Odd results with basic math

Support, Help & Resources
1

Hi, I’m very new to Sonic Pi and loving it.

However, I’ve come across some unusual behaviour with basic math which has me stumped:

Firstly, whenever I try to divide a small number with a larger number, which should yield a result lower than 1, I get 0 - for example:

puts 1/2 = 0

but

puts 2/1 = 2

That’s the first. Secondly, and odd thing occurs when a subtraction gives a result close to 0, from around 0.3 - for example:

puts 1-0.8 = 0.19999…6
puts 2-1.8 = 0.19999…6

but

puts 0+0.2 = 0.2

Does anyone else find this or is it a quirk of my system?

Many thanks, Paul.

2

Try 1 / 2.0 as opposed to 1 / 2
If integers are used in calculations ,
then results are integers .
A ’ float ’ ( decimal ) in the equation
will give decimal results …

3

Awesome, that sorted out the division issue, huge thanks!

However, I’m still getting the strange recurring decimals, even with 1.0 - 0.8 - however, this isn’t causing any problems for me at the moment, and judging by how close they are to the true value, it may never be an issue.

4

Hi there, what’s you’re running into here is caused by the fact that floating point numbers are inaccurate by design. This is due to them occupying a limited size in memory, so they can’t represent every possible decimal number and therefore have limited precision.

For more background Google for “issues with floating point numbers” which will return a number of articles explaining this situation in varying detail. For example: Losing My Precision: Tips For Handling Tricky Floating Point Arithmetic | SOA

5

Thanks for the advice and link, this was not something I was aware of before.

I’m really enjoying using Sonic Pi and have signed up on Patreon. Keep up the good work!

Paul.

1 Like
6

As @samaaron explains, all computer floating point numbers have a finite precision. The default Float is not decimal— it is binary— and has a precision of 53 bits (nearly 16 decimal digits). That should really be enough for our purposes, but it is still possible to manipulate them incorrectly in a calculation, causing loss of precision, or make invalid comparisons like

a = 0.15 + 0.15
b = 0.1 + 0.2
if a==b
  puts "yea"
else
  puts "nope"
end

It is not a problem with the numbers; you just have to use them correctly.

If you really need decimal numbers instead of binary, you can do something like

require 'bigdecimal'
puts BigDecimal("1")-BigDecimal("0.8")

but you probably don’t. Try something (depending on what you are trying to do!!!) like

a = 1 - 0.8
b = 2 - 1.8
puts (a-b).abs < 0.00001

That will not make any computer scientists happy, but it’s OK since your music should usually not care about 0.00001 beats or anything below a few milliseconds, humans cannot perceive such intervals.

1 Like