MathGroup Archive 1995

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Best solution definition

  • To: mathgroup at christensen.cybernetics.net
  • Subject: [mg634] Re: [mg613] Best solution definition
  • From: wmm at chem.wayne.edu (Martin McClain)
  • Date: Fri, 31 Mar 95 11:11:14 EST

Most of what I write is for students, so my vote goes to READABILITY and
CLARITY, unless:

1.  The non-obvious answer is so neat it can't be resisted.  (*But then it
should include in-line comments to help the reader over the non-obvious
spots.*)  And it should never include those comic strip expletives
##.@!%\&@#1#2<>><!&!! so admired in Champaign (except the postfixes // and
/. , which help readability).  In particular, PURE FUNCTIONS  should NEVER
be exposed in public.

2.  The timing is so bad that a human notices it on one execution.

3.  It is an item destined for multiple reuse, where timing is a legitimate
issue.  But even here, two versions should be given: one readable, the
other coded to shave off the microseconds.

>   This is just a question that I have been wondering about.  Frequently 
>when a problem is posed on this group, the way the "best" solution is 
>defined is by timing alone.  I have always thought one of the appeals about 
>MMA is how naturally a solution flows from the typical pencil and paper 
>approach.  Frequently this doesn't lead to the fastest solution but 
>definitely the easiest to read.  When the solution takes advantage of a 
>more cryptic technique, the readability suffers.  How does everyone feel 
>about using timing as the primary yard stick?  How often do you run into 
>calculations where timing becomes a problem?
>   
>   Just for my curiosity,
>   Daryl



  • Prev by Date: Re: Reference works are needed, Timing
  • Next by Date: UnitStep Argument
  • Previous by thread: Re: Best solution definition
  • Next by thread: Re: Polar Plots ...