\begitems
+* \ii thin/space Thin space: \x`\thinmuskip` primitive register, `\,` macro. Used around Op atoms.
+* \ii medium/space Medium space: \x`\medmuskip` primitive register, `\>` macro. Used around Bin atoms.
+* \ii thick/space Thick space: \x`\thickmusip` primitive register, `\;` macro. Used around Rel atoms.
+\enditems
+
+\puttext 7.8cm -3.7cm {\rotbox{90}{Left atom}}
+\puttext 11.5cm -.5cm {Right atom}
+\puttext 8.2cm -4.4cm {\typosize[9/11]
+\thistable{\def\_enspace{ }}%
+\table {l|8c|} {
+ \omit & Ord & Op & Bin & Rel & Open & Close & Punct & \omit \hfil Inner \crlp{2-9}
+ Ord & 0 & 1 & 2 & 3 & 0 & 0 & 0 & 1 \cr
+ Op & 1 & 1 & & 3 & 0 & 0 & 0 & 1 \cr
+ Bin & 2 & 2 & & & 2 & & & 2 \cr
+ Rel & 3 & 3 & & 0 & 3 & 0 & 0 & 3 \cr
+ Open & 0 & 0 & & 0 & 0 & 0 & 0 & 0 \cr
+ Close & 0 & 1 & 2 & 3 & 0 & 0 & 0 & 1 \cr
+ Punct & 1 & 1 & & 1 & 1 & 1 & 1 & 1 \cr
+ Inner & 1 & 1 & 2 & 3 & 1 & 0 & 1 & 1 \crlp{2-9}
+}}
+
+\hangindent=-8.7cm \hangafter=0
+Ord atoms are printed without spaces between them. The spaces are not
+cumulated, so the rule about spaces mentioned above is only a rough idea.
+The exact rule for horizontal spaces is given for each pairs of atoms
+in the table here. The symbol 0 means no space, 1 thin space, 2 medium space,
+and 3 means thick space.
+
+\hangindent=-8.7cm \hangafter-2
+The Bin atom is automatically transformed to the
+Ord atom if no atom precedes or if Op, Bin, Rel, Open, or Punct atom
+precedes. And it is transformed to the Ord atom if Rel, Close or Punct atom
+follows. This corresponds to the empty cells in the table.
+Why such behavior? Compare \"\hbox{$0-3$}" and \"$-3$". The Bin atom in
+the second case behaves like Ord atom because it is \ii unary/minus {\em unary minus}.
+There is no space between the unary minus and the following object.
+
+All medium spaces and thick spaces and some thin spaces from this table are
+omitted if the <math list> is processed in
+\ii script/style,scriptscript/style script or scriptscript styles
+(smaller size). See section~\ref[styles] about math styles.
+
+You can overwrite the default class derived from the nucleus of the atom by
+\TeX/ primitives \x`\mathord`, \x`\mathop`, \x`\mathbin`, \x`\mathrel`, \x`\mathopen`,
+\x`\mathclose`, \x`\mathpunct` and \x`\mathinner`. They can precede a nucleus of
+the atom and they set the class of the atom.
+For example, `x \mathrel+ y` behaves like `x = y` from a spacing point of view but +
+is printed. Another example: `\mathop{\rm lim} z` creates the atom `lim` in
+roman font of class Op. So, the thin space is inserted between lim and $z$.
+
+There are more special kinds of math atoms: fractions, math accents,
+radicals. They are constructed in a special way (see next sections) but they behave
+like Ord atom in the horizontal spacing algorithm.
+
+\secc[styles] Math styles
+
+When a formula (or a sub-formula) is processed by \TeX/ then one from four
+\ii math/style,display/style,text/style,script/style,scriptscript/style
+styles is active: display style ($D$), text style ($T$), script style ($S$) or
+scriptscript style ($SS$).
+
+\ii T/style,D/style,S/style,SS/style The $T$ style is started in in-line math mode `$...$` and the $D$
+style is started in display math mode `$$...$$`. The first level of exponents or
+subscripts is processed in $S$ style and the second and more levels of
+exponents or indexes are processed in $SS$ style.
+There are special rules for math styles when fractions are constructed, see
+section~\ref[frac].
+
+The $D$ and $T$ styles use basic \ii font/size font size, $S$ uses smaller font size (typically
+70~\%) and $SS$ style uses more smaller font size (typically 50~\%). Next
+levels of \"more smaller fonts" are not used due to classical typographic rules.
+
+The \ii nucleus nucleus of \iid Op atoms (big operators, $\sum$, $\int$, etc.) have typically bigger versions
+of the character shape for $D$ style than for $T$ style.
+So, there are four sizes for such math
+objects: one size for each math style. All other math objects (with non Op
+class) are printed only in three sizes: The sizes for $T$ and $D$ styles are equal.
+
+The \iid Op atom puts its \iid exponent and \iid subscript above and below the nucleus in $D$
+style but right to the nucleus in other styles:
+$$
+ `\sum``_{i=1}^\infty` \quad \hbox{gives}\quad \sum_{i=1}^\infty \hbox{ in $D$ style and}
+ \quad \textstyle \sum_{i=1}^\infty \hbox{ in $T$ style}.
+$$
+This default behavior of the Op atom
+can be modified by placing \x`\limits` or \x`\nolimits` or
+\x`\displaylimits` \TeX/ primitive just after its nucleus before the constructors
+of exponent and/or index. The `\nolimits` puts exponent and subscript right
+to the nucleus (regardless of the current style) and `\limits` puts these
+objects above and below the nucleus (regardless of the current style). There
+can be more such primitives in a queue (due to a macro expansion, for
+instance). Then the last primitive in the queue wins.
+If the last primitive is \x`\displaylimits` then
+the default behavior is processed regardless there are \x`\limits` or \x`\nolimits`
+before it.
+$$
+ `\sum\nolimits``_{i=1}^\infty` \quad \hbox{gives}\quad \sum\nolimits_{i=1}^\infty
+ \hbox{ in $D$ style and}\quad \textstyle\sum\nolimits_{i=1}^\infty
+ \hbox{ in $T$ style}.
+$$
+Atoms of all other classes have their exponents and/or subscripts only right
+to their nucleus without any exception.
+
+The primitives \x`\displaystyle`, \x`\textstyle`, \x`\scriptstyle` and
+\x`\scriptscriptstyle` set the given style regardless the default rules. For
+example, you can create a formula in in-line math mode and in $D$ style by
+`$\displaystyle <fomula>$` or a formula in display mode and $T$ style can be printed
+by `$$\textstyle <fomrula>$$`.
+
+If a subformula is placed below something (below a line from root symbol,
+below a fraction line), then the processed style $D, T, S$ or $SS$ is
+\ii cramped/style {\em cramped}.
+The exponents are positioned slightly lower than in
+\ii non-cramped/style non-cramped style. The selectors `\displaystyle`\,\dots
+`\scriptscriptstyle` mentioned above select non-cramped style. The
+non-cramped style is selected when math mode starts too.
+\new \OpTeX/
+You can select a cramped style by the macro \x`\cramped` at the start of the
+math formula or after the math-style selectors: `\scriptstyle\cramped` for
+example.
+
+Several macros need to know what math style is currently processed (for
+example they need to draw something in an appropriate size). But it
+not possible simply due to the syntax of fractions (section~\ref[frac]).
+This syntax requires to process all math lists in two steps: the first step
+expands all macros and creates structured data of processed math list. The
+second step reads the output of the first step, switches between math
+styles and creates definitive output. So, macros (working in the first step)
+cannot know the current math
+style because it is set only in the second step. \TeX/ supports the primitive
+\x`\matchchioce``{<D>}{<T>}{<S>}{<SS>}` which prepares four math lists in the
+first step and only one of these four lists are used in the second step. We
+can put different macros into each of the four parameters of `\mathchoice`.
+Plain \TeX/ supports the macro \x`\mathpalette` which gives a more comfortable
+interface of \x`\mathchoice` to the macro programmer.
+The cramped/non-cramped variants of the current style are kept when `\mathchioce`
+is used.
+
+\new \OpTeX/
+We describe another interface for creating macros depending on the current
+style. You can use \x`\mathstyles``{<math list>}`. It
+behaves like `{<math list>}`, moreover, you can use the following commands inside such
+<math list>:
+\begitems
+* The macro \x`\currstyle`. It expands to
+ `\displaystyle`, `\textstyle`,
+ `\scriptstyle` or `\scriptscriptstyle` depending on the current math style
+ when the `\mathstyles` was opened.
+* The \x`\dobystyle``{<D>}{<T>}{<S>}{<SS>}` is expandable macro. It expands its
+ parameter `<D>`, `<T>`, `<S>` or `<SS>` depending on the current math style
+ when `\mathstyles` was opened.
+* The value of the \x`\stylenum` register is 0, 1, 2 or 3
+ depending on the current math style when `\mathstyles` was opened.
+\enditems
+%
+Example of usage of \x`\mathstyles`:
+\def\mysymbol{\mathbin\mathstyles
+ {\kern1pt\vrule height\mysymbolA width\mysymbolA\kern1pt}}
+\def\mysymbolA{\dobystyle{5pt}{5pt}{3.5pt}{2.5pt}}
+\begtt
+\def\mysymbol{\mathbin\mathstyles
+ {\kern1pt\vrule height\mysymbolA width\mysymbolA\kern1pt}}
+\def\mysymbolA{\dobystyle{5pt}{5pt}{3.5pt}{2.5pt}}
+Test: $a\mysymbol b_{c \mysymbol d}$ or $a\mysymbol b\over c$.
+\endtt
+This example gives Test: $a\mysymbol b_{c \mysymbol d}$ or $a\mysymbol b\over c$.
+
+The \x`\mathstyles` macro mentioned above uses \TeX/ primitive \x`\mathchoice`, so it
+creates four math lists and only one is used. It may take more
+computer time in special cases.
+\new Lua\TeX/
+Lua\TeX/ supports the \x`\mathstyle` primitive
+(no \"`s`" at the end of this control sequence) which
+expands to values 0 to 7 depending on the current style:
+$D, D', T, T', S, S', SS, SS'$
+(where $X'$ means cramped variant of the style). This primitive does
+not use `\mathchoice` but it simply ignores the fraction syntax, so
+`$a\mysymbol b\over c$` cannot work if `\mysymbol` is defined using the `\mathstyle`
+primitive. See section 7.3.1 of Lua\TeX/ documentation for more information.
+
+
+\secc[frac] Fractions
+
+The \iid fraction can be constructed by `{<numerator>`\x`\over``<denominator>}`. If the
+fraction is only a single object in the whole math mode (between dollars),
+you need not use the outer braces, so you can write `$1\over2$` to get $1\over2$.
+
+The \ii numerator,denominator <numerator> and <denominator> are printed in \"smaller" math style than
+current math style. More exactly the following schema is used.
+$D$: $T\over T$., $T$: $S\over S$, $S$: $SS\over SS$, $SS$: $SS\over SS$.
+For example
+$$
+ `{a+b \over c}` \quad \hbox{is printed as }
+ {a+b\over c} \hbox{ in $D$ style and as }
+ \textstyle {a+b\over c} \hbox{ in $T$ style}.
+$$
+
+The \LaTeX/ macro \x`\frac``{<numerator>}{<denominator>}` is not supported in
+Plain \TeX/ nor in \OpTeX/ but you can define such macro if you want.
+
+The syntax with \x`\over` is more preferred because it is more human-readable
+notation. You can write the fraction in the same manner as you can read it.
+You can compare: `$1\over2$` (one over two) with `$\frac12$` (frac twelve).
+
+Besides the `\over` primitive, there are analogical \TeX/ primitives which
+create \"generalized" fractions. The result is similar to `{<above>\over <below>}`
+but there is something extra:
+
+\begitems
+* `{<above>`\x`\atop``<below>}` does `{<above>\over<below>}` but
+ without the fractional rule.
+* `{<above>`\x`\above``<dimen><below>}` creates fractional rule with <dimen> thickness.
+* `{<above>`\x`\overwithdelims``<delim-l><delim-r><below>}` adds the <delim-l> left to
+ the fraction and the <delim-r> right to the fraction.
+* `{<above>`\x`\atopwithdelims``<delim-l><delim-r><below>}` is analogical to
+ `\overwithdelims` but without fractional rule.
+* `{<above>`\x`\abovewithdelims``<delim-l><delim-r><dimen><below>}` behaves as
+ `\overwithdelims` but the fractional rule has <dimen> thickness.
+\enditems
+%
+The `\...withdelims` variants read <delim-l> and <delim-r>, they must be
+declared as {\em math delimiter} in \TeX. They are vertically scalable
+math objects, typically brackets. See section~\ref[delims] for more
+information about math delimiters. Example:
+$$
+ `{n \atopwithdelims() k}`\quad \hbox{ creates }
+ {n \choose k} \hbox{ in $D$ style and }
+ \textstyle {n \choose k} \hbox{ in $T$ style}.
+$$
+The \x`\choose` macro is defined by `\def\choose{\atopwithdelims()}`, so the
+user can write `{n\choose k}` in order to get binomial coefficients.
+
+
+\secc[delims] Vertically scalable objects: math delimiters
+
+The vertically scalable objects are called \ii delimiters {\em delimiters}. For example,
+all types of brackets are declared as delimiters.
+This means that you can use a bracket in arbitrary
+vertical size.\fnote{
+This is not exactly true, because traditional typography says that they
+cannot be scaled continuously but by visible steps. This means that there is
+a sequence of increasing brackets in the font, the reader must see a difference
+between every two sizes of brackets.}
+
+The following objects are declared as delimiters (i.e.\ vertically scalable):
+
+\bigskip
+{\tt \adef!{\bslash}
+\table{l 14c}{
+ \rm source: & ( & ) & [ & ] & \code{\\\{} & \code{\\\}} & /
+ & !backslash & !langle & !rangle
+ & | & \code{\\|} & \cr
+ \rm output: &$($&$)$&$[$&$]$& $\{$ & $\}$ & $/$
+ & $\backslash$ & $\langle$ & $\rangle$
+ & $|$ & $\|$
+}
+\medskip
+\table{l 14c}{
+ \rm source: & !lfloor & !rfloor & !lceil & !rceil \cr
+ \rm output: & $\lfloor$ & $\rfloor$ & $\lceil$ & $\rceil$
+}
+\medskip
+\table{l 14c}{
+ \rm source: & !uparrow & !Uparrow & !dowarrow & !Downarrow
+ & !updownarrow & !Updownarrow \cr
+ \rm output: & $\uparrow$ & $\Uparrow$ & $\downarrow$ & $\Downarrow$
+ & $\updownarrow$ & $\Updownarrow$
+}}
+\bigskip
+\noindent \new Unicode
+If you can produce the characters $\langle$, $\rangle$,\fnote
+ {Do not confuse $\string<, >$ and $\langle, \rangle$. The first pair are Rel atoms
+ with meaning \"less than" or \"greater than", but the second pair are special
+ types of brackets. They are not directly available at computer keyboards without
+ using a keyboard macro.}
+$\lfloor$, $\rfloor$, ... $\updownarrow$, $\Updownarrow$
+directly in your text editor then you can use these Unicode characters in your source instead of control
+sequences `\langle`, `\rangle`, `\lfloor`, `\rfloor` ... `\updownarrow`, `\Updownarrow`.
+For many users (including me), there is more simple to type `\lfloor` than to find
+how to create the $\lfloor$ character in my text editor. Note that there exist
+text editors (Emacs, for example)
+enabling you to type `\lfloor` and this control sequence is immediately
+converted to the $\lfloor$ Unicode character.
+Your source text looks pretty and you can use classical \TeX/ sequences.
+
+\new Unicode
+There are more \ii delimiters delimiters, but it heavily depends on loaded Unicode Math
+font. For example, this document is printed in `latinmodern-math` font and
+there are six more delimiters `\lBrack`~$\lBrack$, `\rBrack`~$\rBrack$,
+`\lAngle` $\lAngle$, `\rAngle` $\rAngle$, `\lgroup` $\lgroup$, `\rgroup` $\rgroup$.
+See section~\ref[objects] for table of all Unicode symbols for math typesetting.
+
+Arbitrary tall formula can be surrounded by a pair of delimiters using
+\x`\left` and \x`\right` \TeX/ primitives: `\left<delim> <formula> \right<delim>`.
+The delimiters are scaled to the height and depth of the <formula>
+and vertically centered to the {\em math axis}.\fnote
+{Math axis is a horizontal line passing through the center of symbols $+$
+ and $-$. All vertically scalable objects are vertically centered with
+ respect to this axis.}
+Example:
+$$
+ `+ \left\{ \sum_{i=1}^\infty x_i \right)` \quad \hbox{ gives }
+ + \left\{ \sum_{i=1}^\infty x_i \right).
+$$
+The pair `\left<delim> <formula> \right<delim>` creates the formula in a
+\TeX/ group. Such group can be nested with another groups.
+Each `\left` must have its `\right` counterpart at the same group level.
+If you don't want to create visible delimiter, use dot instead <delim>.
+Example:
+$$
+ `\left. \int``_0^t e^{x^2}\,dx\, \right|_{t=42}` \quad \hbox{ gives }
+ \left. \int_0^t e^{x^2} \,dx\, \right|_{t=42}
+$$
+
+\new e\TeX/
+You can use \x`\middle``<delim>` inside the <formula> which is surrounded by
+`\left...\right`. Then the given <delim> is scaled to the same size like
+their `\left` and `\right` counterparts.
+
+When a delimiter is used without `\left` nor `\right` prefix, then it is the
+Open, Close, Ord or Bin atom by its natural meaning:
+$(, [, \{, \ldots, \lfloor, \lceil$ are Open atoms,
+$], ], \}, \ldots, \rfloor, \rceil$ are Close atoms,
+$/, \backslash, |, \|$ are Ord atoms and
+$\uparrow, \Uparrow, \ldots, \Updownarrow$ are Bin atoms. You can overwrite
+this default setting, for example `\mathclose(`. If delimiters are used with
+`\left` and `\right` prefixes then `\left<delim>` behaves like Open atom,
+`\right<delim>` behaves like Close atom and the math list
+`\left<delim><formula>\right<delim>` is encapsulated as a single Inner atom.
+The `\middle<delim>` behaves like Open atom at its left side and like Close
+atom at its right side.
+
+The sequence of increasing delimiters can be printed by the following
+macros:
+$$
+ `(` \to (,\quad \x`\big``(` \to \big(,\quad \x`\Big``(` \to \Big(,\quad
+ \x`\bigg``(` \to \bigg(, \quad \x`\Bigg``(` \to \Bigg(.
+$$
+The `\Bigg<delim>` is not the maximal size of the bracket. Try
+`\left(\vbox to5cm{}\right.`, for example. You can see that the font
+\"cheats" from certain sizes, because there are not all infinity number of
+sizes of brackets drawn in the font, of course.
+
+The `\big<delim>` creates Ord atom. We need to create Open atom
+for opening bracket and Close atom for closing bracket more often.
+Then we can use macros
+\x`\bigl``<delim>`,
+\x`\Bigl``<delim>`,
+\x`\biggl``<delim>`,
+\x`\Biggl``<delim>` for creating Open atoms and
+\x`\bigr``<delim>`,
+\x`\Bigr``<delim>`,
+\x`\biggr``<delim>`,
+\x`\Biggr``<delim>` for creating Close atoms. Unfortunately, the source is not
+too attractive when more sizes of brackets are used, but typographic
+traditions say that we have to distinguish brackets by the size in math
+mode if they are in equal types:
+$$
+ `\Bigl( f\bigl( 2(x+y) + z\bigr) \Bigr)'` \quad \hbox{gives }
+ \Bigl(f\bigl(2(x+y)+z\bigr)\Bigr)'.
+$$
+
+\secc Horizontally scalable objects: math accents
+
+Arbitrary wide formula can be covered by \ii math/accent,accent {\em scalable math accent}.
+Example:
+$$
+ \x`\overrightarrow`` {a+b+c+d+e+f}` \quad \hbox{gives } \overrightarrow {a+b+c+d+e+f}.
+$$
+The usage is: control sequence of selected math accent followed by `{<math list>}`.
+
+Standard scalable math accents are:
+\x`\overline` $\overline{abc}$,
+\x`\overbrace` $\overbrace{abc}$,
+\x`\overrightarrow` $\overrightarrow{abc}$,
+\x`\overleftarrow`~$\overleftarrow{abc}$,
+\x`\underline` $\underline{abc}$,
+\x`\underbrace` $\underbrace{abc}$.
+
+An Op atom is created. The exponents and subscripts are centered above
+and below the nucleus of this atom (regardless of the current style). Example:
+$$
+ `\overbrace {b\cdot b\cdot b \cdots b}^{k\times}` \quad \hbox{gives }
+ \overbrace {b\cdot b\cdot b \cdots b}^{k\times}
+$$
+There are scalable accents with a limited maximum width:
+\x`\widehat` $\widehat{abc}$ and \x`\widetilde` $\widetilde{abc}$. If the
+formula is wider than the font can cover then the widest variant from the
+font is used and it is horizontally centered.
+
+\new Unicode
+There are more scalable accents in Unicode math fonts:
+\x`\overparen` $\overparen{abc}$, \x`\underparen` $\underparen{abc}$,
+\x`\overbracket` $\overbracket{abc}$, \x`\underbracket` $\underbracket{abc}$,
+\x`\overleftrightarrow` $\overleftrightarrow{abc}$,
+\x`\overleftharpoon` $\overleftharpoon{abc}$,
+\x`\overrightharpoon` $\overrightharpoon{abc}$,
+
+
+\secc Fixed math accents
+
+Fixed \ii math/accent,accent math accents can be applied to single math object or to the `{<math list>}`.
+The accent is centered (with respect of slanting axis) and the result is a
+nucleus of Ord
+atom. For example `\dot x` gives $\dot x$. The list of fixed math accents
+follows: \x`\acute`` x` $\acute x$, \x`\bar`` x` $\bar x$,
+\x`\breve`` x` $\breve x$, \x`\check`` x` $\check x$,
+\x`\dot`` x` $\dot x$, \x`\ddot`` x` $\ddot x$,
+\x`\grave`` x` $\grave x$, \x`\hat`` x` $\hat x$,
+\x`\vec`` x` $\vec x$, \x`\tilde`` x` $\tilde x$.
+
+\new Unicode
+The additional fixed accents depends on used Unicode math font. The
+`latinmodern-math` supports:
+\x`\ovhook`` x` $\ovhook x$, \x`\ocirc`` x` $\ocirc x$,
+\x`\leftharpoonaccent`` x` $\leftharpoonaccent x$, \x`\rightharpoonaccent`` x` $\rightharpoonaccent x$,
+\x`\dddot`` x` $\dddot x$, \x`\ddddot`` x` $\ddddot x$,
+\x`\widebridgeabove`` x` $\widebridgeabove x$, \x`\asteraccent`` x` $\asteraccent x$.
+
+There exist one special math accent `'` (single quote, ASCII 39)
+which can be appended after a symbol like this: `f'`
+and it creates $f'$ (typical meaning is the
+derivation of the given function). You can put more such accents, for
+example `g'''` gives $g'''$.
+
+\secc Roots
+
+There is a macro \x`\sqrt``{<math list>}` to create \ii square/root square root. For example:
+$$
+ `\sqrt{\sqrt{\sqrt{x+1}+1}+1}` \quad \hbox{gives }
+ \sqrt{\sqrt{\sqrt{x+1}+1}+1}
+$$
+The \ii root $n$-th root is created by the macro \x`\root`` <n>\of{<math list>}`. For
+example `\root k+1\of x` gives $\root k+1\of x$.
+
+\secc Math alphabets
+
+Letters \ii math/alphabet $a\dots z$, $A\dots Z$ and $\alpha$\dots$\omega$ are printed in italic
+in math mode. This follows the traditional typographic rule.
+All other math symbols, digits, and uppercase Greek letters must be
+upright.\fnote
+{French typographic convention says that uppercase Greek letters have to be
+ in italic too. Use `\_itGreek` declaration in this case.}
+These rules are independent of the current variant of surrounding text font.
+
+If we want to use the letters or digits
+in another than this default shape, then we can use
+\ii math/alphabet/selector {\em math alphabet selectors}:
+\x`\mit`, \x`\rm`, \x`\it`, \x`\bf`, \x`\cal`.
+\new \OpTeX/
+\OpTeX/ supports more such selectors \x`\script`, \x`\frak`, \x`\bbchar`, \x`\bi`, see
+section 1.3.3 in the \OpTeX/ documentation. The math selectors have local
+validity in the group.
+
+The control sequences \x`\rm`, \x`\it`, \x`\bf`, and \x`\bi` act as variant selectors
+of fonts in non-math mode (text mode) and they act
+as math alphabet selectors in math mode. This \"overlaying" concept
+is given by Plain \TeX/. Example: math operators lim, sin, cos,
+log, etc.\ must be printed unslanted. We are using `\lim`, `\sin`, `\cos`,
+`\log` etc.\ in math mode in order to comply with this typographic convention. For
+example `\sin` is defined as:
+\begtt
+\def\sin {\mathop{\rm sin}\nolimits}
+\endtt
+The `\rm` is used here as math alphabet selector, no variant selector of
+text fonts.
+
+The list of all predefined `\rm`-like math operators follows: \x`\arccos`,
+\x`\arcsin`, \x`\arctan`, \x`\arg`, \x`\cos`, \x`\cosh`, \x`\cot`, \x`\coth`, \x`\deg`.
+\x`\det`, \x`\dim`, \x`\exp`, \x`\gcd`, \x`\hom`, \x`\inf`, \x`\ker`, \x`\lg`, \x`\lim`,
+\x`\liminf`, \x`\limsup`, \x`\ln`, \x`\log`, \x`\max`, \x`\min`, \x`\Pr`, \x`\scs`, \x`\sin`,
+\x`\sinh`, \x`\sup`, \x`\tan`, \x`\tanh`, You can define another such operator
+analogically.
+
+\new Unicode
+Unicode font can include the following math alphabets:
+\begtt \typosize[9/11]
+\_rmvariables % upright letters A-Z, a-z
+\_bfvariables % bold letters A-Z, a-z
+\_itvariables % italic letters A-Z, a-z
+\_bivariables % bold italic letters A-Z, a-z
+\_calvariables % calligraphic letters A-Z, a-z
+\_bcalvariables % calligraphic letters A-Z, a-z
+\_frakvariables % fraktur A-Z, a-z
+\_bfrakvariables % bold fraktur A-Z, a-z
+\_sansvaraibales % sans serif letters A-Z, a-z
+\_bsansvaraibales % bold sans serif letters A-Z, a-z
+\_isansvaraibales % slanted sans serif letters A-Z, a-z
+\_bisansvaraibales % bold slanted sans serif letters A-Z, a-z
+\_ttvariables % monospace, typewriter letters A-Z, a-z
+\_bbvariables % double struck A-Z, a-z
+\_rmdigits % upright digits 0..9
+\_bfdigits % bold digits 0..9
+\_sansdigits % sans serif digits 0..9
+\_bsansdigits % bold sans serif digits 0..9
+\_ttdigits % monospace typewriter digits 0..9
+\_bbdigits % double-struck digits 0..9
+\_rmgreek % upright Greek letters \alpha-\omega
+\_itgreek % slanted Greek letters \alpha-\omega
+\_bfgreek % bold Greek letters \alpha-\omega
+\_bigreek % bold italic Greek letters \alpha-\omega
+\_bsansgreek % bold sans serif Greek letters \alpha-\omega
+\_bisansgreek % bold slanted snas serif Greek letters \alpha-\omega
+\_itGreek % slanted Greek letters \Alpha-\Omega
+\_bfGreek % bold Greek letters \Alpha-\Omega
+\_biGreek % bold italic Greek letters \Alpha-\Omega
+\_bsansGreek % bold sans serif Greek letters \Alpha-\Omega
+\_bisansGreek % bold slanted snas serif Greek letters \Alpha-\Omega
+\endtt
+%
+Not all Unicode math fonts include all math alphabets listed here. Typically,
+the lowercase letters of calligraphic shape and all letters of
+bold calligraphic shape are missing.
+
+\new \OpTeX/
+\OpTeX/ defines internal math alphabet selectors as mentioned in the
+previous listing of math alphabets and sets as default:
+\begtt \typosize[10/12]
+\_itvariables \_rmdigits \_itgreek \_rmGreek
+\endtt
+%
+Moreover, it defines the alphabet selectors at user level (see section 1.3.3
+of the \OpTeX/ manual). For example
+\begtt \typosize[10/12]
+\def\rm {\_tryload\_tenrm \_inmath{\_rmavariables \_rmdigits}}
+\endtt
+%
+The first part
+\new \OpTeX/
+`\_tryload\_tenrm` is applicable for text fonts and the
+`\_inmath` part is processed only in math mode and sets the math alphabets.
+You can see the file `unimath-codes.opm` where all user-level selectors are
+defined. You can redefine them. For example, \OpTeX/ defines `\bf` as a math
+alphabet selector that selects sans serif bold in math. This is the common
+notation for vectors, tensors, and matrices. If you dislike this, then you can define:
+\begtt \typosize[10/12]
+\def\bf {\_tryloadbf\_tenbf \_inmath{\_bfvariables\_bfdigits\_bfgreek\_bfGreek}}
+\endtt
+
+
+\secc[objects] List of single math objects
+
+\new Unicode
+All \ii single/math/object single math objects are listed in the `unimath-table.opm` or
+`unicode-math-table.tex` file. You can
+look into this file. The codes, \TeX/ sequences, classes, and comments
+for all possible math codes are here. Maybe, your Unicode math font which is loaded
+does not support all these codes.
+\new \OpTeX/
+You can try all codes of the currently loaded font by
+\begtt
+\input print-unimath.opm
+\endtt
+The `unimath-table` is printed with characters available in the loaded font.
+\new \OpTeX/
+If the character is unsupported by the font then the slot is empty and only
+\TeX/ sequence and the class of the code is printed in the table.
+For example, this document loads `latimodern-math.otf` font.
+And the result from `\input print-unimath.opm` looks like the following ten
+pages.
+
+Unsupported characters can be replaced by characters from other Unicode math
+font, see
+\ulink[http://petr.olsak.net/optex/optex-tricks.html\#addumathfont]{OpTeX trick 0030}
+or section~\ref[newfam].
+
+
+\label[listA]\wlabel{}
+
+\bigskip
+\input print-unimath.opm
+\bigskip
+
+\label[listB]\wlabel{}
+
+It isn't very comfortable to find something in the previous table
+if you know the shape. You can try the online web tool \ii Detexify
+\ulink[https://detexify.kirelabs.org/classify.html]{Detexify}.
+You can draw the symbol here and the suggestion of \TeX/ sequence is
+printed.
+
+\def\g#1{{\tt\string#1}~$#1$}
+
+The generic \TeX/ sequences for the Greek letters \ii Greek/letters can be used:
+\g\alpha, \g\beta, \g\gamma, \g\delta, \g\varepsilon, \g\zeta, \g\eta,
+\g\theta, \g\iota, \g\kappa, \g\lambda, \g\mu, \g\nu, \g\xi,
+\g\omicron, \g\pi, \g\rho, \g\varsigma, \g\sigma, \g\tau, \g\upsilon,
+\g\varphi, \g\chi, \g\psi, \g\omega, \g\vardelta, \g\epsilon, \g\vartheta,
+\g\varkappa, \g\phi, \g\varrho, \g\varpi\
+and
+\g\Alpha, \g\Beta, \g\Gamma, \g\Delta, \g\Epsilon, \g\Zeta, \g\Eta, \g\Theta,
+\g\Iota, \g\Kappa, \g\Lambda, \g\Mu, \g\Nu, \g\Xi, \g\Omicron, \g\Pi, \g\Rho,
+\g\Sigma, \g\Tau, \g\Upsilon, \g\Phi, \g\Chi, \g\Psi, \g\Omega.
+
+\new \OpTeX/
+The variant of Greek letters in the output
+(upright, italic, bold, bold sans serif, etc.)
+written by the sequences `\alpha`, `\beta` etc. depends on the math alphabet
+selected by `\_itgreek`, `\_rmgreek`, etc. selectors. The user-level
+selectors `\bf` and `\bi` set `\_bsansgreek` and `\_bisansgreek`, so
+`{\bi\delta}` produces $\bi\delta$.
+
+\new Unicode
+All characters available in the math font can be accessed by \TeX/ control
+sequence or by directly using the Unicode character in the document source.
+Example:
+\begtt \adef/{}
+$$
+ \sum/_{k=0}^\infty e^{(\alpha+i\beta/_k)} =
+ e^\alpha \sum/_{k=0}^\infty e^{i\beta/_k} =
+ e^\alpha \sum/_{k=0}^\infty (\cos\beta/_k + i\sin\beta/_k).
+$$
+\endtt
+or
+\begtt \ttspec
+$$
+ ∑_{k=0}^∞ e^{(α + iβ_k)} = e^α ∑_{k=0}^∞ e^{iβ_k}
+ = e^α ∑_{k=0}^∞ (\cos β_k + i\sin β_k).
+$$
+\endtt
+both gives the same result:
+$$
+ ∑_{k=0}^∞ e^{(α + iβ_k)} = e^α ∑_{k=0}^∞ e^{iβ_k}
+ = e^α ∑_{k=0}^∞ (\cos β_k + i\sin β_k).
+$$
+\medskip
+
+
+\sec Other specialities
+
+\secc The `\not` prefix
+
+You can apply \x`\not` before a following math object.
+The slash $/$ is overprinted such math object, for example
+`$a \not= b$` gives $a \not= b$.
+
+\new \OpTeX/
+If there exists a direct Unicode character for the negation of a relation symbol
+(for example `\ne` creates $\ne$ directly as a character U+2260)
+then `\not<char>` expands to appropriate Unicode character.
+For example `\not=` expands to `\ne` or `\not\in` expands to `\notin`.
+If such character does not exist then
+the centered $/$ is overprinted over the next character.
+
+
+\secc The `\buildrel` macro: text over the relation
+
+The macro \x`\buildrel`` <text>\over <relation>` creates a new atom Rel with the
+<relation> and with the smaller <text> above this <relation>. Example:
+`$M \buildrel\rm def\over= X\cup Y$` gives $M\buildrel\rm def\over= X\cup Y$.
+
+
+\secc Spaces
+
+Spaces between atoms are created automatically as were mentioned in
+section~\ref[class]. But sometimes you have to help \TeX/ to create
+appropriate space. You can use following macros:
+
+\begitems
+* `\,` is \ii thin/space {\em thin space} used around Op atoms, after comma, etc.: \ss\,,
+* `\!` is \ii negative/thin/space negative thin space,
+* `\>` is \ii medium/space {\em medium space} used around Bin atoms: \ss\>,
+* `\;` is \ii thick/space {\em thick space} used around Rel atoms: \ss\;,
+* \x`\quad` is \ii em/space {\em em space}: \ss\quad,
+* \x`\qquad` is \ii double/em/space {\em double em space}: \ss\qquad.
+\enditems
+%
+Of course, you can use \ii direct/space {\em direct space} {\visiblesp`\ `} which is \TeX/ primitive
+and gives interword space: \ss\ \
+or you can use \x`\hskip`` <value>` to put arbitrary space.
+
+The space size of `\,`, `\!` resp. `\>`, resp. `\;` is given by
+`\thinmuskip`, resp. `\medmuskip`, resp. `\thickmuskip` values. You can see
+in the `plain.tex` file that these default values differ very little in their basic
+size but there is no stretchability/shrinkability in the `\,` space, there is small
+stretchability in the `\>` space, and more stretchability in the `\;` space.
+
+The registers \x`\thinmuskip`, \x`\medmuskip`, and \x`\thickmuskip` store so-called
+\ii mu/values {\em mu values} given by math unit `mu`. It is 1/18 em and this unit depends
+on the current font size used in the math formula ($S$ or $SS$ styles use
+smaller font size, the `mu` unit is smaller here). You can use \x`\muskip`
+instead `\hskip` or \x`\mukern` instead `\kern` if you want to use this
+special mu unit. It is allowed only in math mode.
+
+The \x`\quad` and \x`\qquad` spaces have fixed width and they can be used in text
+mode too. (\OpTeX/ allows to use `\,` in text mode too).
+Use `\quad` or `\qquad` if you want to separate more formulas created in
+single math mode. Examples of typical usage of spaces:
+
+\begtt \adef/{} \typosize[10/12]
+$$ \alpha\,(x+y), \qquad \int/_a^b \!\! f(x)\,{\rm d}x, \qquad \Gamma/_{\!i}. $$
+\endtt
+$$ \alpha\,(x+y), \qquad \int_a^b \!\! f(x)\,{\rm d}x, \qquad \Gamma_{\!i}. $$
+
+
+\secc Texts in math mode
+
+If you write `$Hello world!$` (i.e.\ Hello world in math mode), then you get
+$Hello world!$. It is interpreted as the product of variables $H$, and $e$, and
+$l^2$, and $o$, etc., followed by the symbol ! used for factorial.
+The non-ASCII letters (with accents) don't work at all because they are
+never used as symbols for variables. Spaces are ignored.
+
+If you want to write a short text in the math mode, then you can use
+`\hbox{<text>}`. The \x`\hbox` primitive initializes text mode regardless of
+the \"outer mode". Example:
+
+\begtt \adef/{} \typosize[10/12]
+$$ \sum/_{n=0}^\infty (-1)^n a_n \hbox{ converges, if $a_n\searrow 0$.} $$
+\endtt
+$$ \sum_{n=0}^\infty (-1)^n a_n \hbox{ converges, if $a_n\searrow 0$.} $$
+%
+Note the space before the word \"converges". The space before `\hbox` is
+irrelevant. Second notice: the example shows the text mode inside math mode
+and the in-line math mode inside this text mode. The same result can be
+produced by:
+
+\begtt \adef/{} \typosize[10/12]
+$$ \sum/_{n=0}^\infty (-1)^n a_n \hbox{ converges, if } a_n\searrow 0. $$
+\endtt
+%
+The difference can be visible if the formula $a_n\searrow 0$ includes a
+fraction, for example ${1\over2}a_n\searrow 0$. The first example prints the
+fraction in the text style and the second example prints it in the display style.
+
+The disadvantage of `\hbox` is that it starts in the text mode independently
+of the current style, but we want to use smaller font in $S$ or $SS$ styles.
+\new \OpTeX/
+You can use \x`\mathbox``{<text>}` in such situations. This macro behaves like
+`\hbox` but the text is appropriately smaller in $S$ and $SS$ styles.
+Example:
+$$
+ `{\mathbox{cena}\over\mathbox{výkon}}` \quad \hbox{gives }
+ {\mathbox{cena}\over\mathbox{výkon}} \hbox{ in $D$ style and }
+ \textstyle {\mathbox{cena}\over\mathbox{výkon}} \hbox{ in $T$ style.}
+$$
+Note that $\mathbox{cena}\over\mathbox{výkon}$ means $\rm price\over performance$
+and you can write `$\rm price\over performance$` when you are using only
+words without spaces and accented letters. But phrases with spaces or accented letters
+should be printed in text mode using `\hbox` or `\mathbox`.
+
+\secc `\vcenter`
+
+The \x`\vcenter` primitive behaves like `\vbox`, but it can be used only in
+math mode and its result is vertically centered to the math axis.
+For example, matrices, are created by tables in `\vcenter`.
+
+All big objects in math formula are centered on the math axis and the baseline is
+ignored. In the following example, we create a new big math operator by
+`\vcenter`:
+\begtt
+$$
+ \def\myop#1{\mathop{\vcenter{\frame{\vbox to2em{\vss\hbox{ $#1$ }\vss}}}}}
+ F(x) = \myop{x(i)}_{i=1}^\infty
+$$
+\endtt
+\vskip-2em
+$$
+ \def\myop#1{\mathop{\vcenter{\frame{\vbox to2em{\vss\hbox{ $#1$ }\vss}}}}}
+ F(x) = \myop{x(i)}_{i=1}^\infty
+$$
+
+
+\secc Three dots
+
+You can write `$1,2,\dots,n$` to get $1,2,\dots,n$. The \x`\dots` macro puts
+thin space between dots and after the last dot, so the five object: comma,
+dots, comma are exactly equidistant.
+
+Typographic conventions say that you have to use the repeating symbol
+before and after three dots (comma in the previous example) and the three dots
+should be at baseline if the repeating symbol is at baseline. Or they should be
+at the math axis if the repeating symbol is at the math axis. We have to use \x`\cdots` instead
+`\dots` in the second case. Example:
+$$
+ `a_1, a_2, \dots, a_n, \quad a_1 + a_2 + \cdots + a_n` \qquad
+ a_1, a_2, \dots, a_n, \quad a_1 + a_2 + \cdots + a_n
+$$
+\vskip-.7em
+There are \x`\vdots` $\vdots$, \x`\ddots` $\ddots$ and \x`\adots` $\adots$ which
+can be used in matrices.
+
+\new Unicode
+Three dots like the output of the `\dots` macro are present as a single
+character in fonts too. This character is called ellipsis. Font designers
+typically suggest this character with smaller spaces between dots than we
+need in math mode. So the rule about equidistant \"comma, three dots, comma"
+is not met when this character is used. You can try `$1,2,\unicodeellipsis, n$`
+and `$1+2+\unicodecdots + n$`. You get $1,2,\unicodeellipsis, n$ and
+$1+2+\unicodecdots + n$. If you feel that this is better, then you can set:
+`\let\dots=\unicodeellipsis \let\cdots=\unicodecdots`.
+
+The Unicode fonts includes compact variants `\unicodevdots` $\unicodevdots$,
+`\unicodeddots` $\unicodeddots$ and `\unicodeadots` $\unicodeadots$ too.
+
+
+\secc Phantoms and `\smash`
+
+The \x`\phantom``{<math list>}` macro creates an invisible subformula equal to
+the formula generated by <math list>. It has its size, so it can
+interfere with surrounding visible subformulas. This macro is very useful for
+aligning with special requirements. Examples are shown in
+sections~\ref[matrix] and~\ref[elines].
+
+The \x`\vphantom``{<math list>}` has only its vertical size (i.e.\ its height
+and depth), the width is zero. The \x`\hphantom``{<math list>}` has only its
+width.
+
+The \x`\smash``{<math list>}` is the opposite of `\vphantom`. It creates visible
+subformula but it has only its width. Its height and depth are zero.
+
+The result of these macros is the nucleus of an Ord atom. You can use `\mathop`,
+`\mathbin`, etc. primitives to change this class. For example
+`$a\mathrel{\phantom{=}}b$` creates the same formula as $a=b$ with the same
+distance from $a$ to $b$ but without the equal sign:
+$a\mathrel{\phantom{=}}b$.
+
+These macros work in the text mode too. Then their argument is
+the <horizontal list>.
+
+
+\sec Structured objects
+
+\secc[matrix] Matrices
+
+The macro \x`\matrix``{<data>}` creates a vertically centered table of items.
+The <data> includes <items> separated by `&` and rows are separated by
+`\cr`. The number of columns and rows are unlimited. Columns are printed centered
+and separated by the `\quad` space. The vertically scalable
+brackets around the table are not printed. You can use \x`\pmatrix``{<data>}`
+instead `\matrix{<data>}`: the vertically scalable parentheses () are inserted
+around the table. Examples:
+
+\begtt
+$$
+ {\bf A} = \pmatrix{a_{1,1} & a_{1,2} & \cdots & a_{1,n} \cr
+ a_{2,1} & a_{2,2} & \cdots & a_{2,n} \cr
+ \vdots & \vdots & \ddots & \vdots \cr
+ a_{m,1} & a_{m,2} & \cdots & a_{m,n} },
+$$
+\endtt
+$$
+ {\bf A} = \pmatrix{a_{1,1} & a_{1,2} & \cdots & a_{1,n} \cr
+ a_{2,1} & a_{2,2} & \cdots & a_{2,n} \cr
+ \vdots & \vdots & \ddots & \vdots \cr
+ a_{m,1} & a_{m,2} & \cdots & a_{m,n} },
+$$
+or:
+
+\begtt
+\def\qmatrix[#1]{\left[\matrix{#1}\right]}
+$$
+ \qmatrix[a&b&c\cr d&e&f\cr g&h&i] \cdot \qmatrix[x_1\cr x_2\cr x_3]
+ = \qmatrix[b_1\cr b_2\cr b_3].
+$$
+\endtt
+$$
+ \def\qmatrix[#1]{\left[\matrix{#1}\right]}
+ \qmatrix[a&b&c\cr d&e&f\cr g&h&i] \cdot \qmatrix[x_1\cr x_2\cr x_3]
+ = \qmatrix[b_1\cr b_2\cr b_3].
+$$
+
+If you need to align the columns by another way than to center, then you can
+use the phantom. Compare:
+
+\begtt
+$$
+ \pmatrix{ 1 & -1 & 0 \cr
+ 0 & 2 & 13 \cr
+ -3 & 0 & 5 } \quad \hbox{or} \quad
+ \def\0{\phantom0} \def\+{\phantom+}
+ \pmatrix { \+1 & -1 & \00 \cr
+ \+0 & \+2 & 13 \cr
+ -3 & \+0 & \05 }
+$$
+\endtt
+$$
+ \pmatrix{ 1 & -1 & 0 \cr
+ 0 & 2 & 13 \cr
+ -3 & 0 & 5 } \quad \hbox{or} \quad
+ \def\0{\phantom0} \def\+{\phantom+}
+ \pmatrix { \+1 & -1 & \00 \cr
+ \+0 & \+2 & 13 \cr
+ -3 & \+0 & \05 }
+$$
+
+\new\OpTeX/
+Another option to set the right aligned matrix is setting the \x`\lmfil`:
+Its value is used on the left side in each `\matrix` item. The right side is
+set directly to `\hfil`.
+\begtt
+$$
+ \lmfil={\hfill} % left matrix filler = \hfill
+ \pmatrix{ 1 & -1 & 0 \cr 0 & 2 & 13 \cr -3 & 0 & 5 }
+$$
+\endtt
+
+\new \OpTeX/
+If you want to draw a vertical line inside the matrix, you can use `\adef|`
+as in the following example:
+\begtt
+$$
+ \adef|{\kern-.2em&\strut\vrule&\kern-.2em}
+ \def\+{\phantom+}
+ \pmatrix{1 & 2 & 3 | 0 \cr 4 & 5 & 6 | 1 \cr 7 & 8 & 9 | 2 } \sim
+ \pmatrix{1 & \+2 & \+3 | 0 \cr 0 & -3 & -6 | 1 \cr 0 & -6 & -12 | 2} \sim
+ \pmatrix{1 & 2 & 3 | 0 \cr 0 & 1 & 2 | -1/3 \cr 0 & 0 & 0 | 1 }
+$$
+\endtt
+$$
+ \adef|{\kern-.2em&\strut\vrule&\kern-.2em}
+ \def\+{\phantom+}
+ \pmatrix{1 & 2 & 3 | 0 \cr 4 & 5 & 6 | 1 \cr 7 & 8 & 9 | 2 } \sim
+ \pmatrix{1 & \+2 & \+3 | 0 \cr 0 & -3 & -6 | 1 \cr 0 & -6 & -12 | 2} \sim
+ \pmatrix{1 & 2 & 3 | 0 \cr 0 & 1 & 2 | -1/3 \cr 0 & 0 & 0 | 1 }
+$$
+
+If you want to put something before the opening bracket in the matrix, you can
+use another `\matrix`. Example:
+
+\begtt
+$$
+ \adef|{\kern-.2em&\strut\vrule&\kern-.2em}
+ \def\+{\phantom+}
+ \def\r{{\bf r}}
+ \pmatrix{1 & 2 & 3 | 0 \cr 4 & 5 & 6 | 1 \cr 7 & 8 & 9 | 2 } \ \sim \
+ \matrix{\cr 2.\r - 4\cdot 1.\r: \cr 3.\r -7\cdot 1.\r: }
+ \pmatrix{1 & \+2 & \+3 | 0 \cr 0 & -3 & -6 | 1 \cr 0 & -6 & -12 | 2}
+$$
+\endtt
+$$
+ \adef|{\kern-.2em&\strut\vrule&\kern-.2em}
+ \def\+{\phantom+}
+ \def\r{{\bf r}}
+ \pmatrix{1 & 2 & 3 | 0 \cr 4 & 5 & 6 | 1 \cr 7 & 8 & 9 | 2 } \ \sim \
+ \matrix{\cr 2.\r - 4\cdot 1.\r: \cr 3.\r -7\cdot 1.\r: }
+ \pmatrix{1 & \+2 & \+3 | 0 \cr 0 & -3 & -6 | 1 \cr 0 & -6 & -12 | 2}
+$$
+
+Plain \TeX/ defines the \x`\bordermatrix` macro which allows you to create a
+top row above the brackets and left column before the opening bracket.
+\TeX/book shows the following example:
+
+\begtt
+$$
+ M = \bordermatrix{ &C&I&C'\cr
+ C&1&0&0 \cr I&1-b&0 \cr C'&0&a&1-a }
+$$
+\endtt
+$$
+ M = \bordermatrix{ &C&I&C'\cr
+ C&1&0&0 \cr I&1-b&0 \cr C'&0&a&1-a }
+$$
+
+The `\matrix` macro used in $D$ style creates all its items in $T$ style.
+\new \OpTeX/
+If you are using the `\matrix` macro in $T$ style or $S$ style (but not in
+fractions nor another matrices) then the resulting table is appropriately
+smaller and all its items are processed in $S$ style.
+If you are using `\matrix` in $SS$ style then the items are in $SS$ style too.
+The following example shows one-column matrix in script style:
+$$
+ `\sum``_{\matrix{i\in M\cr j\in N\cr k\in P}} x_{i,j,k}` \ \hbox{ creates: }
+ \sum_{\matrix{i\in M\cr j\in N\cr k\in P}} x_{i,j,k}
+$$
+
+
+\secc Cases
+
+The \x`\cases` macro can be used as in the following example:
+
+\begtt
+$$
+ f(x) = \cases{ 1 & for $x\le 0$, \cr
+ 100-x\over 100 & when $0 < x \le 100$, \cr
+ 0 & in other cases. }
+$$
+\endtt
+$$
+ f(x) = \cases{ 1 & for $x\le 0$, \cr
+ 100-x\over 100 & when $0\string< x\le 100$, \cr
+ 0 & in other cases. }
+$$
+
+The `\cases` macro behaves like a special `\matrix` with two left-aligned
+columns and with left vertically scaled brace $\{$.
+The first column is processed in math mode and $T$ style, the second column
+is processed in text mode. We have to use `$...$` in the second column if
+there is math material.
+
+
+\sec Lines in display mode
+
+\secc General principles
+
+The \ii display/math,display/math/mode
+`$$<formula>$$` finalizes previous paragraph, prints centered <formula> on
+single line with a vertical space above and below and opens next paragraph
+with no indentation.
+
+From \TeX/'s point of view, the text above `$$<formula>$$` plus text
+below is a single paragraph interrupted by display <formula>. If there is no
+text above (i.e. the opening `$$` are in vertical mode), then the internal
+`\noindent` is processed first and the empty line above <formula> is created.
+Thus, it is definitely a bad idea to open display mode in vertical mode: never
+put an empty line before `$$<formula>$$`. On the other hand, the empty line
+just after `$$<formula>$$` says that the paragraph is finalized by the
+<formula> and the next text (after the empty line) opens the next paragraph with
+indentation. Summary:
+
+\begitems
+* \"Text above `$$<formula>$$` text below": the <formula> interrupts the
+ paragraph with \"Text above" and \"text below". The \"text below" is
+ without indentation.
+* \"Text above `$$<formula>$$` empty line Text below": the \"Text below"
+ opens new paragraph with indentation.
+* \"Empty line `$$<formula>$$`" is bad practice.
+\enditems
+
+In contrast with the paragraph breaking, there is no built-in algorithm for
+breaking display <formula> to more lines. If the <formula> is too wide then
+overfull `\hbox` occurs and human must decide about splitting the <formula>
+to more lines. The macros `\displaylines` and `\eqalign` are intended to
+such task, see sections~\ref[dlines] and~\ref[elines].
+
+On the other hand, the in-line math <formula>, i.e.\ the `$<formula>$` in a
+paragraph, can be broken after a Bin atom (with penalty `\binoppenalty`) or
+after a Rel atom (with penalty `\relpenalty`). If you don't want to break
+such a formula at a specific place then use `\nobreak`, for example
+`$a+\nobreak b$`. If you want never to break such formulas then you can set
+`\binoppenalty=10000`, `\relpenalty=10000`. (Default values are 700 and 500.)
+
+
+\secc References to display lines
+
+The `$$<formula>`\x`\eqno``<mark>$$` \ii mark prints centered <formula> and the <mark> at
+right margin. The `$$<formula>`\x`\leqno``<mark>$$` prints centered <formula> and
+the <mark> at left margin. Examples:
+
+\begtt
+$$ a^2 + b^2 = c^2 \eqno (1) $$
+\endtt
+$$
+ a^2 + b^2 = c^2 \eqno (1)
+$$
+\begtt \aftergroup\nobreak
+$$ a^2 + b^2 = c^2 \leqno \rm Py $$
+\endtt
+$$\predisplaypenalty=10000
+ a^2 + b^2 = c^2 \leqno \rm Py
+$$
+The <mark> \ii mark is processed in math mode and $T$ style. If you want to put a
+text here then you have to use `\rm <text>` or `\hbox{<text>}`.
+
+\new \OpTeX/
+The auto-generated <mark> can be created by `$$<formula>`\x`\eqmark``$$`. The
+<label> can be used previously declared by \x`\label``[<label>]`.
+Then it is associated to such
+auto-generated <mark> and you can write a reference in the form
+`see equation~\ref[<label>]`. You can write `\eqmark[<label>]` instead of
+`\label[<label>]...\eqmark`. See the \OpTeX/ manual, section~1.4.3.
+
+\secc[dlines] The `\displaylines` macro
+
+This macro creates more horizontally centered formulas in one display mode.
+The syntax is
+\begtt \catcode`\<=13
+$$ \displaylines{
+ <formula> \cr
+ <formula> \cr
+ ... \cr
+ <formula>
+} $$
+\endtt
+Usage of \x`\displaylines` is more
+prefered than doing more display modes just behind each other.
+Example:
+\begtt
+$$ \displaylines{
+ (3x^3 + 4x^2 + 5x + 6) \cdot (x^6 + x^2 + 5) = \cr
+ = 3x^9 + 4x^8 + 5x^7 + 6x^6 + 3x^5 + 4x^4 + 20x^3 + 26x^2 + 25x + 30.
+} $$
+\endtt
+$$ \displaylines{
+ (3x^3 + 4x^2 + 5x + 6) \cdot (x^6 + x^2 + 5) = \cr
+ = 3x^9 + 4x^8 + 5x^7 + 6x^6 + 3x^5 + 4x^4 + 20x^3 + 26x^2 + 25x + 30.
+} $$
+
+The output from `\displaylines` macro is breakable to more pages because
+the lines are not encapsulated in one box. The macro uses a special feature of
+`\halign` primitive. The disadvantage is that you cannot use `\eqno` nor
+`\leqno` nor `\eqmark`.
+
+\new \OpTeX/
+\OpTeX/ provides alternative form: `\displaylines to<dimen> {<data>}`.
+Then the centered formulas are encapsulated in a `\vcenter` box of width <dimen> and
+usage of `\eqno` or `\leqno` or `\eqmark` is allowed.
+The individual lines have the form `\hbox to<dimen>` and formulas are centered using
+`\hss` from both sides. This means that you can set arbitrary <dimen>
+without visual change of the formulas. Use smaller <dimen> value than
+`\hsize` (or `\displaywidth`\fnote
+{`\displaywith` = `\hsize` in most cases but it is
+real display width when `\parshape` or `\hangindent` is used.})
+if you want to center formulas with `\eqno` appended.
+The internal \TeX/ rule says: the formula with `\eqno` (or `\leqno`) is centered
+if its width is less or equal `\displaywidth-4<mark-width>`. So
+\begtt \catcode`\<=13
+$$ \displaylines to\hsize-8em {<formula>\cr<formula>}\eqno (1) $$
+\endtt
+does global centering, because size of `(1)` is less than 2\,em.
+You can do more experiments with this example,
+for example `\diplaylines to\hsize{...}`
+puts the `\eqno` mark to the next line in the display environment. Read
+\TeX/book or \TeX/book naruby, where the precise explanation about such
+positioning is.
+
+\new \OpTeX/
+You can use \"`\displaylines to<dimen>`" for more applications. For example,
+you can put more
+\"`\displaylines to<dimen>`" in single display mode, one next to second in order
+to creating more centered columns with formulas; the width of such columns
+are controlled by the <dimen> parameter.
+Or, you can give an exception for several lines:
+
+\begtt \catcode`\<=13
+$$\displaylines to 10cm {<formula>\hfill \cr
+ <formula>\cr
+ <formula>\cr
+ \hfill <formula>}
+$$
+\endtt
+The example above gives similar result as the \LaTeX/ `multline` environment:
+$$\displaylines to \hsize {
+ (3x^3 + 4x^2 + 5x + 6) \cdot (x^6 + x^2 + 5) = \hfill \cr
+ = 3x^9 + 3x^5 + 15x^3 + 4x^8 + 4x^4 + 20x^2 + 5x^7 + 5x^3 + 25x
+ + 6x^6 + 6x^2 + 30 = \cr
+ \hfill = 3x^9 + 4x^8 + 5x^7 + 6x^6 + 3x^5 + 4x^4 + 20x^3 + 26x^2 + 25x + 30. }
+$$
+
+
+\secc[elines] The `\eqalign` macro
+
+The \ii .eqalign \iis .eqalign {{\code{\\eqalign}}} usage is:
+\begtt \catcode`\<=13
+$$ \eqalign { <left-side> & <right-side> \cr
+ <left-side> & <right-side> \cr
+ ... \cr
+ <left-side> & <right-side>
+} $$
+\endtt
+The `\vcenter` box is created with two columns, left column is right aligned
+and right column is left aligned. Example:
+
+\vbox{
+\begtt
+$$ \eqalign{
+ x + 2y + 3z &= 600 \cr
+ 12x + y - 3z &= 7 \cr
+ 4x - y + 5z &= -5 \cr
+} $$
+\endtt
+\vskip-5.7\baselineskip
+{\hsize=1.35\hsize
+$$ \eqalign{
+ x + 2y + 3z &= 600 \cr
+ 12x + y - 3z &= 7 \cr
+ 4x - y + 5z &= -5 \cr
+} $$}}
+
+\bigskip\noindent
+The tab \"`&`" should be used just before a relation, i.e. `&=` is right, `=&`
+is wrong. All lines are aligned to the used tab.
+
+Maybe you want more precise alignment in the example above. You can use
+`\phantom`:
+
+\par\nobreak\vbox{
+\begtt
+$$ \def\1{\phantom1} \def\+{\phantom+}
+ \eqalign{
+ x + 2y + 3z &= 600 \cr
+ 12x + \1y - 3z &= \+7 \cr
+ 4x - \1y + 5z &= -5 \cr
+} $$
+\endtt
+\vskip-5.7\baselineskip
+{\hsize=1.35\hsize
+$$ \def\1{\phantom1} \def\+{\phantom+}
+ \eqalign{
+ x + 2y + 3z &= 600 \cr
+ 12x + \1y - 3z &= \+7 \cr
+ 4x - \1y + 5z &= -5
+} $$}}
+\bigskip
+Another typical usage of the `\eqalign` macro:
+
+\begtt \typosize[10/12]
+$$ \eqalign{
+ p(x)\,q(x) &= (3x^3 + 4x^2 + 5x + 6) \cdot (x^6 + x^2 + 5) = \cr
+ &= 3x^9 + 4x^8 + 5x^7 + 6x^6 + 3x^5 + 4x^4 + 20x^3 + 26x^2 + 25x + 30.
+} $$
+\endtt
+$$ \eqalign{
+ p(x)\,q(x) &= (3x^3 + 4x^2 + 5x + 6) \cdot (x^6 + x^2 + 5) = \cr
+ &= 3x^9 + 4x^8 + 5x^7 + 6x^6 + 3x^5 + 4x^4 + 20x^3 + 26x^2 + 25x + 30.
+} $$
+
+In \OpTeX/, \new\OpTeX/
+the `\eqalign` macro is more flexible. You can set the
+`\baselineskip` value by the \x`\eqlines` parameter and math style
+by the \x`\eqstyle` parameter.
+For example, you need to put the system of \"equations" as a subscript of a sum
+operator:
+\begtt \typosize[10/12] \adef/{}
+$$
+ \sum/_{\eqlines{\baselineskip=.7\baselineskip}\eqstyle{\scriptstyle}\eqalign{
+ i &\in A \cr
+ \quad j &\in B\cup C \cr
+ m &\in C }}
+ i + j + m
+$$
+\endtt
+$$
+ \sum_{\eqlines{\baselineskip=.7\baselineskip}\eqstyle{\scriptstyle}\eqalign{
+ i &\in A \cr
+ \quad j &\in B\cup C \cr
+ m &\in C }}
+ i + j + m
+$$
+
+\indent \new\OpTeX/
+You can write more equation systems one next second:
+\begtt
+$$\eqalign{
+ x + y + z &= 1 && a + b + c &= -1 \cr
+ u + v &= 20 && f + g &= -20 \cr
+ i &= j
+}$$
+\endtt
+You can use the third column for centered equations
+without aligning point. For example:
+\begtt \catcode`\<=13
+$$ \left( \eqspace=0pt \eqalign{&& <first equation>\cr
+ && <second equation>} \right) $$
+\endtt
+The \x`\eqspace` is additional space used in the third column to separate
+equation systems one next second.
+
+\new \OpTeX/
+\OpTeX/ extensions summary:
+\begitems
+* `\eqlines` and `\eqstyle` set baselineskip and math style of the formulas.
+* `\eqalign` allows more than two columns:
+ The first column is right-aligned (no space). The second is left-aligned (no space).
+ The third column (if used)
+ is centered with `\eqspace/2` at the left and right boundary of the column.
+ The fourth is the same as the first. The fifth is the same as second etc. The number
+ of columns that can be used in `\eqalign` is unlimited.
+\enditems
+
+\secc The `\eqalign` macro with references
+
+You can give common mark to whole equation system by
+`$$\eqalign{<data>}\eqno <mark>$$`. If you want to give marks to individual
+lines of the equation system, then you can use another macro:
+\x`\eqalignno`. The usage is similar as `\eqalign` but the
+third column (if used) is intended to the equation mark. Example:
+
+\begtt
+$$ \eqalignno{
+ x + 2y + 3z &= 600 & \rm(A) \cr
+ 12x + y - 3z &= 7 & \rm(B) \cr
+ 4x - y + 5z &= -5 & \rm(C) \cr
+} $$
+\endtt
+\vskip-\baselineskip
+$$ \eqalignno{
+ x + 2y + 3z &= 600 & \rm(A) \cr
+ 12x + y - 3z &= 7 & \rm(B) \cr
+ 4x - y + 5z &= -5 & \rm(C) \cr
+} $$
+
+The `\leqalignno` macro is similar to `\eqalignno` but the marks are at the left
+margin. The \OpTeX/ extensions of `\eqalign` are not available in
+`\eqalignno` nor `\leqalignno` macros.
+
+\new \OpTeX/
+You can use auto-generated marks by `\eqmark` macro:
+`$$\eqalign{<data>}\eqmark$$` or:
+\begtt
+$$ \eqalignno{
+ x + 2y + 3z &= 600 & \eqmark[A] \cr
+ 12x + y - 3z &= 7 & \eqmark[B] \cr
+ 4x - y + 5z &= -5 & \eqmark[C] \cr
+} $$
+\endtt
+
+
+\sec Concept of loading math fonts
+
+\secc[fams] Math families
+
+\TeX/ can use more than one math font in math mode. This was a
+necessity in the old days when only 128-characters fonts existed.
+Each math font used in math mode has its \ii math/family {\em math family} represented by a number.
+Math family is a collection of three (almost) equal fonts in three sizes:
+first for `\textstyle` and `\displaystyle`, second for `\scriptstyle`
+and third for `\scriptscriptstyle`.
+
+\new Unicode
+When Unicode math font is loaded then it includes all three optical sizes and
+all characters needed for typesetting math formula.
+Theoretically, we can use only one math family with this single font. But more math families
+(i.e.\ more fonts in math mode) is still possible. You can combine
+characters from more fonts (Unicode fonts and old TFM fonts together) in one math formula.
+
+\new \OpTeX/
+\OpTeX/ loads the main Unicode math font into math family 1. The math
+families 2 and 3 are reserved for specific \TeX/nical reasons, family 4 is
+used for `\script` font and families 5, 6, 7, etc. can be used by user for
+loading more fonts. The default macro for loading math fonts looks like:
+
+\begtt \typosize[10/12]
+\_def\_normalunimath{%
+ \_loadumathfamily 1 {\_unimathfont}{} % Main Unicode math font
+ \_loadmathfamily 4 rsfs % \script (old TFM font)
+ \_setunimathdimens % set dimen parameters used in math formulas internally
+}%
+\let\_normalmath=\_normalunimath % this is done when Unicode math is initialized
+\endtt
+%
+Whenever \OpTeX/ needs to resize math fonts (for example in footnotes or
+titles), it calls the `\_normalmath` macro to reload all math
+families to the desired size. If you want to add the next font, you can add
+`\_loadunimathfamily <family> {<Unicode-font>}{<features>}` or
+`\_loadmathfamily <family> <TFM-font>` into the `\_normalmath` macro.
+The example in section~\ref[newfam] shows how to do it.
+
+\secc Two variants of math fonts: normal and bold
+
+All math formulas in the whole document need only one Unicode font (or only one collection
+of math fonts as mentioned in section~\ref[fams]). But this
+is not really true if titles are in bold font. If a math formula is present in
+such a bold title then all characters of this formula must be bolder.
+For example \"normal" variables must be in bold italic in titles, symbols
+like `+` `=` must be bold and \"normal bold" letters
+(e.g., indicating vectors in math formula) must be extra bold in titles.
+It means that all fonts from the collection of math fonts must be bolder.
+We need a second collection of math fonts with bolder
+shape. Unfortunately, it is not always available.
+
+\new \OpTeX/
+If you have bold variant of used Unicode math font, then you can use
+\ii .loadmath \iis .loadmath {{\code{\\loadmath}}} \x`\loadboldmath` command, for example:
+\begtt
+\loadmath {[xitsmath-regular]}
+\loadboldmath {[xitsmath-bold]} \to {[xitsmath-regular]}
+\endtt
+%
+\OpTeX/ uses `\_normalmath` macro for loading collection of math fonts in
+\"normal" cases and `\_boldmath` macro for bold titles. The font declared by
+\x`\loadboldmath` is used in second case. But if the bold variant of the font is
+not available (this is unfortunately more typical), then \OpTeX/ uses faked
+bold of main Unicode math font (the `embolden` internal font feature is
+used). So, the default `\_boldmath` macro defined by \OpTeX/ looks like:
+
+\begtt \typosize[10/12]
+\_def\_boldunimath{%
+ \_ifx\_unimathboldfont \_undefined
+ \_loadumathfamily 1 {\_unimathfont}{embolden=1.7;} % Main faked bold
+ \_else
+ \_loadumathfamily 1 {\_unimathboldfont}{} % Main real bold font
+ \_fi
+ \_loadmathfamily 4 rsfs % rsfs in not in bold, unfortunately
+ \_setunimathdimens
+}%
+\let\_boldmath=\_boldunimath % this is done when Unicode math is initialized
+\endtt
+
+\secc[newfam] Example of using additional math font
+
+The font `bbold10.tfm` includes double stroked characters, for example,
+double stroked plus, double stroked Greek letters and digits.
+Try to run `pdftex testfont`, then answer the question about the name of the
+font: `bbold10` and then type command `\table\end`. The `testfont.pdf` is
+printed with the table of characters of this font.
+Most of these characters cannot be found in Unicode math fonts.
+
+\new \OpTeX/
+We show an example of how to add this font to the collection of used math fonts.
+We can re-define the `\_normalmath` macro by:
+\begtt \typosize[10/12]
+\addto\_normalmath {\_loadmathfamily 5 bbold }
+
+\_regtfm bbold 0 bbold5 5.5 bbold6 6.5 bbold7 7.5 bbold8 8.5 bbold9
+ 9.5 bbold10 11.1 bbold12 15 bbold17 * % using all bbold*.tfm
+\_normalmath % reload the math fonts collection
+\endtt
+%
+The string \"`bbold`" is declared by `\_regtfm` as a collection of all
+`bbold*.tfm` fonts, the optical sizes are supported.
+
+\addto\_normalmath {\_loadmathfamily 5 bbold }
+\_regtfm bbold 0 bbold5 5.5 bbold6 6.5 bbold7 7.5 bbold8 8.5 bbold9
+ 9.5 bbold10 11.1 bbold12 15 bbold17 * % using all bbold*.tfm
+\_normalmath
+\Umathchardef \bbplus 2 5 "2B
+\Umathchardef \bble 3 5 "3C
+\Umathchardef \bbge 3 5 "3E
+
+Finally, we must to declare new \TeX/ sequences for accessing the characters
+from the new font, for example:
+
+\begtt \typosize[10/12]
+\Umathchardef \bbplus 2 5 "2B
+\Umathchardef \bble 3 5 "3C
+\Umathchardef \bbge 3 5 "3E
+\endtt
+%
+The \x`\Umathchardef` \TeX/ primitive declares new \TeX/ sequence used in math
+typesetting. The first parameter is a class number (2 means Bin, 3 means Rel,
+see the table in the section~\ref[class]). The second parameter is a math
+family number. It is 5, see the redefinition of the `\_normalmath` macro above.
+The third parameter is a slot in the font. Now you can try to use these characters:
+$$
+ `a \bbplus b \bbge c` \quad \hbox{gives} \quad a \bbplus b \bbge c.
+$$
+%
+Maybe, you want to declare a special math selector which can be used as
+`$\bball a+b>c$` in order to get $\bbchar a \bbplus b \bbge c$.
+Then you can define:
+
+\begtt \typosize[10/12]
+\def\bball {\bbchar \Umathcodenum `+ \bbplus
+ \Umathcodenum `> \bbge
+ \Umathcodenum `< \bble }
+\endtt
+
+If you want to add all double stroked Greek letters into `\bball`
+selector, then you can do something like this:
+
+\begtt \typosize[10/12]
+\def\setbbgreek #1 {\tmpnum=#1\xargs\setbbgreekA}
+\def\setbbgreekA #1{\Umathcode \_ea`#1 0 5 \tmpnum \advance\tmpnum by1 }
+
+\addto\bball {%
+ \setbbgreek "0B \alpha \beta \gamma \delta \epsilon \zeta \eta
+ \theta \iota \kappa \lambda \mu \nu \xi \pi \rho \sigma
+ \tau \upsilon \phi \chi ;
+ \setbbgreek "7F \omega ;
+ \setbbgreek "00 \Gamma \Delta \Theta \Lambda \Xi \Pi \Sigma \Upsilon
+ \Phi \Psi \Omega ; }
+\endtt
+%
+The \x`\Umathcode`` <input-code> <class> <family> <font-slot>` primitive is used
+here. The control sequences `\alpha`, `\beta`, `\gamma` etc. are macros which expand
+to the Unicode character of appropriate (non-math) Greek letter. We set the
+Umathcode to such character, for example
+\code{\\Umathcode `}$\_rmgreek\alpha$` 0 5 "0B` is processed.
+
+The bold variant of the font `bbold*.tfm` is not available, unfortunately.
+We have to settle for normal version of the font in the `\_boldmath` macro:
+
+\begtt \typosize[10/12]
+ \addto \_boldmath {\_loadmathfamily 5 bbold }
+\endtt
+
+Another approach of using more Unicode math fonts in a single formula is
+shown in
+\ulink[http://petr.olsak.net/optex/optex-tricks.html\#addumathfont]{OpTeX trick 0030}.
+
+\sec Index
+
+Control sequences listed at pages \pgref[listA]--\pgref[listB] are not
+mentioned here again.
+
+\begmulti 3
+\makeindex
+\endmulti
+
+\bye