Difference between revisions of "Vector-space/Versions"

From HaskellWiki
Jump to navigation Jump to search
m
m
 
Line 1: Line 1:
 
== Version 0 ==
 
== Version 0 ==
   
=== Version 0.1 ===
+
=== Version 0.2 ===
  +
  +
==== Version 0.2.0 ====
  +
  +
* <hask>Data.Cross</hask>: generalized cross products
  +
* <hask>distribD</hask> now takes only one function argument and ties its own knot.
  +
* Renamed <hask>distribD</hask> to <hask>distrib</hask>.
  +
* Now there are two power series modules: <hask>Maclaurin</hask> (was <hask>Derivative</hask>) and <hask>Horner</hask>. Now <hask>Derivative</hask> is an indirection to one or the other, for testing. See Doug McIlroy's paper "The Music of Streams".
  +
* Factored <hask>Data.AdditiveGroup</hask> out of <hask>Data.VectorSpace</hask>
  +
* <hask>LinearMap</hask> has a new data representation for linear maps, to replace the simple function representation. Used in <hask>Data.Maclaurin</hask>.
  +
* <hask>Data.Derivative</hask> is now more than an indirection module. It re-wraps a power series type to provide different instances for <hask>VectorSpace</hask> and <hask>InnerSpace</hask>.
  +
* Added support for OpenGL vector types. In ghc-6.8.2, I get an error if a <hask>LMapDom</hask> instance is compiled in any module except for <hask>Data.LinearMap</hask>. Until this code is removed, the <code>vector-space</code> package depends on <code>OpenGL</code>.
  +
* Tests in <code>FieldTrip</code> show a dramatic slow-down, in contrast to my intended speed-up.
  +
* Andy Gill found at least one cause for the slow-down, which has been fixed. I think we have a long way to go. There appear to be lots of redundant dictionary construction.
  +
* <hask>Data.Horner</hask> removed temporarily, until re-sync'd with <hask>Data.Maclaurin</hask>.
   
 
==== Version 0.1.3 ====
 
==== Version 0.1.3 ====
Line 25: Line 39:
 
* Changed repo location from darcs.haskell.org to http://code.haskell.org/vector-space.
 
* Changed repo location from darcs.haskell.org to http://code.haskell.org/vector-space.
   
* '''Oops''' - there's a serious mistake in the derivatives code in versions up through 0.0.1. I'm working on a solution.
+
* '''Oops''' - there's a serious mistake in the derivatives code in versions up through 0.0.1. I'm working on a solution. [Fixed in 0.1.]

Latest revision as of 19:07, 11 November 2008

Version 0

Version 0.2

Version 0.2.0

  • Data.Cross: generalized cross products
  • distribD now takes only one function argument and ties its own knot.
  • Renamed distribD to distrib.
  • Now there are two power series modules: Maclaurin (was Derivative) and Horner. Now Derivative is an indirection to one or the other, for testing. See Doug McIlroy's paper "The Music of Streams".
  • Factored Data.AdditiveGroup out of Data.VectorSpace
  • LinearMap has a new data representation for linear maps, to replace the simple function representation. Used in Data.Maclaurin.
  • Data.Derivative is now more than an indirection module. It re-wraps a power series type to provide different instances for VectorSpace and InnerSpace.
  • Added support for OpenGL vector types. In ghc-6.8.2, I get an error if a LMapDom instance is compiled in any module except for Data.LinearMap. Until this code is removed, the vector-space package depends on OpenGL.
  • Tests in FieldTrip show a dramatic slow-down, in contrast to my intended speed-up.
  • Andy Gill found at least one cause for the slow-down, which has been fixed. I think we have a long way to go. There appear to be lots of redundant dictionary construction.
  • Data.Horner removed temporarily, until re-sync'd with Data.Maclaurin.

Version 0.1.3

  • InnerSpace instance in Data.Derivative

Version 0.1.2

  • Renamed dId to idD and added fstD and sndD.
  • Renamed bilinearD to distribD.

Version 0.1.1

  • Added some missing haddock docs
  • Added a Complex instance of VectorSpace

Version 0.1

  • Changed representation of the derivative part of a :> b. Was a :> (a :-* b). Now a :-* (a :> b), which can also be written a :~> b (leaving the derivative's linearity implicit). The type a :~> b means infinitely differentiable functions from a to b.

Version 0.0.1

  • Oops - there's a serious mistake in the derivatives code in versions up through 0.0.1. I'm working on a solution. [Fixed in 0.1.]