*************************************************************************
*************************************************************************
***********     Virtual Retina :  CHANGES ACROSS VERSIONS     ***********
*************************************************************************
*************************************************************************

                         Spiking, large-scale retina simulation software.

      Author : Adrien Wohrer
      Institution : Group For Neural Theory, Ecole Normale Supérieure
      Contact : Adrien.Wohrer@ens.fr
      Project started December 2004. 


*************************************************************************
************               VERSION 2.2.1                *****************
************                (April 2010)                *****************
*************************************************************************


USAGE OF VIRTUAL RETINA:
=======================

- Modified the general architecture of the retina_package. Executables
  are now stored in VirtualRetina/bin/. Experiments/ are now stored
  in a separate directory outside of VirtualRetina. Etc.

- Totally removed the obsolete <undershoot-version> of the 
  <outer-plexiform-layer>. The undershoot is now added as a subfield of
  the <linear-version> (see EXAMPLE_cat_X_cell.xml).

- Modified implementation of retinal spikingChannels, and associated
  single units (object "OneCell"). As a result, the XML definition files
  for the <spiking-channel> parts have slightly changed (see EXAMPLE 
  retina files).
  Also, the cells' positions are now given in degrees (instead of pixels).
  
- Added a new functionality for normalization in reconstruction from spike
  trains (option -dn 2 in ReconstructRetina, now the default), leading to
  much nicer reconstructions.
  
  
IMPLEMENTATION:
==============

- Modified #includes for compatibility with gcc version 4.3

- Updated to xmlParameters++ version 1.1.1 (with direct use of libxml2 
  instead of c++ wrapper libxml++).
  
- Updated to CImg 1.3.4
  
- Solved bug in Retina::adjustSizes() : When using as input a 
  file "retina.xml" produced by a previous call to program Retina,
  the new retina produced (and new saved "retina.xml") could display
  a different number of cells in its spiking channels.
  
- Solved bug in <square-array> of spiking cells. Now, an array with a
  single spiking cell has the cell centered in (0,0), as it should.
  
- Changed implementation of ExpCascade in BaseRecFilter (to gain
  stability). Also, added graphical test of BaseRecFilter impulse 
  responses: functions testImpulseResponse() and testIntegratedResponse(). 

- various small code modifications and fixes...




*************************************************************************
************                 VERSION 2.1                *****************
************            (July 07 -> April 10)           *****************
*************************************************************************


10-30-2007 :

-------> NEW XML NAMES:
		* in the OPL, 'adaptation-version' is replaced by 'undershoot-version'.
		  corresponding parameters 'adap-***' replaced by 'undershoot-***'

		* at the contrast gain control stage, all names in '**-amacrine**'
		  are replaced by their equivalent '**-adaptation**',
		  accounting for the biological origin of the fast contrast adaptation 
		  modeled here, which is very likely NOT amacrine cells.


10-12-2007 :

-------> SQUARE ARRAY OF SPIKING CELLS: Bug fixed in the sqSpikingChannel (square array of
	 spiking cells), which did not correctly place the spiking cells in the image.


08-21-2007 : 

-------> improved signal representations (cimg_signal_drawer.h and .cc),
	 with real handling of log scales. Corresponding modifications in src/appli executables


08-08-2007 : NEW PACKAGE ONLINE (version 2.1), with

-------> new file architecture (External_Libraries vs VirtualRetina).

-------> Old, rudimentary Makefile replaced by an installation with CMake.

-------> a VirtualRetina/local/ directory is created, with all required bin/ and lib/

-------> Thorough tutorial in pdf, explaining installation, usage and underlying model.

-------> All spatial xml parameters expressed in visual degrees.

-------> Simpler xml names for the different parameters.

-------> Possibility of automatic download and installation of all required libraries,
         with the shell scipt download_build_all.sh


08-06-2007 : Happy birthday Tim

07-24-2007 :

-------> Changed names of the microsaccade generator

-------> Solved numeration bug in Retina_main.cc (size of input_CP , lign 364)


07-19-2007 :

-------> Default synaptic function for the contrast-gain-control
	is taken by default as	
		 V -> V^2
	instead of the old synaptic function
		 V -> pos_part(V-v0) + pos_part(-V-v0)
	used before that.
	You can use <contrast-gain-control-custom>, to have more liberty
	on the choice of the synaptic function.


-------> Change in the xml names for the GANGLION layers :
	the obsolete 'parvocellular-ganglion-layer' and 
	'magnocellular-ganglion-layer' are now replaced by the simple:
	
	'ganglion-layer'.

	But the former names still work.




03-2007 :

-------> All parameters used in the xml definition file are now 
	followed by the unit in which they are expressed.
	example:
		temporal-step is replaced by temporal-step__sec


-------> All spatial parameters are now expressed in visual degrees
	 rather than pixels as before

-------> A pixel/degree conversion is created.