Background

In actuarial notation, one often writes something like

A ___   __

  x:y : n|

Luckily, the actuarialangle package provides the command angl{<duration>} to typeset the “angle” notation. This command is designed in such a way that the vertical space between the angle roof and the duration matches that of overline when used in first-level subscript.

% !TeX program = XeLaTeX

documentclass{article}

%usepackage{newpxtext}

%usepackage{newpxmath}

%usepackage{newtxtext}

%usepackage{newtxmath}

usepackage{actuarialangle}

newcommand*test{%

  & $displaystyle     overline{30},angl{30}$

  & $textstyle        overline{30},angl{30}$

  & $scriptstyle      overline{30},angl{30}$

  & $scriptscriptstyleoverline{30},angl{30}$%

}

pagestyle{empty}

begin{document}

begin{tabular}{l l l l l}

$mathrm{Computer Modern}$ test \

$mathrm{newpxmath}$ test \

$mathrm{newtxmath}$ test \

end{tabular}

end{document}

Problem with unicode-math

With OpenType math fonts via unicode-math, I was aware of the fact that some legacy fontdimen’s, e.g., fontdimen8scriptfont3 (used by angl), were inappropriate. See Problem with substack and substack doesn’t work properly.

My attempt of re-implementing angl is thus to replace 3fontdimen8scriptfont3 with fontdimen53 (overbar_vgap) and to replace the fixed rule thickness of 0.4pt with fontdimen54 (overbar_rule). I was quite happy that my guess work turned out well for Latin Modern Math. But then I tried other fonts and my guess failed miserably:

% !TeX program = XeLaTeX

documentclass{article}

usepackage{unicode-math}

setmathfont{Latin Modern Math}[version=LM]

setmathfont{TeX Gyre Bonum Math}[version=Bonum]

setmathfont{TeX Gyre Pagella Math}[version=Pagella]

setmathfont{TeX Gyre Schola Math}[version=Schola]

setmathfont{TeX Gyre Termes Math}[version=Termes]

setmathfont{Fira Math}[version=Fira]

usepackage{actuarialangle}

makeatletter

ExplSyntaxOn

% The following fix is included but yet to be released, as of 2019/01/03

cs_set:Nn __um_fontdimen_to_percent:nN

  {

    fp_eval:n { dim_to_decimal:n { 65536 fontdimen #1 #2 } / 100 }

  }

% Re-implementing the ``correct'' font dimensions

defacta@angle#1#2{%

  mathord{%

    mkern1mu%

    vbox{hrule height __um_overbar_rule:N #1 hbox{    % <- seems right

        vbox{%

          kern dim_eval:n { __um_overbar_vgap:N #1 }   % <- something missing?

          hbox{$m@th#1#2$}}%

        setboxz@hbox{$#1mathstrut$}%

        vrule width __um_overbar_rule:N #1 depthdpz@}}% <- seems right

    mkern1mu}}

ExplSyntaxOff

makeatother

newcommand*test[2]{% #1: font name, #2 mathversion

  mathversion{#2}$mathup{#1}$

  & mathversion{#2}$displaystyle     overline{30},angl{30}$

  & mathversion{#2}$textstyle        overline{30},angl{30}$

  & mathversion{#2}$scriptstyle      overline{30},angl{30}$

  & mathversion{#2}$scriptscriptstyleoverline{30},angl{30}$%

}

pagestyle{empty}

begin{document}

begin{tabular}{l l l l l}

test{Latin Modern Math}{LM} \

test{TG Bonum Math}{Bonum} \

test{TG Pagella Math}{Pagella} \

test{TG Schola Math}{Schola} \

test{TG Termes Math}{Termes} \

test{Fira Math}{Fira} \

end{tabular}

end{document}

Disclaimer: I have used private macros __um_..., which is considered as bad practice. Please try not to copy and paste the above “fix” unless you know what you are doing.

Questions

Which are the correct fontdimen’s, and how can I use them?