37 lines
1.1 KiB
Plaintext
37 lines
1.1 KiB
Plaintext
From: tim_one at email.msn.com (Tim Peters)
|
|
Date: Sat, 24 Apr 1999 05:17:10 GMT
|
|
Subject: Time complexity of dictionary insertions
|
|
In-Reply-To: <glmpv4vnhh6.fsf@caffeine.mitre.org>
|
|
References: <glmpv4vnhh6.fsf@caffeine.mitre.org>
|
|
Message-ID: <000901be8e11$b4842560$f09e2299@tim>
|
|
X-UID: 751
|
|
|
|
[someone asks about the time complexity of Python dict insertions]
|
|
|
|
[Tim replies]
|
|
> Min O(1), Max O(N), Ave O(1). If the hash function is doing
|
|
> a terrible job (e.g. maps every key to the same hash value), make
|
|
> those all O(N).
|
|
|
|
[one person confuses the issue]
|
|
> C++ STL junkies know this as "amortized constant time".
|
|
|
|
[and another compounds it]
|
|
> So does anyone who has ever studied much at all about algorithms, data
|
|
> structures, and optimization.
|
|
>
|
|
> It's not a C++ thing. It's a computer science thing.
|
|
|
|
This one-ups-man-ship would be a lot cuter if Python's dict insertion were
|
|
in fact amortized constant time <0.9 wink>. It's not, and the answer I gave
|
|
doesn't imply that it is. Insertion in STL hashed associative containers
|
|
isn't ACT either.
|
|
|
|
reassuringly y'rs - tim
|
|
|
|
|
|
|
|
|
|
|
|
|