filename fylfat "d:/txt/txtbig.txt" lrecl=16000 recfm=v;
* create the input fat file;
data sasfat;
informat state $2. num1-num720 best21.;
infile fylfat end=dne;
input state $ num1-num720 ;
output;
if _n_=3 then stop;
run;quit;
proc print data=sasfat;
format num1-num720 21.17;
run;quit;
Norte SAS cannot handle 18 significat digits(you need R or Python for 18 sig digit arithmetic)
1.0021709654943634 Original number
1.0021709654943600 SAs number (lost precistion)
Obs STATE NUM1 NUM2 ... NUM720
1 AK 1.00217096549436000 1.00414185896059000 ... 2.26668456775932000
2 AL 1.00203845526359000 1.00408994140026000 ... 2.70441496537969000
3 AR 1.00187083703867000 1.00367877037134000 ... 2.27810994554249000
Middle Observation(1 ) of sasfat - Total Obs 3
-- CHARACTER --
STATE C 2 AK STATE
-- NUMERIC --
NUM1 N 8 1.0021709655 NUM1
NUM2 N 8 1.004141859 NUM2
NUM3 N 8 1.0059412003 NUM3
NUM4 N 8 1.0075022638 NUM4
NUM5 N 8 1.0089031672 NUM5
NUM6 N 8 1.0103131154 NUM6
...
NUM717 N 8 2.2575557326 NUM717
NUM718 N 8 2.2610511943 NUM718
NUM719 N 8 2.2640534503 NUM719
NUM720 N 8 2.2666845678 NUM720
If you want to maintain precision you van use python (R does not have a native 64 bit integers)
Drop the decimal until last calc then add it with a format;
* there are extended precision packages in R and python;
* did not add the code to place the decimal you need a format;
%utl_submit_py64('
import numpy as np;
big1=np.int64(10021709654943634);
print(big1);
big1=np.int64(80000000000000008/4 );
print(big1);
');
1.0021709654943634
2.0000000000000002
