awk calculations giving rounded answers for larger numbers

Question

I have a Linux bash script which is calculating the total of some values in a file using awk. I am using this syntax to do the calculations:

$ awk "BEGIN { print $first_number + $second_number }"

For calculations where the answer is a number below 9999.99, it displays the answer correctly. Above that value, it rounds the answer to one decimal place, or no decimal places, etc. like this:

$ awk "BEGIN { print 89.54 + 6789.44 }"
6878.98

$ awk "BEGIN {print 1.66 + 68790.77 }"
68792.4

$ awk "BEGIN {print 1.66 + 687990.77}"
687992

$ awk "BEGIN {print 10100.66 + 6787990.77}"
6.79809e+06

I want the answer to show the decimals, and to 2 decimal places in all instances.

DO you know why it is doing the unexpected roundings? Thanks

Answer 1

Expanding on top of James's useful answer, there is a reason why you see that behavior. It is because the exact manner in which numbers are converted into strings is controlled by a predefined variable in awk called OFMT when using print

OFMT's default value is "%.6g", which creates a value with at most six significant digits. If you want print to represent a more accurate representation, change it for e.g. to always represent 2 digits post the decimal point, something like "%2.2f"

$ awk 'BEGIN { OFMT = "%2.2f"; print 89.54 + 6789.44 }'
6878.98

$ awk 'BEGIN { OFMT = "%2.2f"; print 1.66 + 68790.77 }'
68792.4

$ awk 'BEGIN { OFMT = "%2.2f"; print 1.66 + 687990.77 }'
687992.43

$ awk 'BEGIN { OFMT = "%2.2f"; print 10100.66 + 6787990.77 }'
6798091.43

See here and here, which explains the behavior of these conversions.