NAG and IMSL have sold numerical libraries for Fortran
for many years. I think the NAG library is more
comprehensive. Looking at GARCH models (a statistical
time series model used in finance), IMSL has a
procedure only for the symmetric GARCH(p,q) model,
whereas NAG has procedures for both symmetric and
asymmetric garch models and EGARCH
http://www.nag.com/numeric/fl/manual/html/G13/g13_conts.html
.
Unfortunately, the NAG GARCH routines do not exploit
any of the post-Fortran-77 features in their
interface, such as optional arguments or assumed-shape
arrays. Therefore, the interface for the simplest
GARCH routine is
SUBROUTINE G13FAF(DIST, YT, X, LDX, NUM, IP, IQ, NREG,
MN, ISYM, NPAR,THETA, SE, SC, COVAR, LDC, HP, ET, HT,
LGF, COPTS, MAXIT, TOL, WORK, LWORK, IFAIL)
For IMSL, the call can be as simple as
CALL GARCH(W, NP, NQ, XGUESS, X)
IMSL took the trouble to provide a modern Fortran 90
interface for its entire library, while retaining the
Fortran 77 interface for backwards compatibility.
About 15 years after the Fortran 90 standard, NAG
cannot be bothered to create SIMPLE interfaces even
for NEW procedures -- GARCH algorithms are recent
additions to its library. This is especially
disappointing from a company that is a leader in
adding features from the lastest Fortran standard to
its compiler.
I hope that NAG will try to make its libraries easier
to use. Otherwise it deserves to lose business to
companies that do.
____________________________________________________________________________________
Need Mail bonding?
Go to the Yahoo! Mail Q&A for great tips from Yahoo! Answers users.
http://answers.yahoo.com/dir/?link=list&sid=396546091
|