Bxdty

So I have this basic code:

But the fractional alignment looks awful, any tips to fix this?

documentclass[11pt, a4paper]{report}

usepackage{bm}

usepackage{amsfonts, graphicx, verbatim, amsmath,amssymb, amsthm, mathrsfs, eufrak}

usepackage{color}

usepackage{array}

usepackage{setspace}% if you must (for double spacing thesis)

usepackage{fancyhdr}

usepackage{enumitem}

usepackage{tikz}

usepackage{parskip}

usepackage{lipsum}

usepackage{floatrow}

usepackage{commath}

providecommand{U}[1]{protectrule{.1in}{.1in}}





begin{document}



Let $h$ be a linear mapping of $V_1$ into $V_2$. Let

[

h^0 = frac{1}{abs G} sum_{tin G} (rho_t^2)^{-1} h rho_t^1

]

where $abs G$ is the order of $G$. Then

begin{enumerate}[label=(roman*)]

item If $rho^1$ and $rho^2$ are not isomorphic then we have $h^0=0$

item If $V_1 = V_2$ and $rho^1 = rho^2$, $h^0$ is a scalar multiple of the identity, the scalar being $frac{1}{n} Tr(h)$ where $n = dim(V_1)$

end{enumerate}



end{document}

as you notice, it looks awful.

The commath package provides a highly questionable implementation of abs. Don't load this package unless you absolutely have to. (Aside: You do not have to for the example at hand.)

Instead of loading this package, I suggest you

• load the mathtools package instead of the amsmath package. The mathtools package is, for all intents and purposes, a superset of the amsmath package; and



• 
  

  issue the instruction

  
  
  

  DeclarePairedDelimiter{abs}{lvert}{rvert}
  
  

  
  
  

  in the preamble. This sets up a proper implementation of abs{...}. If you need to auto-size the vertical bars, simply use abs*{...}.

  
  

*Addendum": More about how commath implements envert (abs is just an alias for envert): If envert{<arg>} is invoked without an optional sizing argument (a whole number between 0 and 4), the following code gets executed:

 !leftlvert#2rightrvert

It's the ! ("negative thinspace") instruction that messes up the symmetry of the full expression. Either write abs[0]{<arg>} to execute lvert#2rvert or, more safely, don't load commath at all and go for a cleaner definition of abs. The definitions of abs[1]{...} thru abs[4]{...} are also messed up, by the way, as are the definitions of norm and norm[1] thru norm]4]. While it's possible, in principle, to correct the faulty definitions of envert, abs, enVert, and norm, one is better off going for a setup that avoids these complications...

Compare the output of

documentclass[11pt, a4paper]{report}

usepackage{amsmath}

usepackage{enumitem}

usepackage{commath}



DeclareMathOperator{Tr}{Tr}



begin{document}



Let $h$ be a linear mapping of $V_1$ into $V_2$. Let

[

h^0 = frac{1}{abs G} sum_{tin G} (rho_t^2)^{-1} h rho_t^1

]

where $abs G$ is the order of $G$. Then

begin{enumerate}[label=(roman*)]

item If $rho^1$ and $rho^2$ are not isomorphic then we have $h^0=0$

item If $V_1 = V_2$ and $rho^1 = rho^2$, $h^0$ is a scalar multiple of the identity,

      the scalar being $frac{1}{n} Tr(h)$ where $n = dim(V_1)$

end{enumerate}



end{document}

with the output of

documentclass[11pt, a4paper]{report}

usepackage{amsmath}

usepackage{enumitem}



makeatletter

% begin definitions for a "better commath"



newcommand{dif}{mathop{}!mathrm{d}}

newcommand{Dif}{mathop{}!mathrm{D}}



newcommand{spx}[1]{%

  ifrelaxdetokenize{#1}relax

    expandafter@gobble

  else

    expandafter@firstofone

  fi

  {^{#1}}%

}



newcommandpd[3]{frac{partialspx{#1}#2}{partial#3spx{#1}}}

newcommandtpd[3]{tfrac{partialspx{#1}#2}{partial#3spx{#1}}}

newcommanddpd[3]{dfrac{partialspx{#1}#2}{partial#3spx{#1}}}



newcommand{md}[6]{frac{partialspx{#2}#1}{partial#3spx{#4}partial#5spx{#6}}}

newcommand{tmd}[6]{tfrac{partialspx{#2}#1}{partial#3spx{#4}partial#5spx{#6}}}

newcommand{dmd}[6]{dfrac{partialspx{#2}#1}{partial#3spx{#4}partial#5spx{#6}}}



newcommand{od}[3]{frac{difspx{#1}#2}{dif#3spx{#1}}}

newcommand{tod}[3]{tfrac{difspx{#1}#2}{dif#3spx{#1}}}

newcommand{dod}[3]{dfrac{difspx{#1}#2}{dif#3spx{#1}}}



newcommand{genericdel}[4]{%

  ifcase#3relax

  ifx#1.else#1fi#4ifx#2.else#2fior

  bigl#1#4bigr#2or

  Bigl#1#4Bigr#2or

  biggl#1#4biggr#2or

  Biggl#1#4Biggr#2else

  left#1#4right#2fi

}

newcommand{del}[2][-1]{genericdel(){#1}{#2}}

newcommand{set}[2][-1]{genericdel{}{#1}{#2}}

letcbrset

newcommand{sbr}[2][-1]{genericdel{#1}{#2}}

letintoodel

letintccsbr

newcommand{intoc}[2][-1]{genericdel(]{#1}{#2}}

newcommand{intco}[2][-1]{genericdel[){#1}{#2}}

newcommand{eval}[2][-1]{genericdel.|{#1}{#2}}

newcommand{envert}[2][-1]{genericdel||{#1}{#2}}

letabsenvert

newcommand{sVert}[1][0]{%

  ifcase#1relax

  rvertorbigr|orBigr|orbiggr|orBiggr

  fi

}

newcommand{enVert}[2][-1]{genericdel||{#1}{#2}}

letnormenVert

newcommand{fullfunction}[5]{%

  begin{array}{@{}r@{}r@{}c@{}l@{}}

  #1 colon & #2 & {}longrightarrow{} & #3 \

            & #4 & {}longmapsto{}     & #5

  end{array}

}

%%% end of the definitions

makeatother





DeclareMathOperator{Tr}{Tr}



begin{document}



Let $h$ be a linear mapping of $V_1$ into $V_2$. Let

[

h^0 = frac{1}{abs G} sum_{tin G} (rho_t^2)^{-1} h rho_t^1

]

where $abs G$ is the order of $G$. Then

begin{enumerate}[label=(roman*)]

item If $rho^1$ and $rho^2$ are not isomorphic then we have $h^0=0$

item If $V_1 = V_2$ and $rho^1 = rho^2$, $h^0$ is a scalar multiple of the identity, 

      the scalar being $frac{1}{n} Tr(h)$ where $n = dim(V_1)$

end{enumerate}



end{document}

The code has already been presented in https://tex.stackexchange.com/a/135985/4427 and provides a full replacement of commath, avoiding all its weaknesses. If you're tied to commath because you use its macros and features in other places, this can save you. Otherwise define abs with DeclarePairedDelimiter as suggested by Mico.

For instance, you can notice the reduced space between “where” and “|G|” in the commath version.

You should define anyway an operator name for “Tr”.

If you decide to go for this, save the part between

% begin definitions for a "better commath"

and

% end of the definitions

in a file called bettercommath.sty in the working directory and do usepackage{bettercommath}.