Re: Salamin-Brent algorithm

On Jun 22, 11:29 am, James Waldby <n...@xxxxx> wrote:
dgoldsmith_89 wrote:
On Jun 22, 12:26 am, Gerry Myerson ... wrote:...
dgoldsmith_89 ... wrote:
On Jun 21, 3:42 pm, Gerry Myerson ... wrote:
In answer to your first message in this thread, I think you don't
know what the big-oh notation means.

I've seen it used two different ways in two different contexts; I
know the analysis meaning I imagine you think is the exclusively
correct use of the notation, but I have also seen it used the way I
used it.

Can you cite somewhere where it is used to mean something other than
"less than a constant times"?

http://en.wikipedia.org/wiki/Big_O_notation#Equals_or_member-of_and_o...

See the sub-section titled "Infinite asymptotics."

[snip DG comments re lazy or ignorant ]

Perhaps you misunderstand the Wikipedia article. Nothing in its
"Infinite asymptotics" section supports your notion that it matters
what the logarithm base b is in expressions like O(log_b(n)). As
someone else mentioned, log_a(n) = log_b(n) / log_b(a) . Since
log_b(a) is a constant, if a function is O(log_b(n)) it also is O(log_a(n)).
If you claim otherwise, you don't understand big-O notation.

Incidentally, the "anomalies" mentioned in the Wikipedia article
are unrelated to the above, being instead the ambiguity and
overloading in notations like f(n) = O(m^n) or g(m) = O(m^n)
or h = O(m^n).

--
-jiw

Never mind, I guess I'll seek help elsewhere.

DG

.

Quantcast