I need to set subscripts and superscripts automatically in mathematical equations in roman font style for a book publication. I do not want to apply \mathrm{...} to each subscript or superscript separately.

Fortunately I found a solution to globally set subscripts in roman font style in Redefine underscore to produce roman subscript. I made a slight adaptation in order to have all indexes in sequence in roman font style:


Original typeset $x_{abc}^{de}$: original

Roman subscripts with code above $x_{abc}^{de}$: roman subscript

I can, however, not figure out how to apply this code to superscripts. Does anyone have a solution so set both the subscript and superscript font style to roman?

  • can't you just replace _ by ^ and \sb by \sp ? – David Carlisle Jul 19 '17 at 7:47
  • I assume you never raise to the power of a variable, and instead only use descriptive superscripts in your field? Because in $x \mathrm{e}^{\mathrm{x}}$ the two xs are different entities, while in $x \mathrm{e}^x$ they're not. – Chris H Jul 19 '17 at 8:53
  • 1
    Correct, I only use descriptive superscripts. – christiankral Jul 19 '17 at 9:52

I fail to see the rationale for this, however here it is:




  \catcode`_=12 \catcode`^=12




enter image description here

  • I really hope this is because it's a book about a field that only uses descriptive text superscripts rather than powers. I've come across this but can;t recall what uses it. – Chris H Jul 19 '17 at 8:56
  • @egreg, thanks a lot for your solution. It works very well for me. – christiankral Jul 19 '17 at 9:54
  • The (German) publisher insists on roman subscripts and superscripts. So the only choice I have is to either manually re-format (some older) LaTeX code use this trick to globally change the format. – christiankral Jul 19 '17 at 9:58
  • 2
    @christiankral If the publisher tells me that $x_{n}$ must have the “n” in upright type, I tell them they don't know about mathematics. Of course, this depends on what the subscripts/superscripts contain. – egreg Jul 19 '17 at 10:50

Here's a LuaLaTeX-based solution. It doesn't change the catcodes of _ and ^. Instead, it sets up a Lua function and assigns it to the process_input_buffer callback; that way, all instances of _{...} and ^{...} are converted to _{\mathrm{...}} and ^{\mathrm{...}}, respectively, before LaTeX even starts its usual processing.

Whitespace between _/^ and the material enclosed in curly braces is OK. However, the presence of the curly braces is significant: the Lua function will not modify the appearance of y^x and x_n, say.

enter image description here

% !TeX program = lualatex

function make_roman ( s )
  s = string.gsub ( s , "_%s-(%b{})",  "_{\\mathrm%1}" )
  s = string.gsub ( s , "%^%s-(%b{})", "^{\\mathrm%1}" )
  return s 
\AtBeginDocument{\directlua{luatexbase.add_to_callback (
  "process_input_buffer", make_roman , "make_roman" )}}

$x_{abc}^ {de}$ 

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.