[Thread Prev][Thread Next][Index]

Re: [ferret_users] question about @din or @iin



Hi Dawn,
Ferret does not shift half a grid cell for this calculation.  It computes the definite integral as the discrete sum, F(x)* dx where dx is computed from half a grid cell below to half a grid cell above the coordinate point.  This particular example is

1*1(1.5 - 0.5) + 2*2(2.5 - 1.5) + 3*3(3.5-2.5) = 14.0

When Ferret writes that x is from 0.5 to 3.5 this indicates that the sum includes the grid cells from the start of box 1 at 0.5 to the end of box 3 at 3.5 was used in the sum.

Take another example, where the boxes are not of size 1.

yes? define axis/x=1:5:2 xax
yes? let xx = x[gx=xax]
yes? let ff = xx* xx

yes? list ff[x=@iin]
            VARIABLE : XX*XX
                        indef. integ. on X
             SUBSET   : 3 points (X)
 1   / 1:   2.00
 3   / 2:  20.00
 5   / 3:  70.00

yes? list ff[x=@din]
             VARIABLE : XX*XX
             X        : 0 to 6 (integrated)
          70.00

 Here are the box edges and coordinates
yes? list xboxlo[gx=xax], x[gx=xax], xboxhi[gx=xax]
             X: 0 to 6
 Column  1: XBOXLO is XBOXLO (axis XAX)
 Column  2: X is X (axis XAX)
 Column  3: XBOXHI is XBOXHI (axis XAX)
        XBOXLO     X   XBOXHI
1   / 1:  0.000  1.000  2.000
3   / 2:  2.000  3.000  4.000
5   / 3:  4.000  5.000  6.000


So, ff[x=@din]  =  1*1*(2 - 0) + 3*3*(4 - 2) + 5*5*(6 - 4) = 70

Or, if we had said, list ff[x=1:5@din], then Ferret would use half a grid box on the first and last coordinates, or

ff[x=1:5@din]  =  1*1*(2 - 1) + 3*3*(4 - 2) + 5*5*(5 - 4) = 44
yes? list ff[x=@din]
             VARIABLE : XX*XX
             X        : 1 to 5 (integrated)
          44.00



ferret ocean wrote:

hi, all
   I'm not sure if anybody had ask this question, and I didn't find it from Email archive in PMEL website. My question is about @din, which give us an unaccurate value?

let one=x*x
list/i=1:3 one[i=@din]
then Ferret give us:

             VARIABLE : X*X
             X        : 0.5 to 3.5 (integrated)
          14.00

Actually, we know integrate x*x from 0.5 to 3.5 (shift 1/2 grid cell in Ferret), is x^3/3 from 0.5 to 3.5, which means (3.5^3-0.5^3)/3=14.25 instead of 14.00.


why? the same for @iin, the results of each step is integer, no digit. why?

Any help will be appreciated!


Dawn


__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com


[Thread Prev][Thread Next][Index]

Dept of Commerce / NOAA / OAR / PMEL / TMAP

Contact Us | Privacy Policy | Disclaimer | Accessibility Statement