[fitswcs] Polynomial Correction Function & Alternative-WCS
Don Wells
dwells at NRAO.EDU
Wed May 6 17:10:49 EDT 1998
Don Wells writes:
> .. The value of n which includes all terms with (i+j)<=q is obtained by
> setting i=0, j=q in the formula for n given above:
>
> n_q = axis*1000 + ((q + 1) * q) / 2 + q)
> .. The capability for 999 terms in
> the polynomial will support terms up to q=(i+j)=21 (X^21, X^20Y,
> X^19Y^2, .., XY^20, Y^21) for the NAXIS=2 case..
I made a mistake; the NAXIS=2 limit is not q=21 but rather q=43:
CP1985 = <E-format-value> / coeff for X^4*Y^39 for axis 1
CP1986 = <E-format-value> / coeff for X^3*Y^40 for axis 1
CP1987 = <E-format-value> / coeff for X^2*Y^41 for axis 1
CP1988 = <E-format-value> / coeff for X*Y^42 for axis 1
CP1989 = <E-format-value> / coeff for Y^43 for axis 1
> .. The author has not yet derived the general formula for n and its
> inversion formulae for NAXIS>2. Somebody should do this, and should
> determine the q=(i+j) order limit ..
Although I do not yet have a formula for n for NAXIS=3, I do think
that I now have one for n_q:
n_q(3) = axis*1000 + ((q + 3) * (q + 2) * (q + 1)) / (3 * 2 * 1)) - 1
I observe that n_q(2) can also be derived by setting i=(q+1),j=0:
n_q(2) = axis*1000 + ((q + 2) * (q + 1)) / (2 * 1) - 1
(which also gives n_43(2)=989) and that the formula for NAXIS=1 can be
re-written as:
n_q(1) = axis*1000 + (q + 1) / (1) - 1
Comparison of these 3 n_q formulae implies that the probable general
formula for NAXIS>3 is:
n_q(N) = axis*1000 + (q + N)! / (q! * N!) - 1
We have n_16(3)=968, so the 999-coefficient capability will support
terms X^iY^jZ^k up to (i+j+k)=16 for NAXIS=3. If the general formula
given above is correct, we have n_10(4)=1000 [sic!], i.e., we can have
terms up to 10th-order in each axis for NAXIS=4.
-Don
--
Donald C. Wells Associate Scientist dwells at nrao.edu
http://www.cv.nrao.edu/~dwells
National Radio Astronomy Observatory +1-804-296-0277
520 Edgemont Road, Charlottesville, Virginia 22903-2475 USA
More information about the fitswcs
mailing list