This is an R modification by Thomas Lumley
of S code by Mike Meyer
that calls an adaptive quadrature routine by Alan Genz.
Original  http://lib.stat.cmu.edu/S/adapt  [dated Oct 17 1989]
-------
June 2001

Removed one-dimensional integration: R 1.3.0 has a quadpack-based function
in base.


-------
June 1999

Deleted Makefile -- R now does better without one.

Redid function passing to use eval() rather than call_R.

Added option to keep doubling static storage allocation until precision
achieved or R crashes :)

Thomas Lumley
----				   

January 1999:

The Fortran code works without problems
on Solaris 2.5.1 [f77 SC4.2; g77]  and	Linux (Debian 2.0).
There's no reason to keep the f2c-translated C code.

adapt() and hence integrate() now return a class "integrate" object
having a simple print method.

Martin Maechler, ETH Zurich

----

July 1997:

I recommend using the C code (converted with f2c) as I'm not sure about
the correct linking for the FORTRAN.  If you use g77 to compile the
FORTRAN you must specify the -O flag, as otherwise bsrl.f activates a
known bug in g77 causing it to generate invalid assembly language.

In any case, make sure that the test examples work.

Thomas Lumley
thomas@biostat.washington.edu