home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!opl.com!hri.com!spool.mu.edu!sdd.hp.com!cs.utexas.edu!swrinde!emory!sol.ctr.columbia.edu!The-Star.honeywell.com!umn.edu!lynx.unm.edu!zia.aoc.nrao.edu!laphroaig!cflatter
- From: cflatter@nrao.edu (Chris Flatters)
- Newsgroups: comp.lang.fortran
- Subject: Re: Wanted: folk theorems in Fortran Progra
- Message-ID: <1993Jan27.203653.23536@zia.aoc.nrao.edu>
- Date: 27 Jan 93 20:36:53 GMT
- References: <93-01-199@comp.compilers>
- Sender: news@zia.aoc.nrao.edu
- Reply-To: cflatter@nrao.edu
- Organization: NRAO
- Lines: 29
-
- In article 199@comp.compilers, dodd@mycenae.cchem.berkeley.edu (Lawrence R. Dodd) writes:
- >"Steve" Stevenson <steve@hubcap.clemson.edu> writes:
- >> I am trying to make a catalog of folk wisdom in fortran programming.
- >
- >My personal favorite is the one discussed on page 137 of the 1st edition of
- >Numerical Recipes (Fortran edition, of course):
- >
- >BAD
- >
- >P = c(1) + c(2)*x + c(3)*x**2 + c(4)*x**3 + c(5)*x**4
- >
- >GOOD
- >
- >P = c(1) + x*(c(2) + x*(c(3) + x*(c(4) + x*c(5))))
- >
- >It is a little harder to read but you won't be `summarily executed'
-
- I hope that you don't mean folk wisdom in the sense of things that are
- widely believed without justification: there is a good reason for
- preferring the latter version. The former version requires 4
- additions, 4 multipications and 3 applications of the power operator
- (should total 15 or 16 fp operations using an efficient algorithm for
- raising fp numbers to integer powers. The latter version does not
- require application of the power operator. Not only will it be
- approximately twice as fast but it will accumulate less rounding
- error.
-
- Chris Flatters
- cflatter@nrao.edu
-