# How does the combination of \huge and \bm work?

I've recently come up with some strange behaviors of simultaneous applications of \huge and \bm. In particular, the results are not consistent depending on the loaded packages.

MWE 1: (with the minimum loaded packages)

\documentclass{article}
\usepackage{mathtools}
\usepackage{bm}

\begin{document}
$\Delta \qquad \bm{\Delta} \qquad {\Huge\bm{\Delta}}$

$q \qquad \bm{q} \qquad {\Huge\bm{q}}$
\end{document}


Result 1:

In this case, \bm works, but \huge has no influence.

MWE 2: (with the loaded packages corresponding to a specific project of mine)

\documentclass{article}
\usepackage{amssymb}
\usepackage{amsthm}
\usepackage{mathtools}
\usepackage{newtxtext,newtxmath} % for times font
\usepackage[overload]{empheq} % for subnumbering in cases
\usepackage[framemethod=tikz]{mdframed}
\usepackage{bm}
\usepackage{algorithm}
\usepackage[a4paper, left = 1.5cm, right = 1.5cm, bottom = 1.5cm, top = 1.5cm]{geometry}
\usepackage{braket}
\usepackage[T1]{fontenc}
\usepackage[ansinew,utf8]{inputenc}

\begin{document}
$\Delta \qquad \bm{\Delta} \qquad {\Huge\bm{\Delta}}$

$q \qquad \bm{q} \qquad {\Huge\bm{q}}$
\end{document}


Result 2:

This one seems even more strange, and the results are somewhat unexpected!

Are these behaviors justifiable?

Edit:

I have a matrix equation, and there is an \underbrace for each matrix. The names of the matrices, e.g., \Delta and q, are about to be put there. But the labels are so small, and I need to make them bigger. What is the best practice to do this?

• scalebox,
• mathlarger,
• ...
\documentclass{article}
\usepackage{amssymb}
\usepackage{amsthm}
\usepackage{mathtools}
\usepackage{newtxtext,newtxmath} % for times font
\usepackage[overload]{empheq} % for subnumbering in cases
\usepackage[framemethod=tikz]{mdframed}
\usepackage{bm}
\usepackage{algorithm}
\usepackage[a4paper, left = 1.5cm, right = 1.5cm, bottom = 1.5cm, top = 1.5cm]{geometry}
\usepackage{braket}
\usepackage[T1]{fontenc}
\usepackage[ansinew,utf8]{inputenc}

\begin{document}
$$\label{eq:vec1} \nabla \xi(\bm{q}) = \begin{bmatrix} \nabla \xi(q^{0}) \\ \nabla \xi(q^{1}) \\ \nabla \xi(q^{2}) \\ \nabla \xi(q^{3}) \\ \nabla \xi(q^{4}) \\ \nabla \xi(q^{5}) \\ \nabla \xi(q^{6}) \end{bmatrix} = \underbrace{\begin{bmatrix} 2\lambda_{1}&2\lambda_{3}&2\lambda_{3} &0&0&0&0\\ 2\lambda_{3}&2\lambda_{1}&2\lambda_{3} &0&0&0&0\\ 2\lambda_{3}&2\lambda_{3}&2\lambda_{1} &0&0&0&0\\ 0&0&0&2\lambda_{1}&2\lambda_{3}&0&0\\ 0&0&0&2\lambda_{3}&2\lambda_{1}&0&0\\ 0&0&0&0&0&2\lambda_{1}&2\lambda_{3}\\ 0&0&0&0&0&2\lambda_{3}&2\lambda_{1} \end{bmatrix}}_{\huge\bm{\Lambda}} \underbrace{\begin{bmatrix} q^{0}\\q^{1}\\q^{2}\\q^{3}\\q^{4}\\q^{5}\\q^{6} \end{bmatrix}}_{\huge\bm{q}} +2\omega\lambda_{2} \underbrace{\begin{bmatrix} \mathcal{Q}(q^{0},q^{1}) + \mathcal{Q}(q^{0},q^{2})\\ \mathcal{Q}(q^{1},q^{0}) + \mathcal{Q}(q^{1},q^{2})\\ \mathcal{Q}(q^{2},q^{0}) + \mathcal{Q}(q^{2},q^{1})\\ \mathcal{Q}(q^{3},q^{4})\\ \mathcal{Q}(q^{4},q^{3})\\ \mathcal{Q}(q^{5},q^{6})\\ \mathcal{Q}(q^{6},q^{5}) \end{bmatrix}}_{\huge\bm{\Omega}} + \underbrace{\begin{bmatrix} 2\bigl(\lambda_{1}q^{0}_{\mathcal{T}} + \lambda_{3}(q^{1}_{\mathcal{T}} + q^{2}_{\mathcal{T}})\bigr)\\ 2\bigl(\lambda_{1}q^{1}_{\mathcal{T}} + \lambda_{3}(q^{0}_{\mathcal{T}} + q^{2}_{\mathcal{T}})\bigr)\\ 2\bigl(\lambda_{1}q^{2}_{\mathcal{T}} + \lambda_{3}(q^{0}_{\mathcal{T}} + q^{1}_{\mathcal{T}})\bigr)\\ 2\bigl(\lambda_{1}q^{3}_{\mathcal{T}} + \lambda_{3}q^{4}_{\mathcal{T}}\bigr)\\ 2\bigl(\lambda_{1}q^{4}_{\mathcal{T}} + \lambda_{3}q^{3}_{\mathcal{T}}\bigr)\\ 2\bigl(\lambda_{1}q^{5}_{\mathcal{T}} + \lambda_{3}q^{6}_{\mathcal{T}}\bigr)\\ 2\bigl(\lambda_{1}q^{6}_{\mathcal{T}} + \lambda_{3}q^{5}_{\mathcal{T}}\bigr) \end{bmatrix}}_{\huge\bm{C}} = 0,$$
\end{document}


• Command \Huge invalid in math mode – egreg Nov 28 '18 at 15:12
• @egreg: I see. Then, what is the best practice to change the size of stuffs in math mode? scalebox? – Roboticist Nov 28 '18 at 15:13
• There is \mathlarger from the relsize package, if you really need to do it. What precisely is your aim? – egreg Nov 28 '18 at 15:14
• @egreg: I have a matrix equation, and there is an \underbrace for each matrix. The names of the matrices, e.g., \Delta and q are about to be put there. But the labels are so small, and I need to make them bigger. – Roboticist Nov 28 '18 at 15:17
• {\underbrace{whatever}_{\textstyle\bm{\Delta}}} – egreg Nov 28 '18 at 15:19

You get a warning Command \Huge invalid in math mode and the result is thus unpredictable.

Use \textstyle:

\documentclass{article}
\usepackage{newtxtext,newtxmath} % for times font
\usepackage{amsmath}
\usepackage{bm}

\begin{document}

$$\label{eq:vec1} \begin{split} \nabla \xi(\bm{q}) &= \begin{bmatrix} \nabla \xi(q^{0}) \\ \nabla \xi(q^{1}) \\ \nabla \xi(q^{2}) \\ \nabla \xi(q^{3}) \\ \nabla \xi(q^{4}) \\ \nabla \xi(q^{5}) \\ \nabla \xi(q^{6}) \end{bmatrix} = {\underbrace{\begin{bmatrix} 2\lambda_{1}&2\lambda_{3}&2\lambda_{3} &0&0&0&0\\ 2\lambda_{3}&2\lambda_{1}&2\lambda_{3} &0&0&0&0\\ 2\lambda_{3}&2\lambda_{3}&2\lambda_{1} &0&0&0&0\\ 0&0&0&2\lambda_{1}&2\lambda_{3}&0&0\\ 0&0&0&2\lambda_{3}&2\lambda_{1}&0&0\\ 0&0&0&0&0&2\lambda_{1}&2\lambda_{3}\\ 0&0&0&0&0&2\lambda_{3}&2\lambda_{1} \end{bmatrix}}_{\textstyle\bm{\Lambda}}} {\underbrace{\begin{bmatrix} q^{0}\\q^{1}\\q^{2}\\q^{3}\\q^{4}\\q^{5}\\q^{6} \end{bmatrix}}_{\textstyle\bm{q}}} \\[2ex] &\quad+2\omega\lambda_{2} {\underbrace{\begin{bmatrix} \mathcal{Q}(q^{0},q^{1}) + \mathcal{Q}(q^{0},q^{2})\\ \mathcal{Q}(q^{1},q^{0}) + \mathcal{Q}(q^{1},q^{2})\\ \mathcal{Q}(q^{2},q^{0}) + \mathcal{Q}(q^{2},q^{1})\\ \mathcal{Q}(q^{3},q^{4})\\ \mathcal{Q}(q^{4},q^{3})\\ \mathcal{Q}(q^{5},q^{6})\\ \mathcal{Q}(q^{6},q^{5}) \end{bmatrix}}_{\textstyle\bm{\Omega}}} + {\underbrace{\begin{bmatrix} 2\bigl(\lambda_{1}q^{0}_{\mathcal{T}} + \lambda_{3}(q^{1}_{\mathcal{T}} + q^{2}_{\mathcal{T}})\bigr)\\ 2\bigl(\lambda_{1}q^{1}_{\mathcal{T}} + \lambda_{3}(q^{0}_{\mathcal{T}} + q^{2}_{\mathcal{T}})\bigr)\\ 2\bigl(\lambda_{1}q^{2}_{\mathcal{T}} + \lambda_{3}(q^{0}_{\mathcal{T}} + q^{1}_{\mathcal{T}})\bigr)\\ 2\bigl(\lambda_{1}q^{3}_{\mathcal{T}} + \lambda_{3}q^{4}_{\mathcal{T}}\bigr)\\ 2\bigl(\lambda_{1}q^{4}_{\mathcal{T}} + \lambda_{3}q^{3}_{\mathcal{T}}\bigr)\\ 2\bigl(\lambda_{1}q^{5}_{\mathcal{T}} + \lambda_{3}q^{6}_{\mathcal{T}}\bigr)\\ 2\bigl(\lambda_{1}q^{6}_{\mathcal{T}} + \lambda_{3}q^{5}_{\mathcal{T}}\bigr) \end{bmatrix}}_{\textstyle\bm{C}}} \\ &= 0, \end{split}$$

\end{document}


You could use \textstyle\mathlarger{\bm{\Lambda}}, but I don't see a real reason for it (requires \usepackage{relsize}).