Optidef constraint name not aligned properly

Question

I have tried to follow optidef.tex to make the constraint names aligned. Here is the code:

\begin{mini*}|s|
    {\substack{\mathbf{x},\mathbf{u}}}{J = \phi\left(\mathbf{x}\left(t_f\right),t_f\right) + \int_{t_0}^{t_f}\mathcal{L}\left(\mathbf{x},\mathbf{u},t\right)dt}{}{}
            \addConstraint{\dot{\mathbf{x}} =}{\mathbf{f}\left(\mathbf{x},\mathbf{u},t\right)}
            \addConstraint{\mathbf{x}\left(0\right)}{= \mathbf{x}_0,}{\quad\text{(Initial condition)}}
            \addConstraint{\Psi\left(\mathbf{x}_f,t_f\right) }{= 0,} {\quad\text{(Terminal constraints)}}
    \end{mini*}

And here is the corresponding output that is circled:

Can some explain what is going on? I am writing this out in TeXpad.

Answer 1

I'll be happy to remove this post. According to what I find, you just need to move the position of the = sign in

\addConstraint{\dot{\mathbf{x}} =}{\mathbf{f}\left(\mathbf{x},\mathbf{u},t\right)}

to become

\addConstraint{\dot{\mathbf{x}}}{ =\mathbf{f}\left(\mathbf{x},\mathbf{u},t\right)}

Full example:

\documentclass{article}
\usepackage{optidef}
\begin{document}
\begin{mini*}|s|
    {\substack{\mathbf{x},\mathbf{u}}}{J = \phi\left(\mathbf{x}\left(t_f\right),t_f\right) + \int_{t_0}^{t_f}\mathcal{L}\left(\mathbf{x},\mathbf{u},t\right)dt}{}{}
            \addConstraint{\dot{\mathbf{x}} }{=\mathbf{f}\left(\mathbf{x},\mathbf{u},t\right)}
            \addConstraint{\mathbf{x}\left(0\right)}{= \mathbf{x}_0,}{\quad\text{(Initial condition)}}
            \addConstraint{\Psi\left(\mathbf{x}_f,t_f\right) }{= 0,} {\quad\text{(Terminal constraints)}}
\end{mini*}
\end{document}

There are many things one could improve like the \substack or the differential d.

Answer 2

Just for reference, here's a solution which does not employ the optidef package. Instead, it employs 1 align* environment (for the 4 "outer" blocks) and 1 aligned environment (for the group of 3 "inner" blocks which, together, make up the final "outer" block). align* and aligned are both provided by the amsmath package.

\documentclass{article}
\usepackage{amsmath,amssymb,bm}
\newcommand{\diff}{\mathop{}\!\mathrm{d}} % see https://tex.stackexchange.com/a/385862/5001
\newcommand\tf{t\mkern-1.5mu_f} % snug up the subscript-f to the t

\begin{document}
\begin{align*}
&\text{PM: } \textstyle 
  J=\phi(\mathbf{x}(\tf),\tf)
  +\int_{t_0}^{\tf}\!\mathcal{L}(\mathbf{x},\mathbf{u},t)\diff t \\
&\text{General Problem: } \textstyle 
  \min J=\phi(\mathbf{x}(\tf))
  +\int_{t_0}^{\tf}\! \mathcal{L}(\mathbf{x},\mathbf{u},t)\diff t\\
&{\min_{\mathbf{x},\mathbf{u}}} \quad
  J=\phi(\mathbf{x}(\tf),\tf)
  +\int_{t_0}^{\tf} \!\! \mathcal{L}(\mathbf{x}, \mathbf{u},t)\diff t \\
&\text{s.t.} \quad
  \begin{aligned}[t]
  \dot{\mathbf{x}} &= \mathbf{f}(\mathbf{x},\mathbf{u},t) \\
  \mathbf{x}(0) &= x_0 \qquad\text{(Initial condition)}\\
  \bm{\Psi}(\mathbf{x}_f,\tf) &= 0_{\phantom{0}} \qquad{\text{(Terminal constraint)}}
  \end{aligned}
\end{align*}

\end{document} 

Answer 3

I cannot reproduce the output you get with your code.

Besides moving the = to the second argument to \AddConstraint, as suggested in another answer, I propose a different setting, with constraints left aligned, which seems better to me.

Also I removed all \left/\right pairs, which do nothing besides adding unwanted space. A \, before dt is necessary, instead. Whether you want an upright “d” or not is your choice: conform to what is the usage in your field (or to your personal preference).

\documentclass{article}
\usepackage{amsmath,optidef}

\begin{document}

\subsection*{Original}

\begin{mini*}|s|
    {\substack{\mathbf{x},\mathbf{u}}}{J = \phi\left(\mathbf{x}\left(t_f\right),t_f\right) + \int_{t_0}^{t_f}\mathcal{L}\left(\mathbf{x},\mathbf{u},t\right)dt}{}{}
            \addConstraint{\dot{\mathbf{x}} =}{\mathbf{f}\left(\mathbf{x},\mathbf{u},t\right)}
            \addConstraint{\mathbf{x}\left(0\right)}{= \mathbf{x}_0,}{\quad\text{(Initial condition)}}
            \addConstraint{\Psi\left(\mathbf{x}_f,t_f\right) }{= 0,} {\quad\text{(Terminal constraints)}}
    \end{mini*}

\subsection*{Left alignment}

\begin{mini*}|s|
  {\scriptstyle\mathbf{x},\mathbf{u}}
  {J = \phi(\mathbf{x}(t_f),t_f) + \int_{t_0}^{t_f}\mathcal{L}(\mathbf{x},\mathbf{u},t)\,dt}{}{}
\addConstraint{}{\dot{\mathbf{x}}=\mathbf{f}(\mathbf{x},\mathbf{u},t)}{}
\addConstraint{}{\mathbf{x}(0)= \mathbf{x}_0,}{\quad\text{(Initial condition)}}
\addConstraint{}{\Psi(\mathbf{x}_f,t_f)= 0,}{\quad\text{(Terminal constraints)}}
\end{mini*}

\subsection*{Alignment at equals signs}

\begin{mini*}|s|
  {\scriptstyle\mathbf{x},\mathbf{u}}
  {J = \phi(\mathbf{x}(t_f),t_f) + \int_{t_0}^{t_f}\mathcal{L}(\mathbf{x},\mathbf{u},t)\,dt}{}{}
\addConstraint{\dot{\mathbf{x}}}{=\mathbf{f}(\mathbf{x},\mathbf{u},t)}{}
\addConstraint{\mathbf{x}(0)}{= \mathbf{x}_0,}{\quad\text{(Initial condition)}}
\addConstraint{\Psi(\mathbf{x}_f,t_f)}{= 0,}{\quad\text{(Terminal constraints)}}
\end{mini*}

\end{document}

In my opinion, optidef should have an option to make the subscript variables in script size instead of normal size. I opted for a simpler \scriptstyle declaration in place of the more complicated \substack.

As a further simplification for inputting your formulas, I suggest to add something like

\newcommand{\vectorvariable}[1]{\mathbf{#1}}
\newcommand{\vf}{\vectorvariable{f}}
\newcommand{\vu}{\vectorvariable{u}}
\newcommand{\vx}{\vectorvariable{x}}

so, for instance, the first constraint would become

\addConstraint{}{\dot{\vx}}=\vf(\vx,\vu,t)}{}

In case a reviewer tells you that vector variables should be in bold italic, you'll just have to add

\usepackage{bm}

and change a single line in your code, namely

\newcommand{\vectorvariable}[1]{\bm{#1}}

After this, running LaTeX on your document would yield

(or the other layout, according to the one you prefer, of course). Similarly, adding

\newcommand{\diff}{\mathop{}\!d}

and typing \diff t when you want the differential, will allow you to cope with fussy reviewers or supervisors requiring an upright “d” by just changing a single line of code. Note that if you go this way, the \, in front of the differential in integrals should be removed.