This is a consequence of the fact that Stata
works in binary, so that not all decimals can
be held exactly. 0.1 is a case in point.
The heart of the problem is shown by turning
the magnifying glass on the first puzzling result:
. forval i = 0(0.1)20 {
2. di %21.18f round(`i', 0.01)
3. }
0.000000000000000000
0.100000000000000010
0.200000000000000010
0.299999999999999990
0.400000000000000020
0.500000000000000000
0.599999999999999980
0.700000000000000070
0.800000000000000040
0.900000000000000020
.
.
.
19.400000000000002000
19.500000000000000000
19.600000000000001000
19.699999999999999000
19.800000000000001000
19.900000000000002000
You are asking Stata to add 0.1 each time, but Stata
has do the equivalent in _binary_, and with even all the bits
in the world it would usually have to approximate. In this case
another 0.1 would take beyond 20, so it stops there.
Incrementing by 1/8 is not problematic given that 1/8 = 1/2^3.
Read [U] 16.10 every night after brushing your teeth...
For another example, see
http://www.stata.com/support/faqs/data/mod.html
Nick
[email protected]
[email protected]
I am writing a program with a -forvalues- loop in it. I encountered a
strange problem (Stata 8.2) which can be illustrated by this sample code:
forvalues z = 15(1)30 {
di "current maximum is: `z'"
forvalues i = 0(0.1)`=`z'' {
local a = round(`i',0.01)
}
di "a is: `a'"
}
Output is
current maximum is: 15
a is: 15
current maximum is: 16
a is: 16
current maximum is: 17
a is: 17
current maximum is: 18
a is: 18
current maximum is: 19
a is: 19
current maximum is: 20
a is: 19.9
current maximum is: 21
a is: 20.9
current maximum is: 22
a is: 21.9
current maximum is: 23
a is: 22.9
current maximum is: 24
a is: 23.9
current maximum is: 25
a is: 24.9
current maximum is: 26
a is: 25.9
current maximum is: 27
a is: 26.9
current maximum is: 28
a is: 27.9
current maximum is: 29
a is: 28.9
current maximum is: 30
a is: 29.9
Stata stops looping at the second last value in the numlist
from 20 onward. This behaviour continues until 76, from where on
the correct value is returned again. This seems to happen only with
decimal numbers, and only with some of them
(compare to -forvalues i = 0(0.125)`=`z'' -). I can't really see a
pattern. Maybe someone can give me a clue and tell me what to do about it.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
