Decimal places in fixed point arithmetic with negative numbers

Question

I have the code:

\documentclass[tikz,border=5]{article}
\usepackage{tikz}
\usepackage{xfp}
\usepackage{amsmath}
\usetikzlibrary{math}
\usepackage{fp}
\usetikzlibrary{fixedpointarithmetic}
\begin{document}
\tikzset{fixed point arithmetic}
\tikzmath{
    real \x; real \y;
    %
    \x=-1.3754;
    \y = \fpeval{round(\x*5.56,3)};
}
\tikzset{fixed point arithmetic}
\tikzmath{
    real \z; real \w;
    %
    \z=1.3754;
    \w = \fpeval{round(\z*5.56,3)};
}
\vspace{3cm}
O valor de $x$ é: \fpeval{round(\x,3)} \ 
O valor de $y$ é: \y \
O valor de $z$ é: \fpeval{round(\z,3)} \ 
O valor de $w$ é: \w \
\end{document}

whose output is the following:

When I operate negative numbers in tikzmath, the decimal places of the result is incontrolable. However when the calculation deal with positive real numbers, I set the decimal places with \fpeval{round(x,3)} (to three decimal places) and I can to control the precision. My question is how can I fix this? Why fixed point arithmetic is noisy with negative numbers? I would to set the decimal places in the tikzmath code space, not in the text. Many thanks!

Replace last line with O valor de $y$ é: \fpeval{round(\y,3)} works. No idea why — user202729, Dec 15 '21 at 00:55
That having said, coming from other programming languages, use round() to print number to X decimal places feel absolutely like the wrong way. — user202729, Dec 15 '21 at 00:55
I want to control the decimal places in the tikzmath code, not in the last LINE. — Angelo Aliano Filho, Dec 15 '21 at 00:59
Then you already controlled it. -7.647 == -7.6470000000000. What you want is to print the number to 3 decimal places. — user202729, Dec 15 '21 at 01:01
https://tex.stackexchange.com/questions/350446/print-3-decimals-in-tikz-without-using-standard-form suggests pgfmathprintnumber, which works in this case. — user202729, Dec 15 '21 at 01:02
Not controlled in the tikzmath mode when the number is negative. When Is positive the command round works. Try! — Angelo Aliano Filho, Dec 15 '21 at 01:05
It does work in that the resulting value is correct. I don't think there's anything in the documentation that states that the result is guaranteed to be stored in the minimal form. — user202729, Dec 15 '21 at 01:06
Alternatively siunitx can be used too. https://tex.stackexchange.com/questions/599426/xfp-zerofill-at-fpeval — user202729, Dec 15 '21 at 01:21
I think you're misunderstanding something. Either you only care only about the value of the numbers; in that case the produced value and the expected value is exactly equal; or you care about how it looks like in text, then use the text formatting commands. — user202729, Dec 15 '21 at 01:33
I am concerned about the appearance and value of the result. The only thing I don't understand is because Tikz puts too many zeros when the calculation involves negative numbers and I can't get those zeros inside the tikzmath environment — Angelo Aliano Filho, Dec 15 '21 at 01:37
If you are concerned about the appearance then use the text formatting command? ■ While the internal storage of the value is inconsistent; it's not a bug because the value is correct; nevertheless just wait a bit and see if the TikZ developers are interested in changing the behavior (perhaps some efficiency improvement or something). — user202729, Dec 15 '21 at 01:39

score 1 · Answer 1 · answered Dec 18 '21 at 21:18

Use the numprint package. the "autolanguage" option adjusts the unit and thousand separators according to the babel package option, in this case Brazilian Portuguese.

\documentclass[tikz,border=5]{article}
\usepackage[brazil]{babel}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{tikz}
\usepackage{xfp}
\usepackage{amsmath}
\usetikzlibrary{math}
\usepackage{fp}
\usetikzlibrary{fixedpointarithmetic}
\usepackage[autolanguage]{numprint}
\begin{document}
\tikzset{fixed point arithmetic}
\tikzmath{
    real \x; real \y;
    %
    \x=-1.3754;
    \y = \fpeval{round(\x*5.56,3)};
}
\tikzset{fixed point arithmetic}
\tikzmath{
    real \z; real \w;
    %
    \z=1.3754;
    \w = \fpeval{round(\z*5.56,3)};
}
\vspace{3cm}
\nprounddigits{3}
O valor de $x$ é: \numprint{\x}  \
O valor de $y$ é: \numprint{\y} \
\nprounddigits{4}
O valor de $z$ é: \numprint{\z} \
O valor de $w$ é: \numprint{\w} \
\end{document}

Decimal places in fixed point arithmetic with negative numbers

1 Answers1