# Command line calculator that keeps fractional values

I just found a good command line calculator program called bc and was satisfied with it until I discovered it rounds off fractional values, thus causing loss in precision.

According to its man page:

All numbers are represented internally in decimal and all computation is done in decimal. (This version truncates results from divide and multiply operations.)

Could you please suggest an equivalent of bc for Ubuntu Maverick? I need to make advanced command line calculations with variables.

You can set the length of the fractional part with scale=n.

The command echo 'scale=20;752/447' | bc yields:

1.68232662192393736017

Note that even if the number fits within the scale, additional zero's might be appended:

scale=20
1/2
.50000000000000000000

Unfortunately, there is always a rounding issue:

scale=50
1/3*3
.99999999999999999999999999999999999999999999999999
• could the scale value be set as global or I should always set it explicitly whet start bc? Apr 17 '11 at 13:45
• @sergionni: You should always pass the scale variable to bc. The only environment variable that might be useful is BC_ENV_ARGS. This variable puts some arguments to bc, you could create a file with standard variables or settings, and set BC_ENV_ARGS=/path/to/variables/file. Please read the manual page. Apr 17 '11 at 13:57
• I tend to start bc via bc -ql. -q hides the copyright banner (which generally proves distracting) and -l loads the math library and automatically sets the scale to 20. (See man bc.) Apr 11 '19 at 1:47

calc (I believe from package apcalc) does the same as bc, but does not round. It displays similarly to bc, but unlike bc, it understands scientific notation. Example:

> calc
C-style arbitrary precision calculator (version 2.12.3.3)
[Type "exit" to exit, or "help" for help.]

; a=234
; b=a/7
; b
~33.42857142857142857143
; c=b/1e20
; c
~0.00000000000000000033
; c*1e10
~0.00000000334285714286
;

Compare with bc:

> bc -l
bc 1.06.95
Copyright 1991-1994, 1997, 1998, 2000, 2004, 2006 Free Software Foundation, Inc.
This is free software with ABSOLUTELY NO WARRANTY.
For details type `warranty'.
a=234
b=a/7
b
33.42857142857142857142
c=b/10^20
c
.00000000000000000033
c*1e10
(standard_in) 6: syntax error
c*10^10
.00000000330000000000

A little search turns up lots of results, not all of which are relevant, but I am sure that a few trials will get you exactly what you want (wcalc, for instance):

aptitude search calc
i   apcalc                               - Arbitrary precision calculator (original name: calc)
i A apcalc-common                        - Arbitrary precision calculator (common files)
p   apcalc-dev                           - Library for arbitrary precision arithmetic
p   bandwidthcalc                        - file transfer time calculator written in GTK+
p   calcoo                               - Scientific calculator (GTK+)
p   calcurse                             - text-based calendar and todo manager
p   concalc                              - console calculator
p   extcalc                              - multifunctional scientific graphic calculator
p   gcalcli                              - Google Calendar Command Line Interface
i   gcalctool                            - GNOME desktop calculator
p   ipcalc                               - parameter calculator for IPv4 addresses
p   ipv6calc                             - small utility for manipulating IPv6 addresses
p   kcalc                                - calculator for KDE 4
p   keurocalc                            - universal currency converter and calculator - binary package
p   keurocalc-data                       - universal currency converter and calculator - data package
p   lcalc                                - a program for calculating with L-functions
p   libcolor-calc-perl                   - Perl module for simple calculations with RGB colors
p   libdate-calc-perl                    - Perl library for accessing dates
p   libdate-pcalc-perl                   - Perl module for Gregorian calendar date calculations
p   libmath-basecalc-perl                - Convert numbers between various bases
p   libmath-calc-units-perl              - Human-readable unit-aware calculator
p   libmath-calculus-differentiate-perl  - Algebraic Differentiation Engine
p   libmath-calculus-expression-perl     - Algebraic Calculus Tools Expression Class
p   libmath-calculus-newtonraphson-perl  - Algebraic Newton Raphson Implementation
p   libticalcs-dev                       - Texas Instruments calculator communication library [development files]
p   libticalcs2-7                        - Texas Instruments calculator communication library
p   octave-physicalconstants             - provide physical constants values in Octave
i   openoffice.org-calc                  - office productivity suite -- spreadsheet
v   openoffice.org2-calc                 -
p   python-ipcalc                        - perform IP subnet calculations
v   python2.6-ipcalc                     -
p   r-cran-epicalc                       - GNU R Epidemiological calculator
p   rpncalc                              - RPN calculator trying to emulate an HP28S
p   science-numericalcomputation         - Debian Science Numerical Computation packages
p   sipcalc                              - Advanced console-based ip subnet calculator
p   subnetcalc                           - IPv4/IPv6 Subnet Calculator
p   sugar-calculate-activity             - calculate activity for the Sugar graphical shell
p   tapecalc                             - a full-screen tape editor that lets the user edit a calculation
p   transcalc                            - microwave and RF transmission line calculator
p   wcalc                                - A flexible command-line scientific calculator
p   wmcalclock                           - A dock.app which simply tells time and date
p   xsmc-calc                            - Smith Chart calculator for X

I would suggest using Python as a command-line calculator:

\$ python
>>> from math import *
>>> help(sin)
sin(x)

Return the sine of x (measured in radians).

Also I would recommend IPython or IDLE. Both hugely improve usability of the standard shell.

Update: use python3 to avoid truncation surprises:

\$ python2.7

>>> 10/3
3

\$ python3

>>> 10/3
3.3333333333333335
• Best calculator ever. You could do anything. Apr 17 '11 at 16:20
• That truncates calculations too. Jul 29 '11 at 9:56
• It won't truncate anything if you divide e.g. like this: 2.0 / 100. Of course 2 / 100 would equal to 0, because it's an integer division. Apr 1 '15 at 18:06

You lost precision in this sense: if set precision to 10 decimal digits, divisions are truncated to 10 decimal digits, and this is a coherent choice.

If you look for an exact calculator, you need a symbolic system as maxima.

By the way, bc supports variables.

• yes, I know it support variables,that's I'm asking for analog Apr 17 '11 at 13:43

"genius" is the most advanced calculator out there, with both command-line and GUI options available. Check the manual for details, and see http://www.jirka.org/genius.html .

To install, just type:

sudo apt-get install genius gnome-genius

Here is a good one:

spigot -- a command-line exact real calculator

If you have octave installed you can use it at the command line as:

octave --silent --eval 752/447

To shorten the writing you can add the following as alias in .bashrc

alias ose='octave --silent --eval'

and then call it as ose 752/447. The alias/shortcut is arbitrary but you need to restart the terminal to make it effective.

You can install octave using:

sudo apt-get install octave

Of course, with octave you can use all the advanced functions available in it too.

I just made a simple one:

https://github.com/jb55/ratio-cli

\$ ratio <<< '14/15 * 3'
14/5
• does it support variables? Oct 10 '17 at 3:40
• it's dead simple right now, just basic expressions. maybe one day? Oct 10 '17 at 3:47
• the OP specified that it needs t handle variables Oct 10 '17 at 3:55