Description
The gams function JDate(year,month,day) accepts arguments similar to to date functions in other languages. The earliest date is Jan 1, 1, and the latest date is Dec 31, 9999. Gams picked Jan 1, 1900 as day 1, as done by Excel. Unfortunately, Excel treats the year 1900 incorrectly as a leap year and the serial days starting with March 1, 1900 will be off by one day. The arguments are can be positive or negative and do not have to form a proper date. The arguments with fractional values are first converted to a signed integer by using the floor() function. Note about Excle: The Excel DATE(yesr,month,day) function has the following data ranges note that excel incorrectly treats the year 1900 as a leap year. DATE(1900, 1, 1) = 1 incorrect should be 2 DATE(1900, 2,28) = 59 incorrect should be 60 DATE(1900, 3, 1) = 61 DATE(9999,12,31) = 2958465 for years having a value of less than 1900, 1900 + year is assumed The following table shows different ways of entering dates:
Small Model of Type : GAMS
Category : GAMS Test library
Main file : dattim4.gms
$title Julian date test (DATTIM4,SEQ=120)
* The gams function JDate(year,month,day) accepts arguments similar to
* to date functions in other languages. The earliest date is Jan 1, 1, and
* the latest date is Dec 31, 9999. Gams picked Jan 1, 1900 as day 1,
* as done by Excel. Unfortunately, Excel treats the year 1900 incorrectly
* as a leap year and the serial days starting with March 1, 1900 will
* be off by one day.
*
* The arguments are can be positive or negative and do not have to form
* a proper date. The arguments with fractional values are first converted
* to a signed integer by using the floor() function.
*
* Note about Excle:
* The Excel DATE(yesr,month,day) function has the following data ranges
* note that excel incorrectly treats the year 1900 as a leap year.
* DATE(1900, 1, 1) = 1 incorrect should be 2
* DATE(1900, 2,28) = 59 incorrect should be 60
* DATE(1900, 3, 1) = 61
* DATE(9999,12,31) = 2958465
* for years having a value of less than 1900, 1900 + year is assumed
*
* The following table shows different ways of entering dates:
$eolcom //
set i / 1*22/; acronym error
table tvals(i,*) test values
year month day excel correct
1 2005 10 20 38645 38644
2 2005.3 10.3 20.3 38645 38644
3 2005.7 10.7 20.7 38645 38644
4 2005 22 20 39010 39009
5 2005 22 300 39290 39289
6 2005 -5 300 38468 38467
7 2005 -5 -10 38158 38157
8 2005 -5 -10.3 38157 38156
9 2005 -5 -10.7 38157 38156
10 2 -9 -424 0 error
11 1900 1 1 1 1
12 0 0 1 error error
13 2004 2 30 38047 38046
14 2004 14 30 38413 38412
15 2004 0 0 37955 37954
16 9999 12 31 2958465 2958464
17 9999 13 1 error error
18 1900 1 2 2 2
19 1899 12 31 693962
20 1899 12 30 693961 -1
21 1900 2 1 32 32
22 1 1 1 367 -693594
parameter res(i) gams julian serial numbers;
res(i) = jdate(tvals(i,'year'),tvals(i,'month'),tvals(i,'day'));
tvals(i,'gams') = res(i); display tvals;
* convert acronym to UNDF
tvals(i,'correct')$(tvals(i,'correct') = error) = 1/0;
execerror = 0;
set err1(i) gams values are incorrect
err2(i) recoverd values are different;
err1(i) = tvals(i,'correct') <> res(i); display err1;
abort$card(err1) ' incorrect date serial numbers';
* Now we will recover the date components
parameter back(i,*) recovered date components;
back(i,'gams') = res(i);
back(i,'year') = gyear(res(i));
back(i,'month') = gmonth(res(i));
back(i,'day') = gday(res(i));
back(i,'new') = jdate(back(i,'year'),back(i,'month'),back(i,'day'));
execerror = 0;
display back;
err2(i) = res(i) <> back(i,'new'); display err2;
abort$card(err2) 'back operations were incorrect';