Difference between revisions of "Talk:Blow your mind"

From HaskellWiki
Jump to navigation Jump to search
(Mind-blowing considered harmful)
m
 
(10 intermediate revisions by 6 users not shown)
Line 1: Line 1:
==Could these be on their own page?==
 
 
Hi Johannes - Great stuff on your page. Have you thought about making it an actual page in the WIKI? You could then add <pre>[[Category:Idioms]]</pre> and more people would find it. [[User:BrettGiles|BrettGiles]] 16:19, 1 March 2006 (UTC)
 
 
 
==Name?==
 
==Name?==
   
 
Is there a better name for this page? &mdash;[[User:Ashley Y|Ashley Y]] 00:55, 2 March 2006 (UTC)
 
Is there a better name for this page? &mdash;[[User:Ashley Y|Ashley Y]] 00:55, 2 March 2006 (UTC)
   
i completely agree, the name pretty much sucks. but what i really wanted, was to compile a collection of "idioms" that would enlarge the readers perception of what is possible in Haskell and how to go about it. so, i'll have to find a name that reflects this plan. &mdash;--[[User:JohannesAhlmann|Johannes Ahlmann]] 14:13, 2 March 2006 (UTC)
+
i completely agree, the name pretty much sucks. but what i really wanted, was to compile a collection of "idioms" that would enlarge the readers perception of what is possible in Haskell and how to go about it. so, i'll have to find a name that reflects this plan. &mdash;--[[User:JohannesAhlmann|J. Ahlmann]] 14:13, 2 March 2006 (UTC)
  +
  +
== List / String Operations ==
  +
  +
Should this:
  +
  +
transpose . unfoldr (\a -> toMaybe (null a) (splitAt 2 a))
  +
  +
be this instead:
  +
  +
transpose . unfoldr (\a -> toMaybe (not $ null a) (splitAt 2 a))
  +
  +
== Polynomial signum and abs ==
  +
  +
A sensible option for signum and abs for polynomials (with coefficients from a field) would be
   
  +
* signum -> the leading coefficient
== Elegance before pyrotechnics ==
 
  +
* abs -> the monic polynomial obtained by dividing by the leading coefficient
   
  +
As with the abs and signum functions for Integer and for Data.Complex (when restricted to a subring whose intersection with the real numbers is just the integers), the result of this signum is then a unit (ie, a value x for which there exists a y such that xy = 1), and we have
The polynomial algorithms are indeed mind-blowing, like the excess of Mardi Gras. They smuggle in subscripts, which (blessedly) Haskell usually banishes, disguised in a pyrotechnic display of abstraction, replication, concatenation, composition and enumeration.
 
   
  +
* signum a * abs a = a
Gilles Kahn invented a much more docile and beautiful way to do these things with true Haskell elegance long before Haskell:
 
  +
* abs 1 = signum 1 = 1
  +
* abs (any unit) = 1
  +
* abs a * abs b = abs (a * b)
   
  +
== Polynomial negate ==
instance Num a => [a] where
 
fs + [] = fs
 
[] + gs = gs
 
(f:fs) + (g:gs) = f+g : fs+gs
 
   
  +
Why not:
(f:fs) * (g:gs) = f*g : [f]*gs + fs*(g:gs)
 
_ * _ = []
 
   
  +
<haskell>negate = map negate</haskell>
Unlike the mind-blowing version, this code works on infinite series as well as polynomials. &mdash;[[User:dougm:Doug McIlroy]] 20 December 2006
 

Latest revision as of 12:55, 29 May 2013

Name?

Is there a better name for this page? —Ashley Y 00:55, 2 March 2006 (UTC)

i completely agree, the name pretty much sucks. but what i really wanted, was to compile a collection of "idioms" that would enlarge the readers perception of what is possible in Haskell and how to go about it. so, i'll have to find a name that reflects this plan. —--J. Ahlmann 14:13, 2 March 2006 (UTC)

List / String Operations

Should this:

transpose . unfoldr (\a -> toMaybe (null a) (splitAt 2 a))

be this instead:

transpose . unfoldr (\a -> toMaybe (not $ null a) (splitAt 2 a))

Polynomial signum and abs

A sensible option for signum and abs for polynomials (with coefficients from a field) would be

  • signum -> the leading coefficient
  • abs -> the monic polynomial obtained by dividing by the leading coefficient

As with the abs and signum functions for Integer and for Data.Complex (when restricted to a subring whose intersection with the real numbers is just the integers), the result of this signum is then a unit (ie, a value x for which there exists a y such that xy = 1), and we have

  • signum a * abs a = a
  • abs 1 = signum 1 = 1
  • abs (any unit) = 1
  • abs a * abs b = abs (a * b)

Polynomial negate

Why not:

negate = map negate