I guess this is a very simple question, but I couldn't find the answer... How can I define a variable in TikZ? I want to have something like, for example:
\draw[->] (x1,y1)--(x2,y2); Such that x1,x2,y1,y2 are defined somewhere else and can be changed in order to adjust the picture.
Should be simple, right?
\begin{tikzpicture} \coordinate (A) at (2,3); \coordinate (B) at (2,5); \draw[->] (A)--(B) ; \end{tikzpicture} You can add
\newcommand\XA{2} etc. and then \coordinate (A) at (\XA,\YA);
\coordinate s are just special \node s and they are defined global, i.e. you could define A and B in another tikzpicture, e.g. an otherwise empty one at the beginning of the document. Tikz offers a powerful solution to dealing with variables through the library math (TikZ version >= 3.00)
\documentclass{article} \usepackage{tikz} \begin{document} \usetikzlibrary{math} %needed tikz library % Variables must be declared in a tikzmath environment but % can be used outside (almost anywhere) % Do not forget SEMICOLONS after each command (!) \tikzmath{\x1 = 1; \y1 =1; % Computations are also possible \x2 = \x1 + 1; \y2 =\y1 +3; } % Using the variables for drawing \begin{tikzpicture} \draw[very thick, -stealth] (\x1, \y1)--(\x2, \y2); \end{tikzpicture} % Using the variables in math and text modi The coordinates of the second point are: $x_2 = \x2$ and $y_2 =$ \y2 % Note that the name of the variables, e.g., \x1, % can contain a number as the LAST name characters (!) % That would be not allowed with a \newcommand definition \end{document} 
See also this answer for more examples with different uses.
I'm using \def\x{1068} to have x as a variable in my script to generate scalebars onto images with TikZ. That works like a charm for me.
A working example is pasted below, even though the image does not look very nice :)
\documentclass{article} \usepackage[demo]{graphicx} \usepackage{tikz} \usepackage{siunitx} \usepackage[graphics,tightpage,active]{preview} \PreviewEnvironment{tikzpicture} \newcommand{\imsize}{\linewidth} \newlength\imagewidth % needed for scalebars \newlength\imagescale % ditto \begin{document}% %------------- \pgfmathsetlength{\imagewidth}{\imsize}% \pgfmathsetlength{\imagescale}{\imagewidth/1728}% \def\x{1068}% scalebar-x at golden ratio of x=1728px \def\y{320}% scalebar-y at 90% of height of y=356px \def\shadow{11}% shadow parameter for scalebar \begin{tikzpicture}[x=\imagescale,y=-\imagescale] \clip (0,0) rectangle (1728,356); \node[anchor=north west, inner sep=0pt, outer sep=0pt] at (0,0) {\includegraphics[width=\imagewidth]{image}}; % 279px = 1.7819mm > 100px = 638um > 78px = 500um, 16px = 100um \draw[|-|,blue,thick] (791,151) -- (1020,311) node [sloped,midway,above,fill=white,semitransparent,text opacity=1] {\SI{1.7819}{\milli\meter} (1204px) TEMPORARY!}; \draw[|-|,thick] (\x+\shadow,\y+\shadow) -- (\x+78+\shadow,\y+\shadow) node [midway, above] {\SI{500}{\micro\meter}}; \draw[|-|,white,thick] (\x,\y) -- (\x+78,\y) node [midway,above] {\SI{500}{\micro\meter}}; \draw[color=red, anchor=south west] (0,356) node [fill=white, semitransparent] {Legend} node {Legend}; \end{tikzpicture}% %------------- \end{document}% 
\def\varname{initialValue}. I think your example is far too complicated to show use of a simple variable. \dev\varname{initialValue}. \varname/2? \varname*0.5? \x+78+\shadow in the fifth line from bottom. Dividing works too. Try changing said line to \draw[|-|,white,thick] (\x/2,\y) -- (\x/2+78,\y) node [midway,above] {\SI{500}{\micro\meter}}; and you see that the white scalebar moves... I use \pgfmathsetmacro{VARName}{VARValue} inside a tikzpicture environment.
For example:
\begin{tikzpicture} %VARIABLES \pgfmathsetmacro{\PHI}{-15} % Now use \PHI anywhere you want -15 to appear, % can also be used in calculations like 2*\PHI \end{tikzpicture} 
\begin{tikzpicture} \def\PHI{-15} \node (o) at (0,0) {o}; \node (a) at (\PHI mm,\PHI mm) {a}; \end{tikzpicture} also works. However this command is useful to store the result of a mathematical operation such as \pgfmathsetmacro{\PHI}{cos(60)} \pgfmathsetmacro but there is no harm in using it for constants. \pgfmathsetmacro: \begingroup% \pgfmathsetlength\pgfmath@x{#2}% \edef#1{\pgfmath@tonumber{\pgfmath@x}}% \pgfmath@smuggleone{#1}% \endgroup% . The second remark you can use \pgfmathsetmacro outside a tikzpicture environment. Building on the example in the comments to the question, here's a more complicated example that uses some computations to find the desired coordinates and then uses them.
\documentclass{minimal} \usepackage{tikz} \usetikzlibrary{intersections} \begin{document} \begin{tikzpicture} \draw[gray,thin] (-1,-1) grid (5,5); \draw[ultra thick,->] (-1,0) -- (5.2,0); \draw[ultra thick,->] (0,-1) -- (0,5.2); \clip (-1,-1) rectangle (5,5); \foreach \m/\c/\col in { 3/{(2,0)}/blue, 2/{(1,1)}/red, -.25/{(0,4)}/green% } { % Separate the coordinate \c into x and y components \fill[\col] \c circle (2pt); \pgfgetlastxy{\vx}{\vy} % Work out the y-height at one end of the line \pgfmathsetmacro{\ey}{\vy + \m*(5 cm -\vx)} % Work out the y-height at the other end of the line \pgfmathsetmacro{\by}{\vy - \m*(\vx+1 cm)} \draw[thick,\col] (-1,\by pt) -- (5,\ey pt); } \end{tikzpicture} \end{document} 
pgfplotstag, do you have anything particular in mind that concerns drawing things inpgfplots. There are some special things inside itsaxisenvironments, like theaxis csandrel axis cscoordinate systems that might be helpful for you, so maybe you could edit your question to give more details (or remove thepgfplotstag if you actually meant TikZ in general).x1,x2,y1, andy2. If you just want to vary them explicitly, then the\newcommandapproach is as good as any. If they are coordinates, then the\coordinateapproach is best. But if they are to be computed from something else, then something a little more complicated (but not much) is needed; however, exactly what depends on how you want to specify them.tkz-euclidethat he refers to in an answer to another question could be helpful for you? There are quite a few examples on his website.