Division Prediction Program (DPP)

Authors
Anaelle Pierre 
Dmitry Ershov 
Nicolas Minc Team at IJM



3rd party functions
matlab funciton
"subdivide_tri" by Peter A. Karasev 25 Nov 2009




Notes.




DPP
version 1
2016
Author: Anaelle Pierre
---------------------------------------------------------
- DPP as separate scripts
- data structures are images 
- force calculator is slow
- inconsistency in the functions
---------------------------------------------------------




DPP
version 2
edit: oct 2017
Author: Dmitry Ershov
---------------------------------------------------------
- made it object-oriented
	- class DPP_cell - creation and handling cells.
	- class DPP_simulator - creation and handling simulations.
- corrected the yolk show
- corrected the nucleus show
- made a bunch of graphs with explanations.

- corretcted  the inconsistency between Cartesian coordinates and Matlab matrix dimensions 
	- why: in matrices 1st dim is NOT x, but y!!!
- corrected the MT stability calculation
	- why: The formula was really weird and did not work in test cells.
	- fixed formula works well (of course, visually).

TO DO:
- improve test for "if the system is equilibrated"?
	- use the rejected count instead?

- improve force calculator:
	- fix the polarity "as seen from the pole"??? Smoothing?
	- improve the r-iteration: make it [while ~outside] instead of scanning all values
	- r-maximum can be even shorter: generic cell size instead of the image size.
	- clean obsolete matrices?

- save each iteration? each 10 iterations?
---------------------------------------------------------



DPP
version 3
edit: dec 2017
Author: Dmitry Ershov
---------------------------------------------------------
- base ver 2. 
- force calculator is rewritten almost entirely. Waaay faster now.
	- MT azimuthal angles have more intuitive orientation now.
	- azimutal angles are limited by angle limit directly (faster!)
	- all MTs are in one 3D matix (faster!)
	- changed polar density correction to "density-oriented" rather than "field-oriented", it is faster and more realistic. And easy to understand.
	- changed force calculation (in Cartesians; a bit faster)
	- changed Torque_Force calculations according to the new MT orientations (faster!)

- DPP_simulator class 
	- has more features (reset state, visualize simulation etc...)
	- condition on the accepted new state is strict now.
		- to compensate for this Jump_chance needs to be higher now.
		- this gives better stability search.
---------------------------------------------------------


