eqn - format equations for troff or MathML . .


.SY eqn .OP -rvCNR .OP -d xy .OP -T name .OP -M dir .OP -f F .OP -s n .OP -p n .OP -m n [ files . . . ] .YS .

It is possible to have whitespace between a command line option and its parameter. . .


This manual page describes the GNU version of

R eqn , which is part of the groff document formatting system.

eqn compiles descriptions of equations embedded within

troff input files into commands that are understood by

R troff . Normally, it should be invoked using the

-e option of

R groff . The syntax is quite compatible with Unix eqn. The output of GNU

eqn cannot be processed with Unix troff; it must be processed with GNU troff. If no files are given on the command line, the standard input is read. A filename of

- causes the standard input to be read. .

eqn searches for the file

eqnrc in the directories given with the

-M option first, then in

R /usr/lib/groff/site-tmac ,

R /usr/share/groff/site-tmac , and finally in the standard macro directory

R /usr/share/groff/1.20.1/tmac . If it exists,

eqn processes it before the other input files. The

-R option prevents this. .


eqn does not provide the functionality of neqn: it does not support low-resolution, typewriter-like devices (although it may work adequately for very simple input). . .



  • -d xy
        Specify delimiters xand\~\c yfor the left and right end, respectively, of in-line equations. Any

delim statements in the source file overrides this. .



.EQ and

.EN even when followed by a character other than space or newline.


    Don't allow newlines within delimiters. This option allows

eqn to recover better from missing closing delimiters. .


    Print the version number. .


    Only one size reduction. .

  • -m n
        The minimum point-size is\~\c n .

eqn does not reduce the size of subscripts or superscripts to a smaller size than\~\c n . .

  • -T name
        The output is for device name . Normally, the only effect of this is to define a macro namewith a value of\~\c

R 1 ;

eqnrc uses this to provide definitions appropriate for the output device. However, if the specified device is \[lq]MathML\[rq], the output is MathML markup rather than troff commands, and

eqnrc is not loaded at all. The default output device is

R ps . .

  • -M dir
        Search dirfor

eqnrc before the default directories. .


    Don't load

R eqnrc . .

  • -f F
        This is equivalent to a gfont  F command. .
  • -s n
        This is equivalent to a gsize  n command. This option is deprecated.

eqn normally sets equations at whatever the current point size is when the equation is encountered. .

  • -p n
        This says that subscripts and superscripts should be n\~\cpoints smaller than the surrounding text. This option is deprecated. Normally

eqn sets subscripts and superscripts at 70% of the size of the surrounding text. . .


Only the differences between GNU

eqn and Unix eqn are described here. .


eqn emits Presentation MathML output when invoked with the

-T\~MathML option. .

GNU eqn sets the input token

... as three periods or low dots, rather than the three centered dots of classic eqn. To get three centered dots, write

cdots or

R .

Most of the new features of the GNU

eqn input language are based on \*(tx. There are some references to the differences between \*(tx and GNU

eqn below; these may safely be ignored if you do not know \*(tx. .

5.1. Automatic spacing

eqn gives each component of an equation a type, and adjusts the spacing between components using that type. Possible types are: .

  • ordinary
        an ordinary character such as `1' or `\c x '; .
  • operator
        a large operator such as \*(Su; .
  • binary
        a binary operator such as `(pl'; .
  • relation
        a relation such as `='; .
  • opening
        a opening bracket such as `('; .
  • closing
        a closing bracket such as `)'; .
  • punctuation
        a punctuation character such as `,'; .
  • inner
        a subformula contained within brackets;
  • suppress
        spacing that suppresses automatic spacing adjustment. .

Components of an equation get a type in one of two ways. .

  • type  t e
        This yields an equation component that contains\~\c ebut that has type\~\c t , where tis one of the types mentioned above. For example,

times is defined as .

type "binary" \(mu .

The name of the type doesn't have to be quoted, but quoting protects from macro expansion. .

  • chartype  t text
        Unquoted groups of characters are split up into individual characters, and the type of each character is looked up; this changes the type that is stored for each character; it says that the characters in textfrom now on have type\~\c t . For example, .

chartype "punctuation" .,;: .

would make the characters `.,;:' have type punctuation whenever they subsequently appeared in an equation. The type\~\c tcan also be

letter or

R digit ; in these cases

chartype changes the font type of the characters. See the

Fonts subsection. .

5.2. New primitives

  • big  e
        Enlarges the expression it modifies; intended to have semantics like CSS `large'. In troff output, the point size is increased by\~5; in MathML output, the expression uses .
<mstyle mathsize='big'>


  • .IB e1  smallover  e2
        This is similar to

R over ;

smallover reduces the size of e1and e2 ; it also puts less vertical space between e1or e2and the fraction bar. The

over primitive corresponds to the \*(tx

\over primitive in display styles;

smallover corresponds to

\over in non-display styles. .

  • vcenter  e
        This vertically centers eabout the math axis. The math axis is the vertical position about which characters such as `(pl' and `(mi' are centered; also it is the vertical position used for the bar of fractions. For example,

sum is defined as .

{ type "operator" vcenter size +5 \(*S } .

(Note that vcenter is silently ignored when generating MathML.) .

  • .IB e1  accent  e2
        This sets e2as an accent over e1 . e2is assumed to be at the correct height for a lowercase letter; e2is moved down according to whether e1is taller or shorter than a lowercase letter. For example,

hat is defined as .

accent { "^" } .

R dotdot ,

R dot ,

R tilde ,

R vec , and

dyad are also defined using the

accent primitive. .

  • .IB e1  uaccent  e2
        This sets e2as an accent under e1 . e2is assumed to be at the correct height for a character without a descender; e2is moved down if e1has a descender.

utilde is pre-defined using

uaccent as a tilde accent below the baseline. .

  • split (ts text (ts
        This has the same effect as simply .


but textis not subject to macro expansion because it is quoted; textis split up and the spacing between individual characters is adjusted. .

  • nosplit  text
        This has the same effect as .

(ts text (ts .

but because textis not quoted it is subject to macro expansion; textis not split up and the spacing between individual characters is not adjusted. .

  • .IB e  opprime
        This is a variant of

prime that acts as an operator on\~\c e . It produces a different result from

prime in a case such as

R A opprime sub 1 : with

opprime the\~\c

1 is tucked under the prime as a subscript to the\~\c

A (as is conventional in mathematical typesetting), whereas with

prime the\~\c

1 is a subscript to the prime character. The precedence of

opprime is the same as that of

bar and

R under , which is higher than that of everything except

accent and

R uaccent . In unquoted text a\~\c

' that is not the first character is treated like

R opprime . .

  • special  text e
        This constructs a new object from\~\c eusing a

R troff (1) macro named text . When the macro is called, the string

0s contains the output for\~\c e , and the number registers

R 0w ,

R 0h ,

R 0d ,

R 0skern , and

R 0skew contain the width, height, depth, subscript kern, and skew of\~\c e . (The "subscript kern"of an object says how much a subscript on that object should be tucked in; the skewof an object says how far to the right of the center of the object an accent over the object should be placed.) The macro must modify

0s so that it outputs the desired result with its origin at the current point, and increase the current horizontal position by the width of the object. The number registers must also be modified so that they correspond to the result. .

For example, suppose you wanted a construct that `cancels' an expression by drawing a diagonal line through it. .

Ne pas lire - Code non interprété
define cancel 'special Ca'

.de Ca
.  ds 0s \
\D'l \\n(0wu -\\n(0hu-\\n(0du'\
.. .

Then you could cancel an expression\~\c ewith cancel {  e  } .

Here's a more complicated construct that draws a box round an expression: .

Ne pas lire - Code non interprété
define box 'special Bx'

.de Bx
.  ds 0s \
\D'l \\n(0wu+2n 0'\
\D'l 0 -\\n(0hu-\\n(0du-2n'\
\D'l -\\n(0wu-2n 0'\
\D'l 0 \\n(0hu+\\n(0du+2n'\
.  nr 0w +2n
.  nr 0d +1n
.  nr 0h +1n
.. .

  • space  n
        A positive value of the integer\~\c n(in hundredths of an em) sets the vertical spacing before the equation, a negative value sets the spacing after the equation, replacing the default values. This primitive provides an interface to

R groff 's

\x escape (but with opposite sign). .

This keyword has no effect if the equation is part of a

pic picture. .

5.3. Extended primitives

  • col  n  {  . . .  }
    ccol  n  {  . . .  }
    lcol  n  {  . . .  }
    rcol  n  {  . . .  }
    pile  n  {  . . .  }
    cpile  n  {  . . .  }
    lpile  n  {  . . .  }
    rpile  n  {  . . .  } The integer value\~\c n(in hundredths of an em) increases the vertical spacing between rows, using

R groff 's

\x escape (the value has no effect in MathML mode). Negative values are possible but have no effect. If there is more than a single value given in a matrix, the biggest one is used. .

5.4. Customization


eqn is generating troff markup, the appearance of equations is controlled by a large number of parameters. They have no effect when generating MathML mode, which pushes typesetting and fine motions downstream to a MathML rendering engine. These parameters can be set using the

set command. .

  • set  p n
        This sets parameter\~\c pto value\~\c n ; n\~\cis an integer. For example, .

set x_height 45 .

says that

eqn should assume an x\~height of 0.45\~ems. .

Possible parameters are as follows. Values are in units of hundredths of an em unless otherwise stated. These descriptions are intended to be expository rather than definitive. . .ie t \ . TP +2n . TP


eqn doesn't set anything at a smaller point-size than this. The value is in points. .



fat primitive emboldens an equation by overprinting two copies of the equation horizontally offset by this amount. This parameter is not used in MathML mode; instead, fat text uses .

<mstyle mathvariant='double-struck'>



    A fraction bar is longer by twice this amount than the maximum of the widths of the numerator and denominator; in other words, it overhangs the numerator and denominator by at least this amount. .



bar or

under is applied to a single character, the line is this long. Normally,

bar or

under produces a line whose length is the width of the object to which it applies; in the case of a single character, this tends to produce a line that looks too long. .


    Extensible delimiters produced with the

left and

right primitives have a combined height and depth of at least this many thousandths of twice the maximum amount by which the sub-equation that the delimiters enclose extends away from the axis. .


    Extensible delimiters produced with the

left and

right primitives have a combined height and depth not less than the difference of twice the maximum amount by which the sub-equation that the delimiters enclose extends away from the axis and this amount. .


    This much horizontal space is inserted on each side of a fraction. .


    The width of subscripts and superscripts is increased by this amount. .


    This amount of space is automatically inserted after punctuation characters. .


    This amount of space is automatically inserted on either side of binary operators. .


    This amount of space is automatically inserted on either side of relations. .


    The height of lowercase letters without ascenders such as `x'. .


    The height above the baseline of the center of characters such as `(pl' and `(mi'. It is important that this value is correct for the font you are using. .


    This should set to the thickness of the

\(ru character, or the thickness of horizontal lines produced with the

\D escape sequence. .



over command shifts up the numerator by at least this amount. .



smallover command shifts up the numerator by at least this amount. .



over command shifts down the denominator by at least this amount. .



smallover command shifts down the denominator by at least this amount. .


    Normally superscripts are shifted up by at least this amount. .


    Superscripts within superscripts or upper limits or numerators of

smallover fractions are shifted up by at least this amount. This is usually less than sup1. .


    Superscripts within denominators or square roots or subscripts or lower limits are shifted up by at least this amount. This is usually less than sup2. .


    Subscripts are normally shifted down by at least this amount. .


    When there is both a subscript and a superscript, the subscript is shifted down by at least this amount. .


    The baseline of a superscript is no more than this much amount below the top of the object on which the superscript is set. .


    The baseline of a subscript is at least this much below the bottom of the object on which the subscript is set. .


    The baseline of an upper limit is at least this much above the top of the object on which the limit is set. .


    The baseline of a lower limit is at least this much below the bottom of the object on which the limit is set. .


    The bottom of an upper limit is at least this much above the top of the object on which the limit is set. .


    The top of a lower limit is at least this much below the bottom of the object on which the limit is set. .


    This much vertical space is added above and below limits. .


    The baselines of the rows in a pile or matrix are normally this far apart. In most cases this should be equal to the sum of

num1 and

R denom1 . .


    The midpoint between the top baseline and the bottom baseline in a matrix or pile is shifted down by this much from the axis. In most cases this should be equal to

R axis_height . .


    This much space is added between columns in a matrix. .


    This much space is added at each side of a matrix. .


    If this is non-zero, lines are drawn using the

\D escape sequence, rather than with the

\l escape sequence and the

\(ru character. .


    The amount by which the height of the equation exceeds this is added as extra space before the line containing the equation (using

R \x ). The default value is 85. .


    The amount by which the depth of the equation exceeds this is added as extra space after the line containing the equation (using

R \x ). The default value is 35. .


    If this is non-zero, then

ndefine behaves like

define and

tdefine is ignored, otherwise

tdefine behaves like

define and

ndefine is ignored. The default value is\~0 (This is typically changed to\~1 by the

eqnrc file for the

R ascii ,

R latin1 ,

R utf8 , and

cp1047 devices.) .

A more precise description of the role of many of these parameters can be found in Appendix\~H of "The \*(txbook" . .

5.5. Macros

Macros can take arguments. In a macro body, $ n where nis between 1 and\~9, is replaced by the n-th argument if the macro is called with arguments; if there are fewer than n\~\carguments, it is replaced by nothing. A word containing a left parenthesis where the part of the word before the left parenthesis has been defined using the

define command is recognized as a macro call with arguments; characters following the left parenthesis up to a matching right parenthesis are treated as comma-separated arguments; commas inside nested parentheses do not terminate an argument. .

  • sdefine  name X anything X
        This is like the

define command, but nameis not recognized if called with arguments. .

  • include (ts file (ts
    copy (ts file (ts Include the contents of file ( include and

copy are synonyms). Lines of filebeginning with

.EQ or

.EN are ignored. .

  • ifdef  name X anything X
        If namehas been defined by

define (or has been automatically defined because nameis the output device) process anything ; otherwise ignore anything . Xcan be any character not appearing in anything . .

  • undef  name
        Remove definition of name , making it undefined. .

Besides the macros mentioned above, the following definitions are available:

R Alpha ,

R Beta , . . .,

Omega (this is the same as


R BETA , . . .,


ldots (three dots on the base line), and

R dollar . .

5.6. Fonts

eqn normally uses at least two fonts to set an equation: an italic font for letters, and a roman font for everything else. The existing

gfont command changes the font that is used as the italic font. By default this is\~\c

R I . The font that is used as the roman font can be changed using the new

grfont command. .

  • grfont  f
        Set the roman font to\~\c f . .


italic primitive uses the current italic font set by

R gfont ; the

roman primitive uses the current roman font set by

R grfont . There is also a new

gbfont command, which changes the font used by the

bold primitive. If you only use the

R roman ,

italic and

bold primitives to changes fonts within an equation, you can change all the fonts used by your equations just by using

R gfont ,

grfont and

gbfont commands. .

You can control which characters are treated as letters (and therefore set in italics) by using the

chartype command described above. A type of

letter causes a character to be set in italic type. A type of

digit causes a character to be set in roman type. . .


.Tp +2n

/usr/share/groff/1.20.1/tmac/eqnrc Initialization file. .


MathML is designed on the assumption that it cannot know the exact physical characteristics of the media and devices on which it will be rendered. It does not support fine control of motions and sizes to the same degree troff does. Thus: .


eqn parameters have no effect on the generated MathML. .

* The

R special,

R up ,

R down ,

R fwd , and

back operations cannot be implemented, and yield a MathML `<merror>' message instead. .

* The

vcenter keyword is silently ignored, as centering on the math axis is the MathML default. .

* Characters that

eqn over troff sets extra large (en notably the integral sign (en may appear too small and need to have their `<mstyle>' wrappers adjusted by hand. .

As in its troff mode,

eqn in MathML mode leaves the

.EQ and

.EN delimiters in place for displayed equations, but emits no explicit delimiters around inline equations. They can, however, be recognized as strings that begin with `<math>' and end with `</math>' and do not cross line boundaries. .

See the

BUGS section for translation limits specific to

R eqn . . .


Inline equations are set at the point size that is current at the beginning of the input line.

In MathML mode, the

mark and

lineup features don't work. These could, in theory, be implemented with `<maligngroup>' elements.

In MathML mode, each digit of a numeric literal gets a separate `<mn></mn>' pair, and decimal points are tagged with `<mo></mo>'. This is allowed by the specification, but inefficient. . .


R groff (1),

R troff (1),

R pic (1),

R groff_font (5), The \*(txbook.