CVS log for ccp4/src/amore.f
	Help

Request diff between arbitrary revisions / Display revisions graphically

Initialise IFAIL

Improve spacing in previous fix

Allow 3 digit SOLUT_ numbers

Increase allowed number of fixed translation solutions from 8 to 64 (Bug 2047)

update licence banners

remove multiple SAVE for gfortran

EJD: Multiply memory estimates by 1.1

Corrected typos

DROT, SROT and DFFT clash with OSF mextend math library routines, MRTAB not used - Bug 598.3

use numbers saved in SETSIZE to set MRT - bug 393

update Bessel limits and output -EJD

correction for multiple opening of CLMN0 - bug 283

correct calculation of PZ in DIRAXH to use NX+2 not NX

increase MITAB fudge factor to 200000

add new licence banners

More changes from EJD with modifications by PJX: setting number of columns
from MTZ file in RCARDS, printing of data in TABLNDO.

Update from EJD:
1. Add "IF (XXX.ne.0) close(XXX)" lines to fix running under Suns.
2. Allow user to input LOCKED info with POLAR or EULERian angles.
3. Updated to use correct cell (from appropriate dataset).

close UDI

Small change from EJD: needed for compatibility with new libraries.

print out correction

typo in print statement - thanks to Bart Hazes

Clemens Vonrhein: remove leading tabs.

EJD: make sure files are open before closing them

Substantial changes from Eleanor:
1. Now estimates memory allocation automatically (manual assignment is
   still available).
2. TABFUN: new keyword BTARGET to adjust model B factors.
3. TABFUN: NORMALISE keyword to generate normalised model sf's.
4. CLMN: extended number of Bessel functions used to generate spherical
   harmonics.
5. Fixes to Locked Rotation Function.
6. TRAFUN: can work with non-crystallographic translation.

Add licence banner.

Add licence banner.

trailing spaces take lines beyond 72 chars (CV)

replace equality with diff .ge. test. Dealing with reals and rounding
errors

SAVE not needed as "SAVE" already higher up

ejd "amore - there was a bug ; forgotten what!"

bug in Amore which shows itself only when the search is in a
monoclinic space group (eg C2) ejd

bug in Amore which shows itself only when the search is in a
monoclinic space group (eg C2) ejd

changed error warning to failure to prevent confusion - ejd

Solution for problem when model atoms have OCC=0.00

Commented out statements which compilers complained had no path to them.

Minor change: print out memory allocation before quitting because limits
are too low in tabling.

reducing 2 more errors to warnings. EJD

EJD: Fix to the table output under MR2IC (was incorrectly shifted from that
output under TABFUN, reported by Herb Klei).
Also changed a fatal error to a warning.

Title added for SOLUTIONF lines. Subroutine GMPRD removed as is in modlib.
From EJD

buglet from ejd.

Fix to CONV keyword of FITFUN.

More fixes from Eleanor.

EJD: sort rotation function output on the chosen target.

Minor change (fix?) from EJD.

a fix to the previous fix

input fix from ejd

Replaced STOPs with CCPERR 1s.

Fix for crash when using ROTATE ... LOCK (locked rotation function).

Commented out unused routine CLMNS - there is a broken call to SPHBES
inside which Dave Love says causes problems on some compilers.

EJD: Fix for phased translation function - one line added plus lots of
     tidying up of s/r TRANDO!

Small fix from EJD.

EJD: only take one conformation, and put out spacegroup P1

New amore from Jorge/EJD with some corrections from MDW

a small change to help liz with searching

Improved checking for multiple occurrences of atom type when reading form
factors.

Don't call CELDA if no cell from XYZIN

Made error message more informative.

Yet more changes to RWBROOK.

added new rwbrook stuff.

.

If B is 0 or below then it is set to 20.0.

Add fractional centre of mass output.
Amore format to output TABLING CELL

SFREAD moved to rwbrook.f
extra message

Changes from EJD 3-4-96.

Moved Matrix subr. to library.

Change from EJD + some tidying.

Reinstated ROTATE NSR keyword.
Tidied HKLM error messages.

No. of FIXed molecules allowed for TRAING increased
from 3 to 8. Also, printer output corrected for
TRAFUN PKLIM option.

Change from EJD.

Mainly make SIGF optional in sorting.  It's not clear to me what the
sigmas propagated from there to hklpck0 do if sigf is assigned.

Had missed change for bad fft dimension.
Reinstate loop over label 390 in rcards.
Clarify some stuff in FITING.

Tidy up in various ways, avoiding warnings etc.  Still needs some
attention.  EJD comments on previous change:

  First important change: Fix to inverse FFT - Should have called with
  array dimension NX/2 + 1 instead of NX/2..

  Second important change: Fix search for prime factors in fft

  Third important change: It now uses the Hmax Kmax Lmax and NsynMax to define
  packing constants, so it is less likely to overflow. That meant each
  hklpck or TABLEi had to have an extra record giving these values.

  Fourth change - so far udocumented: you can generate Es for your mnodel and
  read these into a TABLE -

New from EJD

Clarify resolution/grid diagnostic.  END keyword.

Fix for modified atomsf.lib

EJD: all 0kl reflections of TABLE = 0.

HKLM keyword reinstated and associated error messages corrected.

Reject reflexions by testing sigF<=0, not F.

don't lose with NaNs in data

EJD: All `sqrt(abs(' <- `sqrt(' -- bletch.
Fix some error messages.  Missing data stuff.

EJD: check for average B <1, which will cause FFT FPE.

Only first subsidiary keyword of SAMPLE was found.
REFSOL wasn't recognised.

Fix places where input can take 1 or 2 numbers (like RESO) and could use
parser arrays from the last card.

typo in error message

Tidy up error stuff.
Re-do rcards for easier groking of what it expects and consistency; NB will
no reject some formerly-accepted input.
Remove `inconsistent resolution limits???' warning.

typpo

EJD changes to zero arrays etc.  (In addition to library change which should
render them unnecessary.)

SIGF is now compulsory.

LMB: fixed bug in peak search which could give singular
matrix on degenerate ridges: program now just ignores this &
puts the peak on the central grid point

correct xloggraph stuff

EJD fix for problem with polar Sgs pointed out by Sheila Gover

unused vars, line too long etc. from Sheila Gover

bad SAVE

EJD: TMIN calculation wrong for trafun

see MTZLIB

Locked rotation function decalred defunct by EJD

Changes from IJT as:
1. The value of "IND" calculated by the program seems to be wrong.  Using
the values for h,k,l etc I make it:

((-182+139)*280-8+139)*280+8+139 = -3334373, not 10934986!

2. Why was the constant NM changed from 25 to 97?  Surely there are never
more than 24 primitive positions (OK 48 for centrosymmetric).  Presumably
it's counting the 96 non-primitive positions in cubic, but the translation
function should only need primitive positions anyway.

NB seems to mean that previous sorting and/or tabling output can't be
used with new version

IJT: previous patch for RF SDs was wrong.

IJT: compute standard deviation of map properly.

EJD: TABFUN accepts NOTRAN or NOSHIFT equivalently

Close scratch file TEMP2 to allow multiple tables o/p in tabling.
(Failed on some systems with obscure filing errors.)

Some tidying; change conversion of e(i) between real and integer in
sort5

more informative error message

Array size parameter NR in SORTING was wrongly set.
EJD: tidying o/p

EJD:
  Step should equal dangr, BUT he has some mysterious factorisation thing
  which occasionally means step finishes up < dangr:
  Just tracked down a problem where dangr = 2.0 step calculates to 1.875
Think this means rotfun could o/p garbage

Mainly dynamic memory allocation.

.

Re-instated scratch files as scratch.  Considerable tidy-up.
Error messages try to be more explicit about dimensioning problems.
Note still serious problems with illegal parameter aliasing --
don't try to optimise!

taken from york (wher it was amore3.f)

SAVEs

version pulled straight from york

comments

.

uninit vars

unused variables

replace /dynamic/ (which was inconsistently sized) with blank common

merge with ejd

ejd latest

smaller dimensions from IJT which might succeed in reasonable memory

enough mods for aix compilation

.

new from EJD

latest from york

new from york

Richard Dodson:
bug in Cheshire Cell calc fixed
ccp4 style map from the ROTFUN and TRAFUN peaksearchs can be kept by
     having MAPOUT defined
new peaksearch routine
can combine several model TABLEs into one GENErated s-f file with kword
     ROTFUN ..
      GENE .. NMOL n

changes for toolpack

new from york