               ==============================================
                         MACSIMUS version file
                   (with history, bugs, and to-do list)
               ==============================================

BUG AND TO-DO LIST
^^^^^^^^^^^^^^^^^^
KEY is priority or severity (0=unimportant, 9=important, X=critical)

cook:
  X @A1 doubled EkinP term ? (M_P scaled?)
  9 Langevin thermostat not sufficiently implemented, Gear only:
    with -m4 gives *** Too low virial pressure vs. tensor accuracy
    consider DOI: 10.1080/00268976.2012.760055
    also, tau.T should be reinterpreted (double it and modify random force)
  9 Andersen thermostat issues:
    AndersenCM thermostat wrong for TIP4P (not SPC) -- why?
    Andersen wrong for RTIL -u9999 (all bonds vibrating)
    Andersen with elst. field suspicious
    Andersen thermostat not as accurate as expected (No.f problem?)
    low-T crashes if restarted with Andersen
  9 Elst+mag field virial of force and pressure tensor unclear (never use NPT)
  9 pressure/pressure tensor accuracy problems for PERSUM, -m4: see projects/tests/constrdyn/cangles/
  9 Ewald issues:
    parm(): "just calculate charges" should be cleaned!!!
    Ewald autoset via error estimate out of order for polarizable molecules with no or small charges (ponder such molecules in elst. field)
    Ewald error estimates use the assumption that charges are distributed randomly which overestimates (several times) the errors for dipolar models
    Off-diag pressure tensor has not been tested for GAUSSIANCHARGES
    +? consider Ewald cutoff=const*L for NPT
  8 if LINKCELL is used on a few molecules, it crashes (see BUG in lc.C)
  8 rehack the .cp creation logic in the .cpi spirit to avoid maincps.c and simmeas.c synchronization and empty column |M| or |J| if lag.M/lag.J not set
  8 rescale="XY" does not work correctly with Lx,y,z set and tau.rho
  8 reconsider slab autocenter by sites, not molecules
  8 the kinetic pressure correction is not without caveats - do not trust the defaults and consult the manual
  8 add pressure tensor components corresponding to the external elst field
  7 NPT + diffusivity/SF should be checked again (perhaps OK for cube only)
  6 POLAR: improve numerical errors: see locEnel-README.txt
  6 tau.sat<0 cannot be started with init=2 because it is changed to init=1 in a wrong place, or the order of setup is wrong
  6 Gear+Nose does not survive direct start, must (a few cycles) run Berendsen first
  5 parallelized structure factor probably out of order
  5 k-vector initialized in Ewald even w/o charges - crash for dilute systems
  5 cutelst Ewald unnecesarily initialized, with alloc problems
  5 SLAB: (?) vdrift in x,y not autoset - doublecheck!
  5 POLAR elst energy/forces rounding problems (<.1K/charge) for shell=1000
  4 cutelst requires PRESSURETENSOR for only shape change (tau.rho)
  3 writing plb may be shifted by h or so for long runs (1us) - rounding problems
  3 keyword LJ in SIMNAME.cpi out of order (in most versions)
  3 extend Rgyr to the whole config. (cf. mXsection,cXsection), doublecheck Rgys vs. moment of inertia
  3 cutelst: should choose which pressure to use for barostat
  3 add ~M^2 term to energy also in non-Ewald
  2 make 14 inclusion in rdf switchable (now: 14=on in plain version, 14=off in link-cell list)
  2 omit 1-4 at all if not needed (e.g., distance14=0)
  2 key="sort+z" problems, use sort="z"
  2 clean the free version (misleading info about boxes...)
  2 clean simcg (allocate M in bigger chunks)
  2 omit M/J from .cp as default?
  1 siminit: loop (n,0,option('m')+3) sdsralloczero(a[n],i): 3 too many?
  1 should warn if positive Emin in ble-file
  1 improve Emax for debug pot -v31
  1 double vector[4] support? (better cache usage, but speed gain ~1% only)
  1 is it necessary to call rhs at init if Verlet?
  1 LJcutoff=0 should turn off LJ (workaround: use option -j#)
  ? technical problem: calc.c uses rndgen.c
  ? virtual volume change for rescale!="xyz" and constrained model? Use kinetic pressure tensor components?
  ? it happended that one zero-length cycle was executed after stop was reached, but I cannot reproduce it anymore...
  ? QQTAB and qtype&4 optimization (for BK3+ions)
  ? problem of large noise in Jx,Jy,Jz caused by water - split measurements?
  ? unroll intermol() for m1==m2 and m1!=m2 ?
  ? LJ cutoff/constrained models/small boxes and the virtual volume change method
  ? LJ cutoff and virtual volume change small boxes
  = WARNINGs *** box has changed by >5% (log-scale) and k-vectors remain unchanged and  *** cutoff=11.000000 exceeding minL/2=10.000000 by 10.00% unnecessarily printed even if no charge
  = doublecheck the Yeh-Berkowitz correction (now in all 3 directions, equivalent to eps'=0)
  = thermostat="NPT" rescale clash when started from init=3
  = sometimes enthalpy = internal energy
  = Selected #defines: printed to stderr is not the same as printed to prt
  - when rescaling (load.n[]), automatic cutoff determination is incorrect (will not be fixed but noted in the manual)
  - automatic corr setup fails if thermostat/barostat changed within one run (e.g., Berendsen -> NPT) - noted in the manual, will not be fixed

blend:
  8 tetrahedral dependants => 5 bonds (one missing)
  8 implement Gaussian charges
  7 "shake" with "dependants" not supported
  7 Gaussian charges not supported
  6 hot keys ,; with -b0/-a0 change the number of steps; remedy: <Esc> sd=999; <Enter>
  5 warn about positive Emin in nbfixes
  4 extend plb2plb also to mol and gol files
  3 blend: more fool-proof logic if .che changes (check time stamp?)

other:
  9 change home-made int4.h into standard stdint.h
  7 `makepept charmm22 ace G ct1' incorrectly replaces CT2 by CT1 and charge in G
  5 showcp -c -l - blocking not needed, -c strange printout, also Cov
  5 showcp change int to unsigned to allow for 2^32-1 records (not 2^31-1)
  5 showcp ignores column number with -a
  4 showcp does not read .cpz with -f -t (full range is OK)
  3 water (hoh) translated into internal name (TIP3P) and TIP3P.mol interferes with molcfg. Workaround: use the model names like TIP3P, not hoh etc.
  3 show: -@ out of order, the follow mode fails
  ? random numbers: Mersenne Twister instead of Ziff?
  ? pgcc does not compile random numbers
  ? show shows last frame as empty for old plb format
  = cppak crash

HISTORY
^^^^^^^
  !! serious/dangerous bug fixed
  ! bug fixed
  + feature added
  * feature changed
  - feature removed

cook 3.7g:
  + center.sp added
show 2.2m:
  + [] always goes by a frame, PhDn,PgUp in the %-mode changes files

cook 3.7e:
  + variable Ttr-Tin added (returned) to the statistics

cook 3.7d:
  !! slab cutoff correction was wrong for more LJ sites than 1 - fixed
  
cook 3.7c:
  + box.follow added
  ! following the periodic b.c. for MSD, conductivity fixed (affects border cases)

cook 3.7b:
  ! el.L added to fix a minor flaw in Ewald initialization
  ! polishing plb2diff and plb2diff.sh

cook 3.7a:
  ! wrong pressure in Berendsen barostat (for GC, cutelst) fixed
  ! virtual volume change fixed
  ! Gaussian charges r-space splines rewritten to be numerically
    more stable
  * long double used in spline table calculation
    (better, but not necessary)
  * pressure output and control improved

show 2.2l:
  ! wrong plb file exported after editing (missing L) fixed
  + 'a' added as alias for Tab (standard orientation)

evu:
  ! bug buffer overflow found by J. Janek fixed 2024-08-27

blend 2.4f:
  ! eigenvalues of gyration and inertia and tensors also sorted

sumerr 08/2024:
  ! confusing help (of example) fixed

cook 3.6w:
  + Hamaker constant
  ! missing shake4.c, shake30.c, xxxdef.c fixed

cook 3.6v:
  ! Widom fixed

ev,evu:
  ! character '!' now accepted (! in '' is not comment)

show 2.2k:
  + white background for NFF,POW is now default, can be turned off by button [bg] or hot key w

cook 3.6s:
  ! cutoff electrostics (COULOMB=2) now (again) available in the spline version:
    A bit faster, but also slightly less accurate than the direct (COULOMB=0) version
  ! Fennell-Gezelter (COULOMB=3) now (again) available
  + Hammonds-Heyes alpha' correction implemented: https://doi.org/10.1063/5.0101450

********** Posted to GitHub **********

cook V3.6r:
  ! enthalpy in NVT/NVE
  ! el. background in FREEBC control
 
cppak:
  ! bug in the linked-list used inernally fixed

examples/*/ added.

cook 3.6s:
  * default slab.geom changed to 0, to suppress warnings if a SLAB version is used for a cubic box
  + Center-of-mass sampling added (tau.CM)

makemake:
  * metamake renamed to metamake.mmk, command !private added

plot 2.0t:
  ! kill all button wrong info fixed

---------------------------------------------------------------------------

cook 3.6r:
  ! wrong enthalpy for NVT,NVE fixed
  ! control of charged system and el. background for FREEBC and STARS fixed

showcp V2.0f:
  ! bug of V2.0e (selection of columns) fixed
  ! cppak memory bug fixed

cook 3.6q:
  ! Saving of plb moved BEFORE removing a molecule (see SLAB, slab.out)
    Thus, the molecule being removed is present in the last frame
    Good for dt.plb=0

plot 2.0s:
  ! missing column 2 filled by the prev value in plot file (w/o :)

ev,evu:
  * modulo is now @, not ` (in tabproc, modulo is %) -- later returned back

showcp V2.0e:
  *- option -z extended, -e removed, lin.regr. of Etot removed

cook 3.6p:
  + option -b changed, the default is now to test SIMNAME.stp every 15 s

cook 3.6o:
  + Hoover NPT (the total energy looks OK, but has not been tested)

cook 3.6n:
  + external forces by calling Gaussian

blend 2.4e:
  ! option -b (could not minimize)
  ! metamake with exp-6,busing now OK (not fool-proof)
  - legacy DOS, FAT, Imworking() removed

cook 3.6m:
  + with SLAB, slab.out: kinetic properties of the evaporated molecule calculated
  ! missing drifts.c added
  + (GUI) font # 4 added (not used in MACSIMUS)

cook 3.6l:
  + magnetic dipole implemented
  + Pcfg (see variable virial) reported in convergence profile
  ! structure slab_s shifted frame (=> slab-thermostat turned wrogly on)
  ! CP record bugs fixed (FREEBC shifted frame, bad size determined)
  * measurement of integrals of motion rehacked
  * drift correction and measurements separated
  ! P unit in protocol (by dt.prt) fixed, other instances were OK
  ! bug (slab structure, thermostat by mistake) from 3.6k fixed

cook 3.6k:
  + magnetic field implemented for moving charges
  + lag.lM and lag.aM added
  + option -k0 added

show 2.2j
  * hotkey '0' changed into the prefix of 2 decimal digits

cook 3.6j:
  ! bug in Berendsen barostat and MTK anisotropic (introduced in V3.6g)
  ! bug missing columns in SIMNAME.cp

rdfg 10/2022:
  ! bug (write to closed file) fixed

show 2.2i:
  - endian change for match (-m) removed, the same for plbmatch.c

cook 3.6i:
  ! z-profile of Gaussian charges fixed
  * off-diagonal pressure tensor for dependants re-checked (cf. norm.c, macros above depend_f)

cook 3.6h:
  * Gear's integration upgraded:
  - development code partially removed
  + Janek's version added
  ! POLAR support fixed (although ASPC+Verlet is generaly recomended)
  + Tests added to the linked-cell list method to catch a particle out of box
  ! bugs (bad pressure units) fixed

cook 3.6g and utilities:
  - legacy cubrt() eventually hard-replaced by cbrt()
  - obsolete REDUNITS switch removed
  * P vs Ptens logic changed:
  + variable `virial' added, pressures reconsidered and cleaned
  ! atomdist and plbcheck fixed

cook 3.6f:
  ! Ballenegger-Arnold-Cerda correction (Yeh-Berkowitz for charged system
    compensated by background) added

cook 3.6e:
  ! versions c,d (pair sums, not linked-cell-list) elst energy not included in total energy

cook 3.6d:
  + unsplit added
  ! POLAR + cprof.c fixed

cook 3.6c:
  ! bug (POLAR density profiles, introduced recently) fixed

makemake:
  * This legacy code was simplified:
    - DOS support removed
    - home is allowed to constain spaces (and some similar characters),
      this is useful under WSL

cook 3.6b:
  ! SHARPCUTOFF fixed (wall for metal not checked, likely wrong)
  + ANCHOR improved

showcp 2.0c:
  + option -z added

cook 3.6a:
  - unused and obsolete switch #define COOK removed
  * some polishing
  * measuring dipole moments cleaned, manual updated
  + ANCHOR extended to CM in selected direction and CM of all
    molecules not already anchored
  * default equalize.sp changed to negative big number = whole configuration
    equalized, the former default (0) equalized only species 0

cook 3.5i:
  !! missing initialization of shift,vshift (results undefined):
     (just by chance they contained numbers << 0)
  + Ewald background added, el.bg and el.epsq added
  * negative option -q now denotes negative charge scaling,
    to redistribute charges use el.bg=2
  + removing molecules in the slab,trickle,droplet geometry:
    slab.out, removemol.n
  * default -r1 changed into -r2
  ! call to molcfg is now performed after reading N[] so that the
    mol-file matches the plb-file (cf. load.N=4)
  ? placement of initNo likely fixed, to be doublechecked (see V3.6a)

cfg2asc:
  + Now various info on cfg-files is extracted and printed

cook 3.5h:
  + radial and cylindric density profiles, see variable slab.geom

showcp 2.0b:
  ! linear regression uses t=DT*n, not n

cook 3.5g:
  * the kinetic pressure correction info improved
  ! kurtosis bug with kinetic pressure correction (fixed)

cook 3.5f:
  ! (not harmful) too long lag for some quantities

show 2.2h:
  ! option make animated GIF added

cook 3.5e:
  ! slab energy corection from z-profile fixed (likely never useful)
  * code for surface tension cleaned a bit

cook 3.5d:
  + option -d-# added (FREEBC only)

utilities:
  + most filters as tabproc, sumetc, ev now accept environment variable FMT
    for the numerical format.  The default is %.8g.  To be extended...

cook 3.5c:
  ! the last cluster was omitted in x-section calculations and periodic
    cluster detection - fixed

cook 3.5b:
  ! linking duplicate of cl fixed
  ! periodic cluster count fixed
  * some compiler warnings suppressed by unnecessary added code

plot,blend,show,jkv:
  + new version of fullscreen (e.g., show -gf) and
    X-geometry extensions (see help (G)eometry)

cook 3.5a:
  ! units polished
  + groups added, partial support for TCF of M and J in groups
  + export of M and J by coordinates and groups to .cp (noint=1 recommended)
  + group[] added to data
  * lag.cond renamed to lag.J, lag.visc renamed to lag.Pt
  * lag.M added

cook 3.4r:
  + cl.mincluster added

blend V2.4d:
  ! ar_dih_limit=-1 caused repeating aromatics=dihedrals in some cases
    (but the same with gromos96), see blendgen find V2.4d

blend V2.4c:
  + on hotkey E = "info", energy printed also on screen

UTILITIES
  + oneline, line, molren added

cook 3.4q:
  ! export Nose-Hoover xi shifted time fixed

cook 3.4p:
  + export Nose-Hoover lambda added

cook 3.4o:
  ! small fixes

cook 3.4n:
  + export of positions of Drude charges (SIMNAME.dlb)
  + true charge density profile for Gaussian charges

cook 3.4l:
  ! In NPT, enthalpy was calculated using the configurational pressure instead of
    parameter P. (NPT is recognized by nonzero tau.P.)
    Var(H) is meaningful with thermostat="NPT" only.
    Caveat: not well documented in the manual.
cook 3.4k:
  ! organizational bug PERSUM in info (mainhlp) and cook/configure.sh fixed
  ! bad print of vshift fixed

cook 3.4j:
  + NPT documentation improved

cook 3.4i:
  ! Epot reported by cook -l had not contained En.el - fixed

show 2.2g, utilities
  + imperfect ad hoc support for cook -n4, to be extended

cook 3.4h:
  ! bug in file name with option -n fixed (.plb was renamed to .plb~)
  + -n upgraded

cook 3.4g:
  + Eelst[] renamed to el.E[]
  + oscillating electric field supported, control quantities added:
    el.f, el.t0, el.phase[]

cook 3.4f:
  ! bug in print with el.kappa=0 fixed

blend 2.4b
  - bug fixed (int div 0 for -g1 -m1)

cook 3.4e:
  - dih.res removed (DIHEDRAL only)
  ! bug in ANCHOR print fixed

cook 3.4d:
  + option -_ added, see help (s)caling
  - option -^ simplified (cannot change the Drude charge anymore)
  + option -u-# added (select vibrating bonds according to wavenumber)
  + scf.domega, scf.margin (for automatic setup of scf.omega)

plot 2.0q:
  ! bit cleaned, hotkeys [ ] 0..9 fixed
  + hotkey U added

cook 3.4c:
  * #define MONITOR changed to #define LOG, +m in cook* name changed to +l
  + Morse potential (expanded to dist^4), #define MORSE, +m in cook name

cook 3.4b:
  ! organization bug in printing EPS0 fixed
  * "ECC Epot" instead of Epot in previous versions
  + "ECC Uscaled" = conventional energy (not available directly in ECC before)

cook 3.4a:
  * ECC pressure and energy transparently printed

cook 3.3s:
  ! bug conversion K->J/mol fixed

cook 3.3r:
  * monitoring individual terms as LJ, Eintra, first molecules etc.
    now available under #define MONITOR and direct pair sum (not LINKCELL)
    some variables (LJ, PLJ) need a record in .cpi
    LJ is in K (not J/mol as before)
  * LJ pressure added (with MONITOR, see above)
  - option -j replaced by No.first and No.rotatefrom (MONITOR, ?)
  + option -j now means scaling of Lennard-Jones parameters
  * some changes in reading in the cfgs recorded by -m0
  * minor code changes
  ! bug in 3.3r only: quantities from .cpi by mistake multiplied by R

cook 3.3q:
  ! re-read mode and -f: energy fixed (P still wrong for GAUSSIANCHARGES)

show 2.2f:
  ! background color bug fixed

show 2.2e:
  * code cleaned (vector eye[] is now the true eye position)
    mid mouse and arrows change the viewpoint wrt screen
    raytracing now better corresponds to the screen
    bar shown a bit better (still cheap and approximate)
  - reverse endian support removed

cook 3.3p:
  + dielectric constant calculation (partly) returned to CUTELST version
  * default el.epsinf changed to 3e33 also for cut-off electrostatics, but
    it is used in postprocessing only (no term ~|M^2| as in Ewald - to be
    added?)

cook 3.3o:
  !! malloc(0) bug fixed: result of malloc(0) used => overflow for >3 species

cook 3.3n:
  ! CPmark error fixed

cook 3.3m:
  + cp-file now supports simulation time info
    cppak and (partially) cp2cp updated

showcp 2.0a:
  + plots show time in ps, help changed, options changed (-m -| -a -p)

show 2.2d:
  ! bug (color searched also in the part of id before /) fixed

show 2.2c:
  ! bug (probably from 2.2a) in wire modes and marking fixed

cook 3.3l:
  + CSVR thermostat (Bussi et al.) simply and inefficiently implemented

show 2.2b:
  + option -P = presenter mode (F4,F5 = hotkey 'i')

cook 3.3k:
  + equalization improved, variable equalize replaced by equalize.{cfg,mol,sp}
  ! conductivity of polarizable models (imprecise inclusion of Drude charges)
  + slab.sym added: slab asymmetry detection
  * new SI definition 5/2019 implemented
  ! small implementation problems with gaussian charges fixed

cook 3.3j:
  * ECC improvement; e.g., Pcorr now split and reported in 2 parts, P2 and P3
  + Limited pressure calculation added to the playback mode (-m0,-m1).
  + Playback mode small changes

cook 3.3i:
  * el.fast renamed to el.ecc, control changed
    pressure checked via virtual volume change for ionic liquids
  ! current density J for polarizable models bug fixed
  + cook diffusivity (e.g., called by plb2diff) now detects and fixes one
    move >L/2 between frames

blend 2.4a:
  * .gol files removed, color and radius info moved before atom-id
  * option -p changed (-p-# now forces gol)

show 2.2a:
  * changed accordingly (backwards compatible)
  ! bug z-viewpoint for raytracing fixed

tomoil removed

blend 2.3f:
  + MC minimization, reasonably good for simple cases with constraints
    WARNING: constraints with dependants not still supported (e.g, TIP4P)!
  + features added to GUI (hot keys []j)
  + option -W added
  ! steepest-descent with constraints fixed - not buggy but slow
  - console-based input of minimization parameteres removed (hotkey F9)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2019-08-01 <<<<<<<<<<<<<<<<<<<<<<<<<<<

show 2.1d:
  ! option -\\ fixed (actually, it is not needed, -I| can be used instead)

cook 3.3h:
  + CLUSTERS: cl.format&32 (print all clusters)
  ! distancecheck removed from the initializer (too slow for large systems)

cook 3.3h:
  * small changes with CLUSTERS

cook 3.3f:
  ! ertest Gauss ga.g12 printed incorrectly - fixed
  + ertest Gauss improved
  + BJERRUM improved

cook 3.3e:
  ! small formal changes in BJERRUM

plot 2.0p:
  ! button [recalc] = Ctrl-U = F9

cook 3.3d:
  ! memory leak fixed

cook 3.3c:
  * small changes; improved: neutrality test, cfg2asc, asc2cfg, control od
    plots spawned by rdfg, showcp

show 2.1c:
  ! bug of previous version for SHELL fixed

cook 3.3b:
  * bjerrum.mode changed

cook 3.3b:
  * Ewald initialization improved

cook 3.3a:
  * WALL version reconsidered:
    - "#define SLIT" now denotes 2D periodic b.c. (z-images do not interact)
    - walls require "#define SLAB 1" and are dynamically set by structure wall.*
    - pressure on walls calculated
  * virial-of-force terms added to wall calculations (the virial pressure
    is NOT the system pressure anyway - use Pwall[0],Pwall[1] instead) as
    well as elst field (only pressure tensor components)
  ! bug in virtual volume change scaling in some coordinates fixed

cook 3.2e:
  + optional thread-parallelism using barriers (may be faster on some systems)

show 2.0m:
  + now Donor=Acceptor accepted, range of accepted bond lengths increased
  + marking now works in all modes (not only "wire")
  ! bug in showing H-bonds in the wire modes

cook 3.2d:
  ! the speed of rescaling using tau.rho is now the same both for swelling
    and shrinking (swelling had been 2x faster)

cook 3.2c:
  + xs.mode="x", "y", "z" added

cook 3.2b:
  ! wrong DT in StaAdd for Mx,My,Mz,M2 fixed, lag.err multiplied ny noint, too

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2018-09-13 <<<<<<<<<<<<<<<<<<<<<<<<<<<

ev, tabproc: Fortran-style numbers (as 2d3) bug fixed

plot 2.0n:
  ! PS output of file names/formulas containing ()\
plot 2.0m:
  + toggle showing error bars, hotkey=[^]

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2018-08-21 <<<<<<<<<<<<<<<<<<<<<<<<<<<

plot 2.0l:
  + batch mode added (option -b)
  + option -t added (do not recompile)

cook 3.2a:
  * default value of the number of degrees of freedom for NVE,Berendsen changed
  + flag corr&64 added to control the above behavior
  * Gear line with errors printed with -v4 only (-c unchanged)

show 2.0k:
  ! export of background (-bg#HEXCOLOR) to nff fixed

cook 3.1m:
  + vibrations for general constraints
  * nm.method flag meanings changed

cook 3.1l:
  * small optimization of constraint dynamics and nmc.c

cook 3.1k:
  ! ewald port to SF: bug in control fixed
  * small technical improvements

cook 3.1j:
  PERSUM default option -x changed to -x0, default corr+=32

plot 2.0k:
  + Ctrl-U and F9 recalculate zoom
  + longer history for undo
  ! smaller bugs fixed

plot 2.0j:
  + option -a added (imperfect way for aspect ratio x:y=1)

show 2.0j:
  * option -v changed to -_
  + option -v added (verbosity level)

cook 3.1j:
  ! the 1-4 control bug mentioned in 3.1g was not fixed in the P1 version
    now fixed
evu.c:
  ! bug fixed: cbrt did not consider units

cook 3.1i:
  + barostat in specified direction includes the respective box in
    statistics (formerly, +Lx,+Ly,+Lz had to be given in .cpi file, which
    can be easily forgotten)

cook 3.1h:
  ! tau.T of Langevin thermostat was wrong (correlation time of velocity of
    ideal gas) so that the correlation time of temperature was tau.T/2.
    Fixed - tau.T is now the correlation time of temperature

cook 3.1g:
  ! linked-cell list was wrong for fully flexible models because of a
    new bug in 1-4 control
  ! metal force field contained incorrectly measurements every step; thus,
    some quantities (as RDF) were multiplied by noint - fixed

plot 2.0i:
  ! parameters to erase (F8) and exported to PLOTCMD have been accepted
    only if shown, allocation error while PLOTCMD (max 3) -- fixed

show 2.0i:
  + cluster analysis added (hot key F5 or Ctrl-T)
  * old F5 (font) renamed to F6
  + plbcenter upgraded to center a periodically split cluster

plot 2.0h:
  + . in ps.def = EOF

install.sh:
  ! bug fixed: compiler was not written to compiler.sh
    (thus default cc was used without optimization)

cook 3.1f:
  * line beginning by * in data input is now copied as a comment
    (avoids problems with WARNINGS in a ble-file)
  ! autocentered slab was by mistake shifted by 0.5 AA - fixed

cook 3.1e:
  + option -k-1 added: with ANCHOR suppresses writing SIMNAME.anc

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2017-10-24 <<<<<<<<<<<<<<<<<<<<<<<<<<<

blend 2.3e:
  ! bug fixed - determination of EvdW (U=0 problem as before)

cook 3.1d:
  + key P in -l added (dump of Drude amplitudes)

plot 2.0f:
  + FIT="e NUMBER" added

cook 3.1c:
  ! U=0 not properly initialized in some places (see 3.1a) - fixed

cook 3.1b:
  ! interaction energy with Eelst[] was inappropriately added to En.pot
    directly, not En.el - fixed
  * scf.eps<0 options and algorithms reconsidered

cook 3.1a:
  + PERSUM version (suffix ps) added: periodic sum, cutoff > box/2
  ? WARNING: large changes in the code were needed, some sub-versions not
             tested, problems cannot be excluded
    Needed changes: U+=0 instead of U=0 in SS_MEASURE and similar
                    U=0 in declarations of U in site-site functions
    Any call to functions LJ, LJM, LJQQ, ... may be affected!

cook 3.0k:
  ! Codata 2010 constants used for h/k in nm.c,nmc.c changed into 2014
  ! init="crystal" fixed for not cube, with pins=4 format changed

cook 3.0j:
  ! small fixes: scf.c: rpol^2 -> |rpol| fixed in divergence output
                 nmc.c: error reporting improved

blend 2.3d:
  ! with option -H the final inertia tensor in kg m2 was wrong - fixed

blend 2.3c:
  ! 1-4 table wrong with blend -v8


plot 2.0e:
  ! stderr was incorrect for the integral of the fit (env INTEG)
  + env FIT added

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2017-07-10 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 3.0i:
  ! Bug in Ewald Drude pair self-energy term (introduced in V3.0a) fixed.
    Typically, the energy error was ~ several J/mol.
    The trajectory was not affected.

plot 2.0d:
  + option -v added (= `v' in GUI, cf. hotkey v)

cook 3.0g:
  ! slab autocenter (see slab.sp) was not working at all, now OK
  ! ST2 was out of order - fixed
  ! "optimized water models" were turned off
    NB: be aware that cutoff<L/2-1 is needed here

cook 3.0f:
  + q added to ble-file data (1st set) = charge scaling
    auxiliary variables x,y,z,a,b,c,aux,i,j,k,n added, too

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2017-05-17 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 3.0e:
  ! bug introduced in V3.0d fixed

cook 3.0d:
  ! module Ewaldtest unnecessarily called for short cutoff - removed
  * default drmax increased to 0.15

cook 3.0c:
  ! int overflow for large systems fixed if cfg loaded from .plb

show 2.0g:
  ! show H-bonds (color not initialized w/o -H option or color change)

cook 3.0b:
  ! bug fixed (wrong CP allocation in the re-read mode)

show 2.0f:
  + H-bonds added to menu, hotkeys h H Ctrl-H Ctrl-C added

cook 3.0a:
  * #ifdef BJERRUM: see bjerrum.c, under development
  * Measurements during simulations / stored configurations unified:
    CHANGED OPTIONS:
    * -m0 = read configurations SIMNAME.0, SIMNAME.1 instead of simulating
       NOT TESTED, CUMBERSOME SIMNAME CONTROL
    * -m1 = read playback instead of simulating
    *  -f = calculate forces (after reading or in addition)
       OLD MEANING OF -f REMOVED, use reread.by
    REMOVED OPTION:
    - -g (remedy: rename files and use symlinks instead of the lost functionality)
  + variables reread.from, reread.to, reread.by added, to control of reading
    frames with options -m1,-m0
  * CLUSTER: input data are now in SIMNAME.cli instead of end of files
    .def,.get
  + cl.mode added (1=on, 2=cluster, 4=configurations)
  + cl.format cl.maxn cl.maxcluster duplicate the respective keywords
    in SIMNAME.cli
  + Cross-section of clusters added
  + diffusion and conductivity changed, variable diff.mode added
  ? Bug in self energy created, fixed in V3.0i

plb2diff 3.0
  * changed to be compatible with cook V3.0a

show 2.0d:
  * -X,-Y changed to -P#,# (because -Y blocked yellow)

cook 2.9f:
  + Gear integrator development code (gear.init, gear.P[], gear.C[])
  ! Urey-Bradley fixed
  ! option -l fixed and extended
  + center.K unified with FREEBC


show 2.0b:
plot 2.0b:
  ! small bugs fixed in the new GUI

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2017-01-31 <<<<<<<<<<<<<<<<<<<<<<<<<<<
December 2016:
GUI based on gtk removed, in-house GUI added
environment variable GUIPATH removed
environment variable GUI added, containing letters:
  g = start blend with menu
  s = start show with menu
  p = start plot with menu
  K = enable "kill all" in plot

plot 2.0a:
  + in-house GUI added, small changes, units in ps.def
show 2.0a:
  + in-house GUI added, small changes in hot keys and buttons
blend 2.3a:
  + in-house GUI added, small changes

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2016-08-05 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.9e:
  ! fully flexible models did not work under linked-cell list because
    the 1-4 predictor was turned off and the bonded forces were not
    calculated - fixed

cook 2.9d:
  + D/r^8 added to exp-6, #define POW used instead of REP12

cook 2.9c:
  * Angstrom is now denoted AA not to confuse with A=Ampere.
    It applies to info and file names; eg., SIMNAME.cm.A-3.z -> SIMNAME.cm.AA-3.z

cook 2.9b:
  + stacking correction, see slab.ext.zero, slab.ext.center,slab.ext.span

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2016-08-05 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.9a:
  * normal modes improved
  ! small fixes

plot:
  + PLOTCMD added

blend 2.2i:
  * some defaults and file names changed for for normal mode vibrations
    (and essential dynamics - untested)

cook 2.8i:
  ! bug fixed: t was ignored if init<=2 - now checked and noted
  ! bug "7 init<-1 wrong - loads incorrect box size" cannot be reproduced, seems to be OK

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2016-03-14 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.8h:
  + Fourier-expansion slab cutoff corrections (slab.K, slab.range) added
  ! cookfree, configure.sh incompatibility fixed

cook 2.8f:
  ! bug cutoff electrostatic (in V2.8e only) fixed

blend 2.2h:
  ! gyration tensor was incorrectly named inertia tensor - fixed
  + inertia tensor added, options renamed (see -I -G -S)

cook 2.9b:
  + Surface tension: Slab stacking corrected, see variables slab.ext.zero,
    slab.ext.center, slab.ext.span

cook 2.9a:
  + Fourier transform slab cutoff correction, see variables slab.K, slab.range

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2015-12-17 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.8e,f:
  ! BUG FIXED: bad Hamiltonian for thermostat="NPT", isotropic case
    (no bug in the trajectory)
  + control of volume, V(t), using external file SIMNAME.box, t added

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2015-11-15 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.8d:
  * some prints cleaned (no dielectric constant if ions)
  + partial currents by ion types added

cook 2.8c:
  * default PRESSURETENSOR changed to 3
  * unit of Wcl in .cp changed to Pa (as statistics), was K/A^2
  * Ein as the 4th column of .cp for NVT removed, rho applies unless
    fitting of sigma, virtual volume change...

cook 2.8b:
  * WARNING "cannot determine smooth cutoff" relaxed, both energy and
    forces beyond C1 are now smoothed, although inconsistent.
    The energy/forces checker reports problems, which are WARNINGs only.
  - No.lcth removed (always the same # of threads for r- and k-space
    with PARALLEL=1)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2015-09-17 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.8a:
  + Support of different energy units in the ble-files (not in the
    ble-files generated by blend). This solves the following caveat:
    ! Some force fields (e.g., Buckingham, see MACSIMUS/sim/buck)
    ! had additional parameters (C and D) in kcal/mol, however,
    ! the same parameter in the table nbfixes must have been given in
    ! internal units (Kelvin, or R/kcal=0.0019872041), i.e., as printed
    ! when cook -v7, or multiplied by kcal/R = 503.219565.
    The new scheme defines the units in structure Epow_s in
    macsimus/sim/<nonbond>/sitesite.c.  Variable  eunit  defines the
    unit in which energies are given in the ble-file.  eunit is given in
    p.u. (loosely, Kelvin), the default is Eunit/kcal = R/kcal (R=gas
    constant).
    The nbfixes table uses the same unit as the site-site table (no
    exception for parameters C,D for the Buckingham potential).
    FOR FORCE FIELD WITH EXTRA ENERGY PARAMETERS, THE NEW nbfixes
    TABLE IS NOT COMPATIBLE WITH THE OLD ONE!
  ? function finite() to detect numeric problems was replaced by isfinite(),
    because finite() did not detect nan in some cases when optimizing on.
    It is assumed and verified that isfinite(0) returns 1,  although the
    man page says it should return 0, which does not make sense.

cook 2.7v:
  ! bug fixed - automatic -a0 (invalid for Gaussian charges) removed from
    the back calculation of diffusivity; note to plb2diff added
  + synopsis slightly changed (cook* SYSNAME.ble SIMNAME.get allowed)
  * Busing-style of the exp-6 potential extended to Gaussian charges
    (good for Baranyai-Kiss water and ions)

cook 2.7u:
  + switch SS_DEBUG changed to variable poteps
  + adjusting size potential parameter extended to cross-terms (fia nbfixes)
  * constants updated for CODATA 2010

cook 2.7t:
  ! reported energy cutoff corrections in the slab version
  * aesthetical changes of the cleaving algorithm, units changed

cook 2.7s:
  + Zone melting ensemble (NLxLyPzzE), see tau.L, box.Lx[], box.Ly[]

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2015-01-08 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7q,r:
  ! Gaussian charges for ions - bugs fixed

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-06-28 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7o,p:
  ! small change/fix of X-section

cook 2.7m:
  + Gaussian charges added (switch GAUSSIANCHARGES, tested for COS)
  + qq-based r-space Ewakd tables added (switch QQTAB)
  + eps14 added (should be improved in future)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2013-08-28 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7l:
  + Tstop added (stop if temperature reached)
  + keywords Ekin added to .cpi

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2013-04-03 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7k:
  * Default el.rshift changed to 3, warnings improved
  ! Bug fixed: normal mode vibrations of constraint models were wrong for
    certain combinations of masses because of numeric underflow in the matrix
    inversion procedure

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-12-11 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7j:
  + Cross-section extended to the whole configuration
  + Normal modes with constraints now may call octave to calculate
    eigenvalues - faster, but more memory consuming.
    New input variable: nm.method (see the manual)

cook 2.7h:
  ! The diagonal components of the pressure tensor for the dipolar Ewald term
    were wrong (its trace and thus the isotropic pressure were correct).
    The error was zero for tinfoil.
    Surface tension with Berkowitz correction was affected, but the
    correction is negligible in any practical setup. The same holds for
    optimizing crystal shapes
  ! POLAR saturation autoset: ref. dipole moment includes the induced part
    (bug: was permanent)

cook 2.7g:
  + Support for automatic determination of el.epsinf and Eelst to
    reach given saturation: new variables sat and tau.sat
    If Eelst=0, el.epsinf is adjusted:
       use keyword "xinf" in SIMNAME.cpi to monitor 1/(2*epsinf+1)
    If Eelst[2]!=0, Eelst[2] is adjusted (must be positive):
       use keyword "Ez" in SIMNAME.cpi to monitor
  + tau.sat<0 will copy the average to the controlling variable
    and start init="append" run

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-06-28 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7f:
  * el.sf default changed to 0 even if option -f
  * el.epsinf<0 no longer means the Berkowitz slab correction - see el.slab
  + el.slab: the Berkowitz slab correction
  - option -\\ removed, use environment NSLOTS and variable No.lcth
  ! no distance check as the default (-d0, was incorrectly -d1)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-06-20 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7e:
  * An attempt was made to improve performance of the linked-cell method,
    esp. if parallel, using different organization of data.
    See compile-time switch LINKCELL.
    The results are very architecture and data dependent.

cook 2.7d:
  * precision loss due to q^2 constant term and Ewald correction of Drude
    charges further improved (see exacterud_sqrt_1())
    Still there is error ~<100 kPa in the pressure tensor (different
    versions may differ because summing contributions in different orders)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-06-01 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7c:
  + el.rshift added (shift erfc splines to avoid jumps, test print)
    el.rshift=3 recommended for normal modes
  + el.rplus added (instead of compile-time switch ERFCPLUS)
  * (in some POLAR versions) precision improved by NOT summing large
     q+q- erfc terms to the alpha/sqrt(pi) sum q^2 constant term
     The error <1K/molecule or 50 kPa in pressure was thus reduced several times

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-05-25 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7b:
  * option -x (optimized code, LJ terms for some water models) changed,
    warnings added - less automatic, but more fool-proof
  * equalize allowed in .def only (ignored if in .get)
  + normal mode vibrations (BUG: not for general constraint models with
    variable angles between fixed bonds, good enough for rigid models)
    New variables: nm.ampl, nm.dr, nm.eps, nm.frames, nm.modes, nm.zero
    See the manual!

Utilities cfg2asc, asc2cfg added (for editing the configurations by hand)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2012-03-31 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.7a:
  + MTK thermostat+barostat
  * format of .cfg files changed, utility cfgconv provided for conversion
  + fundamental frequencies (for vibrating bonds only, to be extended)
    variables nm.dr, nm.eps, nm.frames
  + cutoff correction control (variable corr) extended
  + statistics and convergence profiles of enthalpy, pot+kin energy
  - load.reversed (support of reversed endian) removed

blend 2.2f:
  + dump screen a bit changed
  + -w80 now writes a configuration in the (V2.7a) format of cook

ice.c etc.:
  + ices II,III,V,VI,VII

show 1.1f:
  ! serious bug fixed in showing H-bonds
  * small changes

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-11-11 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6k:
  ! bug fixed (reset of Nose variable log(s) if the thermostat changes)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-10-26 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6j:
  + variable key added, to be used to execute commands and (later) also
    replace quit and sort
    e.g., key="cp" (without ending ;) will pop up convergence profiles
  + three versions of cook added to the basic installation

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-10-18 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6i:
  + variable el.Perr added (to control WARNING in the case of pressure tensor trace/virial discrepancy)
  * variables densprof.* changed to slab.*
  * however, densprof.slab -> slab.mode
  + slab-based thermostat: slab.T slab.Tz0 slab.Tz1
  * slab-based forces in the z-direction moved from center.* to slab.*
  + non-Nose thermostat forces Nose variables:=0
    (NOT to remember these variables if other thermostat was run meanwhile)
  + 3D structure factor added, el.sf=1 = radial SF, el.sf=3 = 3D SF
  * format and name of the radial SF changed
  + corr extended:  1=cutoff corrections included in final results, not in P for the barostat
                    2=cutoff corrections included (also in barostat)
    the default is now corr=2, which the same as before

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-07-13 <<<<<<<<<<<<<<<<<<<<<<<<<<<
plb2diff:
  + fit of MSD(t) to a+b+c/t^1/2 (w. hydrodynamic tail) now optional
    with PLB2DIFF=3 (old linear fit = PLB2DIFF=2 = default)

cutprt:
  ! bug fixed (new table not recognized - some output omitted)

staprt:
  + -kr added (relative standard error)

cook 2.6h:
  + rho printed with standard error estimate
  * option -[ removed, use data:
    load.n[] = replicate factors
    load.L[] = change L option
    load.N   = change N option
    load.tr  = load transformation
  * option -] removed - use variable equalize instead (default = no equalization = 0)
    CAVEAT: only some variables allowed to calculate equalize, e.g.:
            allowed:  noint=5 h=.01/noint equalize=0.8*(h>0.0015)
            error (NOT REPORTED): equalize=cleave.K>10
    (see the manual or code for details)
  * option -@ changed to -@WHO@WHERE (send e-mail if job finished)
    old -@ available as  No.occup

plot:
  * mouse changed: left=digitize (no doubleclick needed)
                   right=place file info
  + ctrl-a,b.. = assign 0 to parms

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-05-11 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6g:
  * harmonics and full correlation function of water (old feature) revisited
  + el.centroid added (default=0, old default=1):
    1 = old default = dipole moments of ions are w.r.t. molecule charge centroid
        problem: total dipole moment is not a sum of ion moment
    0 = dipole moments of ions are w.r.t. coord. center
        total dipole moment = sum

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-04-22 <<<<<<<<<<<<<<<<<<<<<<<<<<<
show 2.0f:
  + dump EPS added (hot key E)
  * old duplicate hot key E removed (use tabulator to rotate by 90 deg)
    Hint: this is the tabulator in bash: $'\t'
    Example (show slab horizontally):
      show -I%$'\t' molfile slab.plb
  + environment variable SHOWPAL added - number of palette entries
    (default = 249, practical minimum is about 100)
    Example (slow connection):
      SHOWPAL=66 show molecule
  + oraNge (browN) color added, used for marking, -N<PATTERN> in recoloring

cook 2.6f:
  + fluctuating charge FQ4 model
    o direct and LINKCELL supported
    o BUG: FQ4--DRUDE interaction not yet
    o CAVEAT: not tested with POLAR&4
  * some variables renamed:
    epsp -> scf.eps
    omegap -> scf.omega
    epspx -> scf.epsx
    epspq -> scf.epsq
    maxitp -> scf.maxit
    new variable: scf.omegax (relaxation parameter for virtual volume
                  change method)
    negative scf.eps, scf.epsx: abs. value determines the number of
                                iterations (scf.maxit irrelevant)
  * output names of box and dipole moments changed
  * formula for the dielectric constant improved (see the manual)
  * legacy code allows loading old names and dielectric constant
    recalculation
  * change of defaults: el.epsinf=1 for cutoff electrostatics

  several bugs fixed by T. Trnka:
  ! -v logic in blend
  ! slab geometry problems: auto set the number of degrees of freedom with z-forces
  ! init cfg with z-forces and with shell-core model (crash - term missing)
  ! z-profile if not autocenter (crash - undef. variable)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-02-22 <<<<<<<<<<<<<<<<<<<<<<<<<<<
statics.c (cook, autocorr, showcp):
  * In the statistics module, the "automatic" determination of error
    estimates was improved.  It is now uniformly good even for cases
    like long negative corelations (e.g., Ptzz for liquid slab+water).
    There is a certain tendency to underestimate the error, though.
    The new formula takes a weighted average over several blocked
    estimates, both without c_1 and with it. (The old version took a
    maximum, which in case of negative correlations overestimated the
    error, however, is more prone against underestimation.)
    The default blocking was increased to 2^15.
Several bugs fixed thanks to Tomas Trnka:
blend 2.2d:
  ! determination of the vdw radius for exp-6 was wrong - fixed
  + verbose output of pair potential parameters
  * -v (verbosity) option changed
cook 2.6e:
  ! load cfg with POLAR<->NONPOLAR conversion and old version was fixed
  ! En.Hz requires PRESSURETENSOR

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-02-17 <<<<<<<<<<<<<<<<<<<<<<<<<<<
show 2.0e
  * -a changed: now -aAN determines max rotation angle as 2*PI/AN,
    the initial rotation angle is initial (PI/8)/AN
    default AN=4
    E alternate for TAB (standard orientations)
cook 2.6d: (wrong version number)
  ! automatic centering bug fixed

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2011-02-01 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6d:
  ! number of degrees of freedom for AndersenCM and MaxwellCM was wrong - fixed
  + simplified Langevin thermostat (canonical in the long tau.T limit)
  + Gibbs interfacial energy by the cleaving method - see the manual
  * density profile measurements improved (as regards z-periodicity)
  + automatic centering of the z-profile (see densprof.sp)
  ! the slab for surface tension calculations (slab corrections) need not
    be centered
  * densprof.zmax=0 means scaled density profile in [0,Lz] divided
    into density.grid
  * dt.plb<0: frame not written for t=0 (init>=2)

show 1.0d:
  + key K (markallmode) added
  + save plb (^S) saves the marked subset (if SELECTION)
  * I = invert marking, TAB = reset (orientation)
  ! bug fixed (buffer overflow, discovered by Radek Vostal)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-11-22 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6b:
  ! caveat fixed: now can read playback and then write it again
  + name of input playback accepted (argument PLBNAME.plb; if missing,
    SIMNAME.plb is used), can be further used even if further
    processed (option -[)
blend 2.2b:
  + GUI support, hot key Ctrl-S (save) added

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-10-18 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.6a, blend 2.2a:
  * new simplified interface to implement potentials and combining rules:
    see MACSIMUS/sim/ssdecl.h and example MACSIMUS/sim/lj/
  + Buckingham potential extended
  + The van der Waals radii (exported via GOL-file to show) are now
    calculated because the Buckingham parameters are not directly
    related to the atom size.  If the atom-atom potential does not
    have a minimum, R=r1*1.12, where U(r1)=1 kcal/mol, is used instead.
    (This radius is then rescaled by option -p).
  * parameter sqrt_rule changed to comb_rule

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-10-07 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.5h:
  + Buckingham support added to cook and blend (see sim/buck+12s), various
    combining rules; the D/r^12 term can be turned off by switch -DNO12REP
  + as above - the shell-core polarizable version
  ! wrong masses in some parts of the algorithm with mass equalization -]
    (probably only some normalization was imprecise)
show 1.1b:
  ! numbering improved, plb can shring during viewing, better looking box..

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-09-29 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.5g:
  ! bug fixed (POLAR: -p control was wrong)
  ! POLAR=1 off-diagonal pressure tensor added
    off-diagonal pressure tensor symmetry tested for off-plane dependants and POLAR=1

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-09-08 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.5f:
  ! bug fixed (could not read files > 2GiB)
  ! bug fixed (surface tension output)
  + Nose-Hover for Verlet+SHAKE implemented via velocity predictor
    switch SHAKEVELOCITY renamed to VERLET
  * options -_ and -p changed:
    -pKPC (dec. digits):
      K = k (predictor length, parameter of ASPC or similar)
      P = polar method (order of the method):
          0=Car-Parrinello-like (0th order, no extra thermostating)
          1=no prediction
          2=ASPC
          3,4=higher-order (with Gear only)
      C = velocity predictor length, Verlet+SHAKE (-m2) + Nose-Hover only
          (default C=2 => energy drift O(h^5))
          constraint predictor order, Gear only (-m#, #>2)
          (default = -m value)
     -_ = inactive now - changed into P above
  + support for out-of-plane dependants (as Lone pair in some water models)
    incl. the pressure tensor
    Not available via blend!

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-06-01 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.5e:
  + diffusivity now also for NPT (cubic box), uses L from plb
  * epslc replaced by box.rmin, code changed to shift normalization

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-05-21 <<<<<<<<<<<<<<<<<<<<<<<<<<<
show 1.1a:
  gtk-based GUI by Karel Matas added:
  - should be compiled with -DGUIMENU
  - runtime environment variable GUIPATH should point to macsimus/guimenu

cook 2.5d:
  ! several bugs fixed by Tomas Trnka
  ! cutoff electrostatics bug of 2.5c fixed
  ! bug in setting atomatom2() fixed

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-04-14 <<<<<<<<<<<<<<<<<<<<<<<<<<<
blend 2.1l:
  + 1-4 interactions with POLAR implemented: all charge-charge terms
    are multiplied by factor14 (as if diel.const.=1/factor14)
  ! shell-core (repulsive antipolarization) changed:
    polar=2 removed from .par file (while polar=4 still means ADIM)
  + table "shellrep" added in par-files.
    The shell-core term is the interaction between (A) a polarizable
    atom and (B) repelling atom.
    (A) is usually anion with a large polarizability, it has nonzero
    kappa (arep) in the table "polaratoms"
    (B) is usually cation with none or small polarizability. These
    cations are now listed in table "shellrep".
    WARNING: not compatible with older versions - you must add table
    shellrep to such force fields!
  + 1-4 polarizable interactions (scaled by factor14) are added to the polar
    versions. The shell-core term is ignored for 1-4
  ! infinite loop w. -b0/-a0 fixed
  + -q also suppresses warning "fractional charge"
    hint: use -q100 to suppress the warning and keep charges unchanged

cook 2.5c(?):
  + bit optimized for monoatomic molecules and no LINKCELL
  * verbose output set in case of POLAR bad iterations if this happens
    for the second time
  * POLAR: option ^ changed (-^+EEEEOO: +EEEE=shell, OO=omegap in %)
  + POLAR: 1-4 polarizable terms added
    supported: FREE, cutoff elst, Ewald; LINKCELL, shell-core (POLAR&1)
    suppoted but not tested: ADIM (POLAR&64)
  - POLAR&32 removed (cf. table "shellrep" in the par-file)

show 1.0d:
  + dihedral angle info printed if 4 atoms clicked

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-03-11 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.5b:
  + another warning for a small non-rigid molecule (forgotten -h in blend)
    pressure tensor double checked for POLAR&1
  ! the suggested timestep was calculated from the initial masses even if
    option -] which might have lead to a misleading info - fixed
  ! memory leak for PARALLEL=1+Ewald fixed

blend 2.1k:
  * options -a0, -b0 improved: minimization with constraints is now smooth
  + constrained dihedrals added. They should be written in file SPECIES.jet as:
    IDi IDj IDk IDl ANGLE
    the angle is in degrees in interval (0,180)
    BUGs : cos(angle) is used internally, 0 and 180 are not allowed and
           the sign of ANGLE is not distinguished

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-03-02 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.5a:
  + code (re)parallelized using threads
    PARALLEL=1: linked-cell list parallelized
    PARALLEL=2: ewald k-space runs in parallel with pair forces (r-space)
    example (SGE):
      qsub -cwd -b y -pe shm 2 ./cookewslcp0P2 polwater slab -t
  - switch CLOCK removed
  * time measurement (-t) upgraded
all:
  * option -DSCR removed from modules in show, blend and util
    (#include in che-files still works)
    it is left for cook because it enables $iFILE in input data
molcfg:
  ! problems with molcfg called from cook again...fixed
plot:
  * wrong input data treated as NaN and not plotted

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-02-16 <<<<<<<<<<<<<<<<<<<<<<<<<<<
blend 2.1j:
  * -h no longer automatic with registered waters (warning printed)
cook 2.4p:
  + info/warning added for inappropriate timestep h (based on min. mass)
plbreplicate:
  ! fixed

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-02-08 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.4o:
  * init=4="slab" make a slab (cheap random with z-center bias)
  + check of a registered water model
blend 2.1i:
  + variable "water" exported in the ble-file (=registered water model)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2010-02-06 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.4n:
  ! bug (wrong pressure tensor components in cook 2.4m) fixed

>>>>>>>>>>>>>>>>>>>> Releases 2010-01-05, 2010-01-22 <<<<<<<<<<<<<<<<<<<<<
WARNING: bug in pressure tensor components

show 1.0c:
  + editing of a configuration added, toggle mode by m (the same as blend)
  + save one frame by ctrl-s
  + unmark by M or ctrl-u (compatibility with blend)
  * hot keys m M (match configuration) renamed to ' "
  + if the mol-file does not match the plb-file, the mol (and gol) file info
    is either repeated or truncated
  + if mol/gol file is missing, just spheres are used
  * DOS support finally removed
    (show 1.0b/DOS frozen as show4dos.c and showidos.c)
blend 2.1h:
  * `x' no longer allowed as the character for crossing bonds in a che-file,
    use `+' instead (`x' interferes with the `:ID' constrct)
  + blend now prints a "soft warning" if the par-file contains "void" data
    lines which do not start with a table name (like if an accidental Enter
    appears within the data)
  * hot key M (used to be floodfill) now means unmark all (cf. show)
    (floodfill is midclick - the middle button is now common, so that
    it is not necessary to use a key)
cook 2.4m:
  + option -] added: equalize masses of atoms in molecules
  + option -q-# changed: now a charged system is neutralized by adding
    a constant charge to all charges (old meaning "do not rotate" is now -j-#)
  * option -j-# changed: now it means "do not rotate while initializing"
    (formerly -q-#)
  + variable  initvel  added: reinitialize velocities of initvel molecules
    (the same as former -j-#)
    initvel is reset to 0 after velocities assignment
  * the definition of the pressure tensor was standardized - now without
    factor 1/3
    Note that the reported values were 1/3 of the pressure, which was
    written in the manual, but misleading
  + time autocorrelations functions:
    time scale is always in ps, conversion to SI should include factor 1e-12
    off-diagonal pressure tensor (-> viscosity), if lag.visc
    current (-> conductivity), if lag.cond
    velocity-velocity (-> diffusivity), #ifdef RGYR, if lag.v
      (for molecules to lag.nv, lag.dim coordinates)
  * lag ranamed to lag.err, lag2 to lag.n
  + molcfg not called if -f specified
cp2cp:
  ! bug found (option -d did not work properly)
    options replaced by (more general) -r and -h
showcp:
  * cleaned, some options changed:
    -e (all columns) removed
    -[ -] changed to -e
    -c extended
    -xDT becomes -hDT -x
    -hDT set time step (=noint*h)
  + can print also time autocovariances (-c2)
  + -n added: number of blocking 2^
cook, showcp, autocorr, staprt:
  + now dt recorded and time axis printed
    (may be confused if old/new sta-files are mixed)
plb2diff:
  ! bug fixed (int overflow if too long plb-file and too many blocks)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-11-30 <<<<<<<<<<<<<<<<<<<<<<<<<<<

blend 2.1g:
  + option -g-# : no graphics, minimization protocol printed
    by # steps (default = guess)
show 1.0b:
  + show hydrogen bonds. Example:
      show iceVMq3288 -DH4I -AO4I -HB2.4
    (donor site type=H4I  acceptor=O4I  distance limit=2.4 A  color=Blue)
  - option -_ removed (not clear what it was good for)
  * bond allocation was made dynamic
    (option -i is now the initial number of bonds - may be removed in future)
  + box around (-\\ = -|) can be now any color (e.g., -\\B)
cook 2.4l:
  unchanged (no bug in Ptens)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-11-18 <<<<<<<<<<<<<<<<<<<<<<<<<<<
  * getline() renamed to mygetline() because of clash with a new GNU library
cook 2.4l:
  + el.diff added
  * tau.rho now forces rho in column 4 of SIMNAME.cp (the same as tau.P)
  * info plb: ... moved to -v4 verbosity level
  * (somewhere around this version)
    statistics not automatically calculated for variables in .cpi
    (can be returned back by + prepended, e.g., "+Lz")
blend 2.1f:
  + option -w80 added (make .cfg for cook)
plbcut:
  ! serious bug fixed in array allocation
show 1.0a:
  + option -J added (more flexible "show shell"). Example:
      show -j6 -Jb xxx
    will show waters (= molecules with 'b' in their id's) around the 1st
    molecule
    Note: 'b' appears if molcfg is automatically called from cook and
    water is the second molecule (N[1])
hbonds:
  + extended to allow a solute

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-09-18 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.4k:
  + variable corr added (whether LJ cutoff corrections are to be included)
  ! severe bug fixed: LJ without charges in nonpolar versions had bad
    cutoff and the molecules "exploded". This bug was introduced while
    "cleaning" the code, several versions affected.
  * the 1-4 predictor in the LINKCELL version is now safer and prints
    more helpful warning messages
  + Andersen and Maxwell thermostats working on centers of mass
    they are safe with constrained dynamics
  * some renaming: "intermolecular" degrees of freedom/kin. energy/temperature
    are now called translational (calculated from the center of mass)
    "Tmol" in .cp is renamed to "Ttr"
    "Tkin inter" (in .sta) renamed consistently to "Ttr"
    "Tkin intra" (in .sta) renamed consistently to "Tin"
    "T" (in .sta and .cp) renamed to "Tkin", to be distinguished from
      the thermostat value Tkin
    "Tinter_intra" renamed to "T_tr_in"
  * the default rescale was changed to 15
  + examples/sample.def: file with the most common variables and options
show:
  options added:
    -[# frames moved in keys []; -[-# the same in %
    -]0 MOLNAME.gol will be ignored
  filename argument logic cleaned:
    - 1st arg can be MOLNAME or MOLNAME.mol or MOLNAME.gol or MOLNAME.plb
    - optional 2nd arg must be with extension, preferably PLBNAME.plb
      (.plb is then stripped off in derived names)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-09-09 <<<<<<<<<<<<<<<<<<<<<<<<<<<
cook 2.4j:
  ! In the linked-cell+polar versions, the RDF contributions were calculated
    every step instead of every cycle. The resulting RDFs were thus noint
    times larger and the code somehow slower.
  ! Wrong number of inter/intramolecular degrees of freedom (with Berendsen
    thermostat and NVE) - fixed.
    (This bug in versions 2.4k and older introduced an inaccuracy of the
    order of 1/N to Tkin inter and Tkin intra. The correct values can be
    easily obtained by multiplying by wrong_degrees/good_degrees)
  * -k option was changed and manual updated
blend 2.1e:
  + hot key @m writes file MOLNAME.mark with marked sites
  + hot key @k writes file MOLNAME.keep with kept (frozen) sites
  + -k extended (-k4 reads FILE.keep, -k8 reads FILE.mark)
mol2mol:
  this utility can provide various operations with mol, gol, and plb files
  It can also read MOLNAME.mark etc mentioned above
show:
  ! bad 'L' (write plb) if new format - fixed
  + match extended to a selection of sites (cf. blend above)

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-07-30 <<<<<<<<<<<<<<<<<<<<<<<<<<<
showcp:
  + Braille subset of UTF-8 used to print a better pseudograph
    can be turned on/off by option -u (former -u renamed to -n)
  * the y-range of plot is now within min:max of the original data, not
    of the blocked data (=the same behavior as the pseudographs)
plot:
  * number of variables available for fitting increased to 10
  * some hot keys had to be changed:
    i: initial scaling is now U (or k)
    I: fit is now:
       ctrl-t in double,
       t in long double (recommended),
       T in highprec
    gG: show/hide grid is now zZ or oO

    Example of fitting:
    ^^^^^^^^^^^^^^^^^^^
    tab 0 1 0.01 | \
      tabproc "A" "exp(A)+rnd(1)*0.1" 0.1 | \
      plot -:A:B:-:C :A:"a+b*A+c*A^2"
    Then, in the plot window, type `t'.
    It should compile a fitting program and plot a fit.
    The result is around a=1.0, b=0.8, c=0.8.

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-07-22 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 2.4i:
  ! malformed error message (rescaled too much) fixed
  + bulkmodulus added (for NPT and similar; if not set, ideal gas value used)

most software; e.g., showcp and ev:
  * terminal sizes are now determined from environment variables LINES and
    COLUMNS, not SCROLL
    Hint: in bash, add to your .bashrc:
      shopt -s checkwinsize
      export LINES
      export COLUMNS

blend 2.1d:
  ! temporary file now created using mkstemp(), which is save if run
    concurrently (on several nodes). (The price paid is that you must
    by hand remove temporary files if blend crashes.)
  ! format for constrained angles fixed (was slightly less accurate)
  ! blank lines allowed between header and parameter_set in a che-file
  ! undefined chirality is a warning only

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-06-11 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 2.4h:
  ! -[-XXX (allow box change without rescaling while load) does not work
    FIXED (thanks to T. Trnka)
  ! rhoinit does not work - FIXED
  ! small bug fixed: report of energy/particle
  * padding spaces (to 4 chars) removed from names listed in SIMNAME.cpi
    (e.g., Lx)

blend 2.1c:
  * second virial coefficient improved

pdb V1.4b:
  ! incompatibility fixed - alternate locations can be now also in order
    AAABBB, not only ABABAB as before. Other orders (BA, etc.) are warning.

show:
  * init size according to box if available FIXED
  ! bug (something like box around for old .plb) FIXED

>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-05-18 <<<<<<<<<<<<<<<<<<<<<<<<<<<

tabproc, plot:
  * 1000 columns now allowed (used to be up to Z=26)
    columns over Z=26 are accessible as #27 or c27, etc.

blend 2.1b:
  + second virial coefficient (options -V, -T, -R) added

cook 2.4g:
  * writing of playback files rehacked
    IMPORTANT CHANGE:
      Option -y-1 now means "write one molecule", NOT write the whole
      configuration. Do not use any option to write the full playback.
    Changes of -y:
      -yBIG (now the default!): write playback and call molcfg to prepare
            data for show
      -y##: write ## molecules only (into single SIMNAME.plb, not .p##),
            incl. a call to molcfg
      -y0:  do not write playback
      -y-##: as above, but do not call molcfg
    Input data changes:
      dt.plb=0: write at start and when a sweep (; in the data) is finished

  * normalizing drift and degrees of freedom cleaned:
    - variable `norm' removed
    + variable `drift' added
    + variable `conserved' added


>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-04-07 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 2.4f:
  + option -w extended to -w#, #>1:
    The same as -w1 and the available disk space is checked before selected
    files are written.  If not enough space is available, it waits several
    minutes (more for the second time) and then checks again.  The available
    space is compared to a rather pessimistic estimate of space needed plus
    # kB added (to be even more pessimistic). -w2 should be enough unless
    your co-workers often fill the disk in a very greedy way.  The algorithm
    is not 100% reliable, though.  It relies on a system call to df; if this
    fails for any reason, it is considered as no space and the program keeps
    waiting.  A warning is printed after space has appeared again; for
    apparent reasons, nothing is printed when the no-space condition is
    encountered.

  FIXED BUGS:
  ! tau.E+thermostat should be error - FIXED (warning+ thermostat off)
  ! something's wrong with reporting rho if tau.rho: after switching to
    tau.rho, the target rho is reported, but not actually reached
    (V there - not calculated) - FOUND, V was not correctly calculated
    (might have affected pressure...)
  ! xxx="xxx" in get data not error if key not defined - FIXED

  NEW:
  + plb2rdf, utility to calculate RDF from plb


>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-04-07 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 2.4e:
  + The Ewald error estimates updated for non-cube; see also the BUGS section
  * some electrostatic (Ewald) parameters renamed:
      alpha    -> el.alpha
      epsk     -> el.epsk
      epsr     -> el.epsr
      epsinf   -> el.epsinf
      minqq    -> el.minqq
      erfcgrid -> el.grid
      etest    -> el.test
  - parameter K removed (see el.kappa below)
  + parameters added:
      el.kappa: k-space cutoffs are K[] = el.kappa*L[]
                (old K was in the z-direction: old K = el.kappa*L[2])
      el.diag: include diagonal correction (good for salts),
               formerly available via #define INCLUDEDIAG


>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-02-20 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 2.4d:
  + The linked-cell list implemented now also for POLAR, tested with POLAR=0
  + Exclusions + EXACTERU (COULOMB=-2) implemented now with the link-cell:
    r-space errors caused by splines for Ewald summation of models with
    partial charges are now smaller


>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-02-10 <<<<<<<<<<<<<<<<<<<<<<<<<<<

cook 2.4c:
  The logic controlling box size and rescaling was cleaned.
  * The "Reference box" is given by density rho and box L[0], L[1], L[2] in
      the input data:
    - if nonzero rho is given and some of L[i] are not given (or are zero),
      the remaining L[i] are calculated to reach given rho
    - particularly, if only rho is given and no L, a cubic box with
      calculated size (L[0]=L[1]=L[2]) is used
    - if either rho==0 or rescale&7==0 (i.e., no rescaling is allowed),
      L[i] are not changed; if any of them iz zero, this is error
    This "Reference Box" is used for:
    - setting cutoff (if not given explicitly)
    - setting Ewald parameters alpha,K from errors epsk,epsr (unless they
      are given explicitly)
  * The "Actual Box" is the box size during simulation. It may differ from
      the "Defined box" because:
    - such a box is read from previous run (init=0,1,2), initialized using
      initrho (init=3,4,5), read from ascii (init=10,11,12), read from
      plb-file (init<0), etc.
    - an NPT ensemble is used (or its variant with a component of the
      pressure tensor) which leads to variable L during simulation.
      This is set by nonzero tau.P
    - If nonzero tau.rho (but not tau.P) is specified, the system rescales
      the box gradually to reach "Defined Box"
    - If none of tau.P,tau.rho is specified, the "Defined Box" may differ
      from the actual box forever; this is not recommended (and warnings are
      printed)
  * Option -[-NNN forces (some) components of the Actual Box to be replaced
    by the "Define Box" values

  Options change:
    -[-NNN changed - see the manual
    -@ now reserved for LINKCELL only, removed from POLAR
    -_ now reserved for ASPC (parameter k) with Verlet+SHAKE only
    -w2 (write if cfg has changed) no longer supported,
        default = -w = always write
        (The reason is that it is difficult to trace all changes that a
        user may do with the configuration and determine whether the
        configuration has changed; a cfg may be restored from a backup
        SIMNAME.cfg~ if necessary.)

NEW and updated playback-related utilities:
  + plbstack - stack configurations at the level of plb-files
  + plb2cfg - convert plb to cfg (loadable by cook)
  + cfg2plb - convert cfg to plb (if you have forgotten -y-1)
  + plbreplicate - replicate cell at the level of plb-file
  * plbbox - change box
  * plbinfo - also reports prints all plb-related utilities


>>>>>>>>>>>>>>>>>>>>>>>>>>> Release 2009-01-15 <<<<<<<<<<<<<<<<<<<<<<<<<<<

blend 2.1a, pdb 1.4a:
  + new format (variable L) of .plb implemented
  * .plb made the default format, .3db (and .3dt) deprecated
    (.3db and .3dt can be still read/written on demand)
  * blend writes also .gol file as the default
    IMPORTANT NOTE: use distinct names for cook!
  + suboptions -y4, -y8, -y16 implemented (see manual)


See blend/new and sim/new for the detailed history until 2008
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

>>>>>>>>>>>>>>>>>>>>>>>>>>> Brief older history <<<<<<<<<<<<<<<<<<<<<<<<<<<

>>>>> 2008
Box scaling, pressure tensor (2.4a)

>>>>> 2006
Widom insertion particle method

>>>>> 2005
Slab geometry, surface tension

>>>>> 2005
Rectangular box (not cube)

>>>>> 2004
ANCHOR, WALL
* ICT Prague

>>>>> 2001
CHARMM22, GROMOS96 and better manual -- version number set to 2.0a.
* Evanston, Prague

>>>>> 2000
Cluster analysis, conductivity
* Evanston, Prague

>>>>> 6-12/1999
New and extended polar version, Busing force field, axial polarizability
* Evanston, Prague

>>>>> 6/1997
MPI parallelization

>>>>> 7/1996
Better control over end-to-end distance, cross section, radius of gyration
* Evanston

>>>>> 11/1995
Playback and batch control improved (option -#)
tcfg tprt changed into dt.prt dt.plb
dt.plb instead of option -y
Packed convergence profile (and playback soon...)
* Evanston

>>>>> 2-10/1995
Support for shared memory parallel computers (Convex, SGI)
* Odense

>>>>> 7/1995
Decoupled inter/intra-molecular friction thermostat
* Odense

>>>>> 6/1995
Isobaric ensemble (Berendsen style) supported.
Cosinus shear stress to measure viscosity
* Odense

>>>>> 4/1995
Polarizable dipoles
* Odense

>>>>> 1/1995
PEO and COOK merged into 1 package
* Odense

>>>>> 11/1994
Linked-cell list method version, free-boundary conditions version (C1.0)
* Odense

>>>>> 10/1993
PROSIS/COOK Version C0.5
* Odense

>>>>> 3/1993
Easy implementation of different systems (V2.0)
* Institute of Chemical Processes * Prague * Czech Republic

>>>>> 8/1992
Serial version V2.0, parallel version V0.1 (see ppeo.man)
* Odense

>>>>> 7/1992
Compatibility changes to the transputer parallel version (V1.2)
* Odense

>>>>> 5/1992
Bug fixes (V1.1)
* Odense

>>>>> 4/1992
PEO V1.0,
* UTZCHT CSAV * Praha 6 - Suchdol * Czechoslovakia

>>>>> 1/1992
Poly(ethylen oxide)
* Dept.Math. & Computer Sci. * Odense University * Denmark

>>>>> 10/1991
Constraint dynamics (Lagrangian and Hamiltonian formalism)
* Dept.Chem. * Northwestern University * Evanston * Illinois * USA
