Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site mit-vax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!unc!mcnc!philabs!cmcl2!seismo!harvard!think!mit-eddie!mit-vax!csdf
From: csdf@mit-vax.UUCP (Charles Forsythe)
Newsgroups: net.graphics,net.math
Subject: Re: Mandelbrot set problem
Message-ID: <1110@mit-vax.UUCP>
Date: Sat, 9-Nov-85 16:47:50 EST
Article-I.D.: mit-vax.1110
Posted: Sat Nov  9 16:47:50 1985
Date-Received: Mon, 11-Nov-85 07:00:44 EST
References: <2346@flame.warwick.UUCP>
Reply-To: csdf@mit-vax.UUCP (Charles Forsythe)
Organization: MIT, Cambridge, MA
Lines: 21
Xref: watmath net.graphics:1259 net.math:2506

In article <2346@flame.warwick.UUCP> kay@flame.UUCP (Kay Dekker) writes:
>How should one choose the maximum number of iterations for any particular
>region of the complex plane?  Sure, one can always pick a value that's
>"big enough" for any particular view ("big enough" meaning, intuitively,
>that not too many pixels are marked as lying within the set when in fact
>they aren't), but having such a large value makes the browser far too slow
>when that many iterations aren't needed.

The "browsers" that I've seen choose a maximum number of 1000, and I 
think it's arbitrary. The programs also check, on each iteration, 
whether or not |z|>2. It seems to me, you could increase the maximum 
iterations to any number. That way, you could intelligently evaluate 
points that require more than 1000 iterations, while also including 
those that require less. The only problem is that you increase your wait
-- I think that 1000 was picked because they figured that was "long enough."
 


-- 
-Charles