changelog.txt

2018-11-01 Egor Sobolev <egor@impb.psn.ru>
 * updated make rules to compile mpi version with mpicc
 * added utility to convert usage output to html
 * changed mpi.h to mpi/mpi.h in include directives
 * added include directive for rng.h
	
2016-07-28 Egor Sobolev <egor@impb.psn.ru>
 * added parameters for PDB chain heating procedure
 * changed heating procedure for PDB chain to simple one
 
2016-04-22 Egor Sobolev <egor@impb.psn.ru>
 * removed linking of MKL or GSL
 * added source code of pseudo-random grenerator mt19937 and ziggurat algorithm
 * added 1st order matrix integral sheme and changed options -4 for 4rd order to -g <int> to specify one of 1,2 or 4 order.
 * added total energy functional (E)
 * added uniform initial charge distribution (-y U)
 * fixed skipped minus in S functional
 * added options to control norm correction: -H <thr> - norm correction threshold and -n <tm> - time interval interval for norm correction. Norm will be corrected if |sum(b^2)-1| > thr or at regular time intervals tm
 * changed -O3 to -O2 for GCC since in GCC 4.9.2 -ftree-loop-optimization included in -O3 generates segmentation fault
 * changed procedure writes 2-int header (n, n0) in final point file to syncronize processes
 
2015-04-20 Egor Sobolev <egor@impb.psn.ru>
 * fixed calculation of first initial point offset when restart from control point (-y S -c)
 * fixed position in final point set file when restart from control point (-w <file> -c)
 * changed matrix exponential approximation, shifting and scaling
 * added rational approximation of small degree for matrix with small norm
 * added option to enable 4th order Magnus (-4)

2015-04-15 Egor Sobolev <egor@impb.psn.ru>
 * added functionals x = sum(n * |b(n)|^2, n=1...N), dx = sum(n * d|b(n)|^2, n=1...N)
 * improved trajectory log records
 * added doc/exitcodes.txt
 * modified thomas procedure to process sub- and superdiagonal as complex with inverse real part and equal imaginary part
 * modified expmv procedure to shift on smalest diagonal number and scale on l-norm
 * modified numeric sheme to calculate new quantum values before last runge step
 * added 4th order calculation of quantum matrix integral on step with Magnus expansion and Simpson's rule
 * added entropy functional S=P[n]*log(P[n])
 * removed cblas and clapack calls
 * fixed bugs rnd.c related to gsl 
 * refactor chain class to provide only dv and dv_hst and implemented implicit du calculation inside schemes
 * fixed reading initial states (-y S) after eof in mpi mode
 * fixed error of restarting with final point set writing

2015-02-24 Egor Sobolev <egor@impb.psn.ru>
 * fixed bug of reading quantum parameters from file
 * added memory usage info into log exclude seq & pbd_equlibrate
 * added storing phase variables into file at end point of trajectory for each sample to continue trajectory if it will be need
 * added abstract inteface for initial states
 * unified all implemented initial states
 * added initial state as set of trajectory points so it allows to continue trajectory from stored set of endpoints
 * added opiton -y to choose type of initial state: 1-one on n0 (-0) site, P-one state, S-set of states

2015-02-23 Egor Sobolev <egor@impb.psn.ru>
 * released abstract interface to result functionals
 * unified all implemented functionals
 * added input file of output sheme (-O) allows to set output functional and corresponding output steps
 * added common output step (-Q) allows calculate long time with rare output
 * added options (-0) to set origin site
 * removed old commented source code
 * changed format of output file to describe variable set of output functionals
 * fixed etrans_read.m to read new format
 * added help into etrans_read.m

2015-01-19 Egor Sobolev <egor@impb.psn.ru>
 * added electric field term (-U)
 * added current functional j(t)
 * added equations parameters in output file
 * restart with parameters from restart file
 
2015-01-06 Egor Sobolev <egor@impb.psn.ru>
 * added Peyrard-Bishop-Dauxois model of classical chain
 * default values of options depend on choise of model
 * added functions for serializing equations
 * unified interface to get initial states of classical chain by direct generation or by heating procedure
 * removed by site parameters in classical equation

2014-10-26 Egor Sobolev <egor@impb.psn.ru>
 * added functionals b4^-1(t), enq(t), enu(t), env(t)
 * added output step of functinals (-w)
 * added random initialization of u and v which takes into account the dispersion
 * removed heating procedure and options
 * random phase of standard inital charge distribution (1 on central site)

2014-10-25 Egor Sobolev <egor@impb.psn.ru>
 * release of an abstract interface to the classical subsystem
 * added wrapping interface for random streams
 * hiding implementation of options processing
 * structuring the code of the main loop of integration

2014-01-21 Egor Sobolev <egor@impb.psn.ru>
 * removed output step for charge transfer functionals. it is just equal
   to the time step (-h)
 * zero value of -Q and -f produce only initial and final frames of
   probability and displacments
 
2013-07-26 Egor Sobolev <egor@impb.psn.ru>
 * fixed measurement of the time spent on calculations between control
   points

2013-07-24 Egor Sobolev <egor@impb.psn.ru>
 * replaced Intel MKL by GSL and LAPACK
 * added compile-time selection of math library
 * added selection of compiler

2013-07-22 Egor Sobolev <egor@impb.psn.ru>
 * added compile-time selection of real type for single trajectory
   procedures: float or double
 * added downloading of argtable2 sources in Makefile

2012-11-30 Egor Sobolev <egor@impb.psn.ru>
 * changed the range of "good" approximation from 2.0 to 0.75 * M_PI 
 * unified code for first adjustment of time step and code for
   adjustment time step on iterations 
 * fixed the interval of linear extrapolation of quantum equations
   matrix in the procedure for adjustment of time step 
 * added option -Q - frequency of quantum frames output 
 * fixed estimate of the number of trajectories computed between
   control points 
 * released 0.12