I would like to convert a number that is stored in scientific notation into a floating point decimal, so I can then perform some comparisons on the data. This is being done in a bash script - here is a small snippet of the code:
while read track_id landfall_num gate_id pres_inter
do
if [[ $landfall_num == 0001 ]]
then
start_flag = true
echo DING DING $start_flag
if [[ $pres_inter < 97000 ]]
then
echo Strong Storm From North $track_id, $gate_id, $pres_inter
fi
fi
done < $file
My problem is that my <
operand is selecting basically all of the pressure values, which are stored in scientific notation, when I use <
, and none when I use >
. I am looking at atmospheric pressure measurements in pascals rather than millibars.
Here is sample output:
Strong Storm From North 0039988 0017 1.0074E+05
Strong Storm From North 0037481 0018 9.9831E+04
Neither of these storms should be meeting the selection criteria!
First thing, bash cannot do arithmetic using floating point arithmetic. Second thing, bash doesn't know scientific notation (even for integers).
First thing you can try, if you're absolutely positively sure that all your numbers are integers: convert them to decimal notation: printf
will happily do that for you:
printf -v pres_inter "%.f" "$pres_inter"
(%.f
rounds to nearest integer).
then use bash's arithmetic:
if (( pres_inter < 97000 )); then ....
This solution is wonderful and doesn't use any external commands or any subshells: speed and efficiency!
Now if you're dealing with non integers you'd better use bc
to do the arithmetic: but since bc
is retarded and doesn't deal well with the scientific notation, you have to convert your numbers to decimal notation anyways. So something like the following should do:
printf -v pres_inter "%f" "$pres_inter"
if (( $(bc -l <<< "$pres_inter<97000") )); then ...
Of course, this is slower since you're forking a bc
. If you're happy with integers, just stick to the first possibility I gave.
Done!
For numeric comparisons, you need to use:
-eq
instead of ==
-ne
instead of !=
-lt
instead of <
-le
instead of <=
-gt
instead of >
-ge
instead of >=
And to fix some syntax issues:
while read track_id landfall_num gate_id pres_inter
do
landfall_num=$(printf "%f", "$landfall_num")
if [[ "$landfall_num" -eq 1 ]]
then
start_flag="true"
echo "DING DING $start_flag"
if [[ "$pres_inter" -lt 97000 ]]
then
echo "Strong Storm From North $track_id, $gate_id, $pres_inter"
fi
fi
done < "$file"
Updated:
This is how you convert a scientific notation number into an integer:
printf "%.0f\n", 1.0062E+05
for example, prints:
100620
And so:
landfall_num=$(printf "%f", "$landfall_num")
will convert landfall_num
from scientific notation into a normal decimal integer.
I think the best way to do it is using awk. For example, in the case of the number "1.0074E+05"
echo "1.0074E+05" | awk -F"E" 'BEGIN{OFMT="%10.10f"} {print $1 * (10 ^ $2)}'
the output:
100740.0000000000
obviously you can use less precision then 10 decimal :)