# How to draw straight lines between nodes in commutative diagram without using TikZ matrix library?

I want to draw a commutative diagram with TikZ as follows:

\usepackage{tikz}
\usetikzlibrary{positioning}

\begin{tikzpicture}[auto]
\node (S1) {$\sigmaSet \cap X_N$};
\node (S2) [below= 2cm and 4cm of S1] {$\sigmaSet \cap X_N$};
\node (S3) [below= 2cm and 4cm of S2] {$\discreteSigmaSet$};
\node (U1) [right= 2cm and 4cm of S1] {$\potentialSpace$};
\node (U2) [below= 2cm and 4cm of U1] {$Y_M$};
\node (U3) [below= 2cm and 4cm of U2] {$\R^M$};
\draw[->] (S1) to node {$F_f$} (U1);
\draw[->] (S2) to node {$\widetilde{F}_f$} (U2);
\draw[->] (S3) to node {$\widehat{F}_f$} (U3);
\draw[->] (S1) to node [swap] {$I$} (S2);
\draw[->] (S2) to node [swap] {$\theta_{B_N}$} (S3);
\draw[->] (U1) to node {$P^{\psprod{E}{\cdot}{\cdot}}_{Y_M}$} (U2);
\draw[->] (U2) to node {$\theta_{C_M}$} (U3);
\end{tikzpicture}


which looks like this: Note how the horizontal arrows are not straight, but end too high on the right side.

Now my question is how I can make TikZ draw the horizontal arrows as straight lines. Is there a way to accomplish this without having to use the matrix library? I want to keep the code above as simple as possible.

• The arrows are straight (they’re just not horizontal). The problem is the placement of the nodes which is not consistent in the way you use it. Why don’t you simply use a matrix (which is used for tikz-cd too)? The on grid option does help here too (as it uses the center anchor for both nodes: the one that is referenced and the one that is placed)) but has other disadvantages (overlapping nodes). Jun 19, 2013 at 13:06
• Adding the on grid option looks pretty good to me. Jun 19, 2013 at 14:56

You can use the (p |- q) notation, meaning the intersection of a vertical line through p and a horizontal line through q.

## Code

\documentclass[tikz,border=2mm]{standalone}
\usetikzlibrary{positioning}

\begin{document}

\begin{tikzpicture}[auto]
\node (S1) {$\sigma \cap X_N$};
\node (S2) [below= 2cm and 4cm of S1] {$\sigma \cap X_N$};
\node (S3) [below= 2cm and 4cm of S2] {$\sigma$};
\node (U1) [right= 2cm and 4cm of S1] {$\gamma$};
\node (U2) at (U1 |- S2) {$Y_M$};
\node (U3) at (U2 |- S3) {$R^M$};
\draw[->] (S1) to node {$F_f$} (U1);
\draw[->] (S2) to node {$\widetilde{F}_f$} (U2);
\draw[->] (S3) to node {$\widehat{F}_f$} (U3);
\draw[->] (S1) to node [swap] {$I$} (S2);
\draw[->] (S2) to node [swap] {$\theta_{B_N}$} (S3);
\draw[->] (U1) to node {$P^{\hat{E}{\cdot}{\cdot}}_{Y_M}$} (U2);
\draw[->] (U2) to node {$\theta_{C_M}$} (U3);
\end{tikzpicture}

\end{document}


## Output • @Joerg: You're very welcome. Jul 11, 2013 at 9:58

Would tikz-cd help?

I've mocked up the symbols you didn't define.

\documentclass{article}
\usepackage{tikz-cd}
\begin{document}

\begin{tikzcd}[column sep=large,row sep=large]
\Sigma \cap X_N
\arrow{r}{F_f}
\arrow[swap]{d}{I}
&
\mathcal{P}
\arrow{d}{P^{\Pi(E){\cdot}{\cdot}}}
\\
\Sigma \cap X_N
\arrow{r}{\tilde{F}_f}
\arrow[swap]{d}{\theta_{B_N}}
&
Y_M
\arrow{d}{\theta_{C_M}}
\\
\Sigma
\arrow{r}{\hat{F}_f}
&
R^M
\end{tikzcd}
\end{document} • Oh, I didn't realize there was a package for commutative diagrams. All examples I found online were using Tikz's matrix library. Thanks a lot for the hint, I will vertainly give it a try. But coming back to my original question: is there a simple Tikz option to accomplish the same with my existing code above? P.S.: Sorry for not providing all symbol definitions. Jun 19, 2013 at 11:53
• @Joerg You can still edit your question to include a minimal working example (MWE). Jun 19, 2013 at 12:58