CVS log for ccp4/src/fft.f
	Help

Request diff between arbitrary revisions / Display revisions graphically

applied Claus flenburg fix

Bug fix for squared values input to FFT resulting in incorrect output (from Clemens) Bugzilla #2339

Add MDW warning that subroutine RPCFTK incompatible with bounds checking

update licence banners

EJD reports cases where fft dies on hkl limits
She suggests adding 1 to HMAX etc
I have done more conservative change of using NINT instead
of INT - this effectively adds 1 in half of cases

Correcting output inconsitent with documentation when
using PATTERSON mode

fftbig.f, please see fftbig.f log for previous history

fft depreciated in favour of fftbig

set correct NSPFFT - bug 677

set correct NSPFFT - bug 677

add new licence banners

changes from Clemens

EJD: 1) More output for rejected reflections, 2) uses cell for appropriate
dataset, 3) removed call to fft10 (as this s/r is broken)
PJX: Commented out FFT3, FFT16, and FFT17 calls (in FFTCTL), they are not
used by anything.

Prevent user selecting PATTERSON and also assigning phases.

some changes from EJD

make DANO option explicit

F1 and F2 might be type 'E'

cv - removeal of save and mcols > 500 - and a couple of other changes!

cv - removeal of save and mcols > 500 - and a couple of other changes!

update comment to go with previous change

test reset to be ge.19, consistent with documentation (CV)

fixes from ejd

EJD: fft for PHASED TRANSLATION. This only worked properly if you set the FFTSPG = P1
( or used fftbig!) I have changed FFT to do that by default.

fft changes from ejd
a) it does a phased translation checking both hands, and
b) it can read in a modified I for calculating a patterson ( essential to
use ECALC output )

K and L are mixed up for the column types

EJD: set the default axis ordering to Z,X,Y for P1, monoclinic space
groups, plus space groups 16, 17 and 18, and to Y,X,Z for all higher
symmetry space groups.

Parameter IDIM increased to 1000000.

Bad format statments for g77

Changes in checking for allowed column types (EJD).

Added XYZLIM ASU which automatically sets limits of map to that of
asymmetric unit for map spacegroup.

Set MAXZSIZE in s/r SHFMAP.

Clarified log output describing exclusion conditions for reflections.

EJD: New s/r HARKER to print out Harker vectors and sections.

Fix to automatic FFT spacegroup selection - from EJD.

abcoeff file printed with f1*k1 and f2*k2 i.e. without the weight. Header
to file changed to avoid confusion.

User used FILLIN option with SIG2 assigned to SIG1=MNF.
So LOGMSS checks added for SIG2.

EJD: changes to CHKSYM.

From EJD:  small changes to fft; differences now listed for original hkl
   And for Patterson in P1bar permutation of AXES was not allowed.
   Use P1 instead - it gives the same answers...

Also, removed ccpasz because function was null.

Minor changes. Including changes from EJD.

A few FORMAT changes + check for SYMM card if hklin not given. Plus change
from Ian to keep F +ve.

Automatic fftspacegroup calc. improved, could get the wrong SG previously.
P222 and P2221 removed as valid FFTSPACEGROUPS.

.

Some major changes from Eleanor. P2 did not work properly and has been
removed as legal FFT SG.

PRE: new GRID keyword to specify the sampling
grid as a fraction of the resolution

fix to make FILL keyword work.

files CLOSED.

Warning messages only once. Error in MNF checking.

Missing number stuff added. Also, FILLIN keyword added. Only works with MNFs.

Pass program input column types to lrassn, except where they can be F or J.

Trim adjacent null comments

Mainly: Change the handling of missing data.  Now, If any program
label is assigned but the datum is missing, we reject the reflection
on the basis that columns will be assigned iff they're required for
calculation or rejection purposes.  This assumption avoids lots of
error-prone special-casing later, but it could catch the unwary.
(Done after consulting PRE and KDC.)

Also some tidying up, particularly in GETHKL and fix buf in GPTINT
call for free R flag input.

Don't lose on missing datum for F1 in GGETF.  (Needs checking.)

Logical order changed for missing no.

Trivial message change

EJD: Missing data stuff (but still not all correct checks).  192
symops.
DL: check for zero reflexions accepted, assorted error messages
etc. changed.

grid limits for 144 and 18r are now correct.

unsed vars

removed calculation of rms in PRNTYZ because it was wrong. Also it is already
calculated.

Abcoeffs now written out correctly. The coeffs are written out twice but
as it is only a test I didn't fix it.

Must assign SIGn with EXCL SIGn but sigma column is not compulsory.

unused vars

just error message

Somehow lost changes for r1.38. Redone.

see MTZLIB.

Error from Eleanor in P61/P65, grid for z is now 12n.

Correct error reported by Howard with free R conjunction which needed
to be short-circuited to work.

IJT: add LRCLOS in CLOSEIT so it works on VMS

PRE, following Pete Dunten: correct error for double difference map
reflexions with |Fph(calc)|=0.

EJD: USEFREE must NOT be tested when reading back FFTHKLSCR.

.

EJD: add calls to PGDEFN.
DL: fix uninitialised RSMFFT if symmetry not explicit.

remove spurious write

IJT: extra translation check in CHKSYM for patterson

A default Patterson SG of 3 was wrong.
EJD change to Patterson calc.

Defaulting the FFTSG for Pattersons didn't work generally (if at all)

tidying, free R stuff

tidy fftctl and remove multiple LRCLOS on i/p file (reported by IJT)

version pulled straight from york

PRE: Modified s/r RDSCAL in parser.for, to allow sensible input
	of multiple scales [Bfactors] from the same line.
	Changed fft.for, fftbig.for, icoefl.for to use this option, &
	check for failures if input illegal

PRE minor fix to HKLMAX option. Also fix to allow
explicit symmetry operators without bombing

PRE: added HKLMAX option for anisotropic resolution (strongly discouraged
for normal use). Altered some statement numbers in s/r RCSFFT

EJD cosmetic

EJD: more information when grid is wrong

don't write abcoeffs title always (to unopened file).
Write the title in fftbig.

EJD: typo

12/5/93 Modified fft to fix bug reported by Ian Tickle: changed mtz stream
	numbers so that scratch file is written & read on stream 2
	Tested on translation function, but NOT EXHAUSTIVELY TESTED (Phil)

EJD:
    Serious error in FFT for spacegroups P65 ( and 61 ..)
    P65 Correct for sections z=1/3 to 1/2 only.
    P61 Correct for sections z=0 to 1/6 only.
    Took opportunity to correct error in fftkw.abcoeffs list and add title

remove FACTRZ for fftlib

LMB Modifed (also fft.1 man page) to add
	EXCLUDE F1MAX F2MAX option, to exclude large Fs, particularly
	useful for DelAnom Pattersons

put fndsmp in fftlib

remove setlim, setgrd from fft, sfall to fftlib

     Changes to P1 grid; make it divisible by 2 2 2;
  problems with changing map axes otherwise.

PRE: set Patterson spacegroups properly
(modified from EJD version). Calls s/r PATSPG from symlib

EJD: changes to get Patterson spacegroups correct

*** empty log message ***

PRE (?): mod to fft/fftbig for Eleanor, set default limts to those
from fft spacegroup if true spacegroup not recognized

IJT: Patterson/bias fix

correct errors in P2, C2 from PJB/PRE

from PRE: use true Laue group to determine default grid, not from FFT spacegroup

from PRE: I'd screwed up the LIST command so that
LIST 0  caused listing of all reflections to ABCOEFFS

standardisation

from PRE:
Modified fft (under test with EJD): changed SETGRD algorithm again,
based on Laue group from EJD. Seems to work. Also changed SETLIM slightly

replace ccpvrs with ccprcs

EJD/PRE changes (see 1.2.2 branch)

add ccprcs

standardise PRE version

cosmetic change from PRE

from PRE:
1/9/92  mods to EJD's fft.for, restoring some of earlier code for
	defaulting grid & limits

(a) SETGRD: setting a default grid is dependent on the FFT
space-group, not the Laue group as Eleanor had changed it to, since
the strange sampling limits are specific to particular spacegroups
	s/r SETGRD now takes arguments defining minimum sampling
(=2hmax+1 etc) & SAMPLE which defines how fine to sample

(b) SETLIM: sets asymmetric unit depending on true space-group as
before, not FFT space-group. If you're going to extend the map, it
makes sense to calculate as much of the cell as possible, but I don't
think this should be the default.

(c) Symmetry checks. Note that AXIS permutation is only allowed to FFT
spacegroups P1, P1bar, P222, P212121, C222, I222, Pmmm: for P212121
the permutation must be cyclic: this is now checked. Also removed
warning message from CHKSYM, & modified some error messages (still not
perfect)

    Late:
       I hope I have fixed FFTSPG 47 problem.
       Havent agreed to Phils wish that the limts should be that of
     real spacegroup, and not those of FFT spacegroup. I feel it is a
     waste to calculate whole unit cell then not output it to save
     EXTEND trouble of generating it.. People only keep extended ,maps
     anyway...
       Havent done anything about confusing messages. I want to stop
     people permuting axes, except in P1.

DL: fixed various portability problems

fetched from York:
    Trying to fix fft SETGRD SETLIM stuff. Have changed FFT and SFALL to
  use same subroutines.

From: Dave 'Mapmaster' Schuller <SCHULLER@edu.umn.med.dccc>
Allow fft to use figure of merit weighting like the example in the namual,
even when the FOM is assigned type M, like in the manual.

From: Evans "P." <pre@uk.ac.cam.mrc-lmb>
I think we abolished the column type M since it's the same as W. Does
anyone remember? (despite the manual!)

From: CCP4 suite <ccp4@uk.ac.york.yorvic>
   yes - we did. Need to correct the manual!!
    But the fft fix would do no harm..

prior to proper attempt at version control

Roughly correposnds to first `official' release of MTZ version for un*x
and VMS around 14--15/5/92.

Initial revision