RUMP Enhancements Log File

<News>
<item> Use of the Ziegler2005 stopping powers has been simplified.
<item> Limits on the number of layers is removed.
</News>

This file is to contain changes, enhancements, and now fixes to the RUMP
code from the earlier FORTRAN version.  Specifically, it should document
all known differences between results.  As of 1/1/94, the basic functions
of the DOS version were completely transferred.  Changes now are
enhancements over the DOS version.

MOT - 7/2010
    Modified resonance code to read default R33 format files from the ibandl
    web site.  www-nds.iaea.org/ibandl is now recommended as the default
    source for all resonance files.  SIM STATUS also now lists the reaction
    in conventional format - 16O(a,a)16O

MOT - Wow - haven't documented anything for 3 years!

MOT - 5/21/2007 Minor fix
    Still another small problem in the pulse pileup code.  Error shows
    up only in such environments as Au on C.  Pileup not extended far
    enough.

MOT - 5/15/2007 Minor enhancement
    The search for user DLL's now scans the RBS config path rather than
    the GENPLOT config path.
    
MOT - 4/2/2007  *ERROR - SIGNIFICANT FIX*
    The pulse pileup calculation has incorrectly "reduced" the overall
    spectrum incorrectly since its first implementation in 1994.  (Only
    the difference between sum and sum2 in the code).  This was found
    as part of the IAEA intercomparison with NDF and SIMNRA - which
    were correct.  As a practical matter, it should not have
    significantly impacted any but extremely high currents with high Z
    elements and long integration times.  In rump, for current*tau
    greater than 2000, the effect is less than 1% and would be almost
    completely disappear with a change in the correction factor.

    RUMP disagrees with NDF/SIMNRA on the absolute level of the pileup
    by a factor of approximately 3.  Rump uses a triangular shape to
    approximate the typical bipolar shaping electronics -- going
    through 0 at tau.  From there, it is an exact convolution of the
    pileup mechanism.  NDF and SIMNRA use different approximations to
    the pulse shape.  Someday will get enough data to clean up.

MOT - 3/19/2007
    Added simulation direct ability to include a background in the
simulation.  The "background <buffer>" command in sim will cause the
counts in the specified buffer to be added to all simulations -- just
before the optional addition of noise to the spectrum.  This improves
the ability to do fits with multiple/plural scattering addressed as a
set of background counts.  Direct result of the IAEA intercomparison
work.

MOT - 2/9/2007
    Documented change.  Slight change in charge of an electron and
    dielectric permittivity of free space (sub 0.1% change).  Just to be
    consistent with latest values.  Modified sigma.c, sim2.c.  Now using
    the values from NIST 2002 CODATA.
        (e^2/4) goes from 1299.242 to 1295.9358 keV^2-barn
        (1/q)   conversion from barn/sr to internal unit goes from 6.242 to 6.2415

MOT - Numerous undocumented changes - sorry.

MOT - 10/5/05
    Modified UserStop format.  See new code examples.

MOT - 10/5/05
    Added variable "RUTHERFORD_FORWARD" check.  If set to any value,
    forward scattering is pure Rutherford.

MOT - 3/31/05
      Increased the maximum size of a spectrum from 2048 to 16384.

MOT - 8/26/02
      Revised error message handling on data file reads.  Now will point out
      the source of the error when reading RBS files.  Helps to discern the
      problems in some cases.  No impact on 99.99% of the users.

MOT - 3/16/02
      Added a new data output/input format - Excel.  Actually tab-delimited
      format for reading by Excel.  SAVEAS *.xls will now write the file in
      a tab-delimited format.  The first part of the file are headers
      containing a full specification of the RUMP headers.  Following that
      is a four column data output.  
          Column 1 - channel number (0 through n)
          Column 2 - actual counts (from the RUMP file)
          Column 3 - converted energy (MeV)
          Column 4 - Normalized yield (counts/msr/uC/keV)
      If the file is not modified, it may also be read back into RUMP.  In
      the data section, only the second column is interpreted.
	/* Excel format [1.0] from RUMP */
	Original File	f:\cgs\rump\tests\mb\ast.xls
	Spectrum	RBS
	Identifier	Simulation of Si-O/Si
	Date	
	Charge	10.000000
	MeV	3.100000
	Conversion	4.950000	1.600000
	Theta	7.000000
	Phi	9.000000
	Omega	3.400000
	Corr	1.004129
	Choff	1.000000
	FWHM	35.000000
	Current	8.000000
	Tau	5.000000
	Geometry	Cornell
	Beam	4He++
	
	Channel	Counts	Energy (MeV)	Normalized Yield
	0	0	0.0016	0
	1	1	0.00655	0.0130647
	2	1	0.0115	0.0130647
	249	744	1.23415	8.88451
	250	744	1.2391	8.8814
	1022	0	5.0605	0

MOT - 3/1/02
      Changed slightly the way writes are handled in RUMP.  First scan is made 
      on the specified extension.  If this is found in the list of write routines,
      then write is made by that format (.rbs => BINARY, .txt => EXCEL, .dat => ASCII)
      This makes it easier to specify the output format.

MOT - 10/24/01
      Fixed problem with GENPLOT command.  The energy conversion failed
      on transferring a spectrum to GENPLOT.  Problem with macros versus
      function calls - I should know better.  A freshman programming errror.

PR/MOT - 2/23/01
      Change to extrapolation of H and D forward scattering crosssections.
      In previous code, if the incident energy were above the limits for
      either the Ziegler or Quillet H/D scattering formulas, the code switched
      to Rutherford throughout the energy bounds.  Now, it will instead just
      change to using the 1/E^2 energy dependence of Rutherford above the
      valid energy range of both routines.  The cross section will be 
      continuous in energy, just going to simple 1/E^2 above the 4.0 MeV
      (Ziegler) or 2.7 MeV (Quillet).

PR/MOT - 2/23/01
      Error in Quillot calculations for ERD cross sections.  +1.278E-1 is
      now corrected to -1.278E-1 in two places of sigma.c.

MOT - 6/26/00
      Changed "menu" read/save as routines so that they use the current
      working directory when first opened.

MOT - 5/15/00
      Fix problem with reading via the drop down menu.

MOT - 4/27/00
      Clarified and added non-important option to read and write ASCII
      module.
         READ   <file> -ASCII [-onecolumn | -1column | -twocolumn | -2column]
         SAVEAS <file> -ASCII [-onecolumn | -1column | -twocolumn | -2column]
         SWALLOW              [-onecolumn | -1column | -twocolumn | -2column]
      Default is single column mode, each line contains exactly one channel of
      data.  In two column mode, the first column is taken to be the channel
      number and the second is the count data.  Swallow expects the data to
      follow on the command lines (or macro file), ending with a blank line.

MOT - 4/13/99
      Removed "Rutherford Integral" from spectral calculations.  It was 
      creating too many problems with complex resonances and new handling
      of cross sections.  Fixed some minor problems with resonances, including
      forcing code to just switch to Rutherford on either side of a table
      range.
  
MOT - 12/29/99
      Added screening correction to Rutherford cross section at low E.
           J. L'Ecuyer et al, Nucl. Instr. and Methods 160, 337 (1979)
           sigma *= (1 - 0.049*Z1*pow(Z2,4/3)/E)
      where E is in keV.  This should sligtly reduce the deviation at
      low energies for the RUMP simulation tail, but is generally not
      visible even at low energies.  Will make a few percent difference 
      in gold and similar heavy metal simulations.

      For Si at 3 MeV, the maximum deviation in data at low E is ~0.1%.
      For Au at 1 MeV, the deviation is about 3%.

      Because of the potential problem this change may create, it can
      be turned off by setting a variable NO_SIGMA_SCREEN to any value.

MOT - Modified ASCII read routine so that the first line not recognized
      as data columns will be copied into the "ID" buffer.  If not line
      is found, then the ID will simply be "! <filename>" as before.

MOT - Added new stopping power formulations, especially the data of
      Kalbitzer's group (Konac particularly for Si).  The data is in a
      new file "newstop.kal" in the RUMP data directory.

MOT - Default stopping power representation is now a SQRT form rather
      than linear.  This provides significant improvement over the linear
      form, and it seems now to be sufficiently stable to make the 
      default.

MOT - Aded Kalbitzer format stopping power tables.  The extension for a
      file of this format is .kal.  They can be force loaded (changed)
      using the command
            CONFIG STOP_LOAD <file>.kal
            CONFIG STOP_LOAD -unload
      will release the tables so you can revert to RUMP values.

MOT - Changed BACKGROUND command to include a -inplace options so no more 
      buffers get created.  Internally documented using the -? option.
      [Also fixed so channels can be entered manually.  This hasn't worked
      for a LONG time - no one ever complained :-) ]

MOT - Added option to GENPLOT <buffer> command to allow transfer in
      channel rather than energy mode.  
          GENPLOT <buffer> [-channel | -energy]
      Default, as previous, is -energy.

MOT - Added option to ASCII write to do as two column mode.
         SAVEAS <file> -ASCII -TWOCOLUMN
      will save it in channel / counts format (plus the headers).
      Read will have some problems!  This also modifies the
      Swallow command so that it can read two-column data.
         SWALLOW [-2column]
         SAVEAS <filename> [-ASCII [-TWOCOLumn] ]
         READ   <filename> [-ASCII [-TWOCOLumn] ]

MOT - As per recommendation (and my agreement), PERT will now set the
      correction factor that it computes in a run.  The message is
      slightly modified to read: 
          Estimated correction factor set for buffer: 1.0042
      as opposed to 
          Estimated correction factor for buffer: 1.0042
      This is now the default behavior.


MOT - CAVEAT: The following work is "in progress".  The format may be
      modified based on feedback from either Vickridge or Vizkelethy.

      Loading resonance files:
           SIM RESREAD <filename>
      This ultimately will be moved out of SIM (where it doesn't
      belong) and into CONFIG.  But the backward compatibility will
      be maintained.

      Work on resonance files continues.  Resonance files may now 
      specified in roughly the same format as proposed for IBA reaction
      cross sections (I.C. Vickridge 1993).  The format has some minor
      differences to help RUMP identify files and to permit use of
      relative cross-sections.

      See the Idaho State University site
         http://ibaserver.physics.isu.edu/sigmabase/documents/ian_prop.html
      for a copy of the DSIR Physical Sciences Report 33 original text.  All
      of the tags presented in that document are allowed.  

      In addition, 4 new tags are defined.
         VERSION: DSIR 33a       -- Must be first line in file.  Identifies
                                    the file as a resonance file conforming
                                    to the DSIR report 33 format (modified).
         UNITS: mb/sr            -- Specifies the type of data for sigma.
                rtr                 Default is mb/sr.  "rtr" means relative
                                    to Rutherford -- ratio of actual cross
                                    section to Rutherford.  Useful only
                                    for resonance reactions, eg. 16O(a,a)16O.
         DATA:                   -- Indicates start of an indeterminate # of
                                    data points.  Original standard requires
                                    NVALUES: <n> to specify precise number.
         NVALUES: <n>            -- The NVALUES: tag may be used instead of the
                                    DATA: tag as per original standard.  If
                                    a positive number of points is given, then
                                    a maximum of that many data values will be
                                    used.  The remainder of the file will be
                                    ignored once this limit is reached.  Setting
                                    <n> to 0 or a negative value is equivalent to
                                    the DATA: tag.
         END_DATA:               -- Optional tag to mark end of data.  RUMP
                                    returns to reading header tags after this
                                    tag and thus resonance files may be 
                                    "stacked" together (ie. several angles)

      In addition, all tag lines may be preceeded by /* or C so that the files
      can be read simply by GENPLOT for plotting.  Within the data block, blank
      lines or lines beginning with the comment character(s) will be ignored.

      RUMP interprets only the COMMENT, UNITS, REACTION, DISTRIBUTION, QVALUE, 
      THETA, ENFACTORS, SIGFACTORS, DATA, and END_DATA tags.  Other tags
      are simply read and ignored (including the MASSES tag).  No specific
      order for the tags is required, and tags may be repeated (only last 
      kept though).  Defaults exist for all important tags, corresponding to
      oxygen resonance at 168 degree laboratory angle in mb/sr.

      Reactions must be fully specified, ie. 16O(4He,4He)16O or 16O(a,a)16O.
      If the reaction is not a simple scattering event (Q non-zero or masses
      change), it will invoke an error and abort reading the file.

      Finally, RUMP requires resonance data to be strictly sorted in ascending
      energy.  Any point violating this condition (energy less than or equal
      to previous point) will simply be discarded.  Several of the databases
      from the Sandia site have this problem.

      WARNING: The COMMENT: tag must be terminated with a blank line.  This 
               is the most common error I made in converting files to this
               format.

      Example:
          VERSION: DSIR 33a
          COMMENT: Alphas on O16; 2000-9000 keV; LAB angle = 170;
                   Cheng et al., 1993 A; RPC; 
                   Ratio to Rutherford vs. alpha lab KE in keV.

          NAME: Cheng et al.
          ADDRESS1: 
          ADDRESS2: 
          ADDRESS3: 
          ADDRESS4: 
          ADDRESS5: 
          ADDRESS6: 
          SERIAL NUMBER: 
          REACTION: 16O(a,a)16O
          DISTRIBUTION: Energy
          MASSES: 16 4 4 16
          QVALUE: 0
          THETA:  170
          SIGFACTORS: 1.0               (SIGMA FACTORS: also permitted tag)
          ENFACTORS: 1.0 0.0            (ENERGY FACTORS: also permitted tag)
          UNITS: relative
          DATA:
          2000    1.00
          2050    0.98
          2100    1.00
              .......
          8940    3.27
          8960    2.29
          8980    2.29
          9000    9.53
          END_DATA:

          THETA: 160
          DATA:
            ......
          END_DATA:

MOT - Distribution for NT now includes resonance files for (boron, carbon
      fluorine, nitrogen, oxygen, and silicon) from the Sandia site.
      [See Sandia for updated versions of these files or for the original
      reference data].  The actual data has been "edited" to remove duplicate
      points and correct obvious errors (transcription).  Files are annoted
      with changes.

MOT - Added CURRENT as variable for PERT.

MOT - Despite the potential for data faking, SIM now has a command to
      add statistical noise to the simulation.  Use
              SIM NOISE [ON | OFF]
      The default is off.  When on, statistical noise is added and the
      raw data is truncated to integral counts.  Useful for comparing
      validity of low count peaks.
          LOOP 10 RECALC OV 0
      The recalc is required to force a new simulation with different
      random numbers for the statistical result.

MOT - Getting annoyed hitting COMPRE instead of COMPARE and getting the
      COMPRESS command.  It is so seldom used that it now requires ALL
      8 letters.  Should consider making COMPRE and alias for COMPARE.

MOT - It is no longer necessary to specify a normalization window in PERT.
      If never given, it defaults to assuming the value given is exactly
      correct.  Also, as long as NORM is not being used, you can now
      vary the CORRECTION parameter over the full extent.  This can be
      very useful matching full spectra.

MOT - Major changes happening in SIM again. 

      FUZZ is now enabled on every layer rather than limited -- this also
      means that FUZZ now longer gets lost when layers are inserted or
      removed.  New command:
         FUZZY <amount> <iterations>
             If amount is <=0 or iterations is zero, same as UNFUZZ.
             Units for amount of fuzzing are the same as the layer itself.
         UNFUZZy
             Cancels fuzzing for this layer
         FUZZ [ <ilayer> <amount> <iterations> ... repeat ... / ]
      SIM SAVE uses the FUZZY command to re-enable.
      PERT changes only slightly.
         PERT VARY FUZZ <layer #> <min value> <max value>
      whereas it used to take an index instead of layer #.  New PERT 
      HAS NOT BEEN extensively TESTED but I don't expect problems.

MOT - Major changes happening in SIM again. 

      Computers are fast enough that MAX_PATH has been by default 
      reduced to 200 /cm2 rather than 5000 /cm2.  The default value 
      can be changed using the new command:
         DFLT_MAXPATH <default_maxpath>
      This value will be inserted as the MAXPATH for all new sims
      (reset command).

MOT - Major changes happening in SIM again. 

      Global impurity profiles.  This is a new feature allowing very
      complex impurity profiles to be introduced atop a basic sample
      structure.  It can be used to implement an element or compound
      profiling within the sample.
         G_COMPosition  -  specifies the composition of the global
                           impurity profile (multiple elements)
         G_CURVE        -  name of a GENPLOT curve that contains the
                           atomic fraction as a function of depth.
                           The depth MUST be in units of 10^15 at/cm2.
         G_START        -  Profile starts (x coordinate measured) at 
                           this  layer #.  Default is 0 which means 
                           the physical surface of the sample.  All
                           absorber layers are ignored.  If specified
                           as non-zero, then X coordinate for profile
                           is taken to start at that layer # -- or
                           physical surface -- whichever comes last.
         G_MODE         -  knot fitting mode.
                              0 is normal cubic interpolation
                              1 is a linear interpolation (bore-ing)
                           
      The actual impurity profile is taken as a spline of the given
      curve (knots).  A very few points will represent a smooth 
      profile in default mode.  Setting mode can change to piecewise
      linear instead of cubic if desired.

      This global impurity profile is added into the spectrum AFTER
      the diffusing species and matrix and extends across layer
      boundaries.  It is not affected by FUZZ except for layers
      before the G_START.

MOT - Major changes happening in SIM again. 

      New equation mode - SPLINE.  The five coefficients for this
      equation are the knots of a spline profile with X coordinates
      at (0,1/4,2/4,3/4,1.0) fraction through the layer.  See GLOBAL
      diffusant equation for more knots on a spline profile.

MOT - Pulse pileup modeling is sufficiently safe now to disable the old
      backup mode.  Prior to 12/5/98 build, if tau were 0, the routine
      would automatically switch to DOS mode compatibility.  This no 
      longer happens and instead no pileup is performed.  You must
      specifically request DOS mode pileup modeling using the
      SIM PILEUP DOS if so desired.

MOT - Major changes in the internal handling of read/write.  Should be
      generally better -- and certainly more supportable.  Let me know
      of any problems.  Only functional change is:
           WRASCII test.asc
           WRITE test.asc -ASCII
      are no longer exactly identical.  Later produces a better file
      for reading via either "READ test.asc" or "XEQ test.asc".  WRASCII
      retains compatability.  The latter, however, is subject to
      ongoing change to support general ASCII data storage of RBS data.

MOT - Reading multiple data formats (distinguished by extension) is now
      cleanly implemented on NT/95 via DLL's loaded with RUMP.  Sites can
      write their own DLLs to handle reading of individual data formats. 
      SUNY Albany, Bell Labs, and a WEB-RBS format have been implemented
      as examples.  See WEB site (www.genplot.com) for download details.

      Modified command:  
           read -CONFIG
	   read -RELEASE
           read <path> -FORMAT <ID>

      The -CONFIG option will list out all the currently loaded modules.
      These modules are read from the entire RumpConfigPath when the first
      file read is requested.  Any file in the path an extension of ".rdr"
      is assumed to be a RUMP reader DLL.
       
      The -FORMAT <ID> option overrides the extension determination of
      data file format.  The ID may be determined from the -CONFIG list.
           read test.rbs -format SUNY
      will read the file test.rbs assuming the SUNY format despite the
      .rbs extension (which normally means RUMP format).

      Existing files formats are: SUNY, LUCENT, MCA.
           suny.rdr       SUNY Albany binary data format     .dat
           lucent.rdr     Lucent Bell Labs ASCII format      .mca
           mca_read.rdr   Livetime read of Cornell MCA       .wrbs

      If you have mca_read.rdr, try plot 1.wrbs.  This will download the
      current data in sector 1 of the Cornell 1 MeV accelerator MCA.


MOT - The "description" written with a .lcm file failed due to lack of CRLF.
      Code now properly writes a new line as well as the text.

MOT - Documented the STOP_USER <routine_name> package for loading alternate
      stopping power calculation routines.  See the MyStop directory for
      details.
    
             CONFIG STOP_USER [<routine> | -]

      Loads the specified dll to calculate stopping powers before resorting
      to the Ziegler default.  The "-" form deletes the current loaded
      dll (if any), and in addition releases any and all stopping power
      tables currently defined.  This causes all stopping powers to be
      recalculated.


MOT - Modified the stopping calculation routine to include a kludge for
      element 93 as Mylar.  This is a holdover from the DOS code where
      Mylar was explicitely added to the tables for FRES users.  Forcing
      users to use the stopping exception load method is considered
      poor compatibility -- just hope no one really wants to start working
      with the transuranic elements.

MOT - Added command "VERsion" to print the current revision information.
      Serial number verification now implemented.

MOT - Working with the calibration of spectra is a major pain in the butt.
      This will change very quickly.

MOT - MAJOR BUG FIX:
      Straggling with forward recoil was in real trouble -- a constant was
      not being properly set under inconsistent conditions.  I think this
      has been fixed.

MOT - Added command to SIM:
        FRES_ABSORBER <num_layers>
      which is synonymous with the ABSORBER command except that in this
      form, the absorber foil is ONLY used if the spectrum type is FRES;
      if not, the absorber layer is totally ignored.  This allows the same
      simulation description to be used for the RBS glancing angle and 
      FRES spectrum without defining/destroying the absorber definition.

MOT - Added command:
         SPECtrum [RBS | FRES | PIXE | NUCLEAR] 
      to set the type of spectrum in a file.  This is currently used only
      with the stopper foil modified command (see above).

MOT - Modified the read/write to keep the RBS/FRES/PIXE spectrum type
      indicator.  This will be incorporated with the ABSORBER feature in
      SIM so that the absorber need not be added/removed when working with
      joint RBS/FRES simulations.

MOT - Fixed problem with STRAGGLE 1 (and possibly other STRAGGLE values)
      with "pointer error" or crash.  Overwriting via negative indices (bad!).

MOT - Fixed problem with background not doing the proper graphs.  It worked,
      but the graphing failed.

MOT - Major problem with Pert has been fixed.  It looks like it is working
      now.  Problem existed with Poisson stastics as well as coding errors.

MOT - Added code so that ~C and ~A on PMDRIV screen gets to the RUMP data
      information also.

MOT - It has been suggested that PERT SINGLE is supposed to immediately start
      a fit on a single parameter.  Since I've forgotten about DOS, let's make
      it so.  PERT SINGLE assumes that an immediate "VARY" operation will be
      specified with the requested change.  PERT will then immediately do a
      fit and return.  If any existing "VARY" setups exist, they will be 
      retained (thus you can mix multiple and single operations).
          PERT SINGLE THICK 1 1000 3000
      will do an immediate variation on the thickness of layer 1 between 1000
      and 3000 units using the existing definitions for error and normalization
      windows.

MOT - Re-Fix MATCH command so properly sets the MODIFY (not DIRTY) flag on buffers.

MOT - Fix MATCH command so properly sets the DIRTY flag on buffers.

MOT - New command via Revesz request.  Available at the SIM level.

         Usage: MATCH [options] <low chan> <high chan>
         
                Command sets correction of current buffer so that it matches the integral
                of the simulation over the range.  Cursor is used if low channel is /.
         
                Options, which may preceed or follow channel numbers include:
                    -?                 this message
                    -Buffer <num>      match to a buffer other than simulation
                    -NOSet             just report value, don't set as new CORR factor
                    -Silent            quiet mode

      Command changes the correction factor on a spectra so that the integral
      of counts over the region matches the current simulation (or another 
      buffer is so specified).

MOT - Fixed crash in NT on simulation.  This error probably also fixed an
      error which would show up if you used diffusing species in the top
      layer.

MOT - In honor of Roger pointing out the density error, the element
      command now accepts a list.  "ELEMENT Si,O,Ca,Th230" works.

MOT - MAJOR FIX: Any specific material units (SiO2, AlP, ...) as a 
      thickness would fail -- defaulting to /CM2 units independent
      of the given density.  How come no-one reported this?  Isn't
      anyone doing quantitative work with compounds?  Aargh ...
      (my personal excuse is that I'm only interested in Si so never
      worry about compounds :-) )

MOT - MAJOR CHANGE in how density of compounds is approximated.  Now
      correctly averages the cm^3/atom instead of atoms/cm^3.  This will
      change the conversion of convenience units (A, um) to true at/cm^2
      in simulations.  For compatibility, a backward mode is available
      with the command:
          CONFIG DENSITY_CALC COMPATIBLE RETURN
      which can be inserted into the rump.ini file.  The change does not
      affect use of CM2, M/CM2, and specific compound densities -- only 
      physical units in compounds.

      Also, to see the calculated density and thickness conversions (both
      to at/cm2 and to A, an option is available on SIM SHOW -DETAIL.


MOT - By request, if the ID of a spectrum is empty, the filename will be
      associated with the IDS command.

MOT - Fixed plotting so in PORTRAIT mode can still do an ELEMENT command.

MOT - Added ability read and write via pipes "| zcat file.gz" and
      automatic handling of gzip'd files (will read and write).  On
      read from a pipe, the character is changed to ! so that a rewrite
      can be trapped and failed (as opposed to desirable pipe write).
      However, once a file is marked as a pipe for reading or writing,
      it can be done repetatively as desired.  These subtleties apply
      primarily to binary read/write although ascii read by pipe is
      possible also.  The gain though is minimal on most RBS files, 
      1040 bytes vers 1400 bytes (so 30%).  But's there for anyone who
      wants it.

MOT - Removed listing of internal addresses from PERT listing.  Not
      terribly relevent to most users.

MOT - Fixed SPLOT so it works on layer numbers.  It's about time!

MOT - Fixed the use of sthickness -- the units of this parameter now has
      meaning again.  Although it was requested, code assumed that the units
      were the same as the layer thickness.  Sorry.

MOT - The equation problem finally has me annoyed enough to fix.  This now
      involves some major changes in the way the equations work to make them
      consistent as well as increase the quantification of the parameters
      which are used.  In the current implementation, the quantitative
      meaning of the parameters changes with the unit of thickness specified
      for the layer.  Also, layer and species compositions were unnormalized
      so additional factors enter even in the case of the constant form.

      Basic concepts:  (1) Try to use the natural unit, if one exists, for each
                           parameter.
                       (2) In absence of natural unit, use consistent set for
                           distances, time, surface coverages, etc.
                       (3) The diffusion specifies the atomic fraction of
                           the diffusing species throughout the layer
                       (4) Result should not depend, if possible, on the 
                           sublayer division.  This will involve integral
                           forms of the diffusion equations rather than the
                           concentration form.
      Specifics:       (1) diffusivity  - cm^2/s
                       (2) velocity     - cm/s
                       (3) time         - seconds
                       (4) distances    - Angstroms
                       (5) dose/thickness - units can be specified.  Default
                           depends on equation (angstroms or atoms/cm^2)

      Internal work in RUMP is almost always based on thicknesses of 1E15
      atoms/cm^2, with atomic density used occasionally to convert to the
      equivalent thickness.  For implants and gaussians, the natural unit is
      atoms/cm^2 anyway, and only in the thin-film case is there an ambiguity.
      Since most users do not think in atoms/cm^2, while serious workers
      always do, it is necessary to have two different possibility values for
      the thin film thickness.  For this reason, requests for dose or for
      thickness (when it can be interpreted as a dose) can be given units
      as well.

      Old verions of RUMP added species into the layer composition definition
      without normalizing the relative values.  So a layer of 100 Si with a
      species of 1 Sb was equivalent to a layer of 1 Si with 0.01 Sb.  This
      problem leads now to having the layer concentrations normalized so that
      fractions specified in the equations are in atom fraction.  A diffusant
      fraction of 0.01 SiO2 into an Al2O3 layer is 1/3 of a molecule of SiO2
      per 100/5 molecules of Al2O3.

      * For THINFILM, GAUSSIAN, IMPLANT and EDGEWORTH, the film thickness
          can be given units.  If none given, angstroms will be assumed
          for thin films and /cm2 for gaussian and edgeworth.  It would be
          nice to do the same for thickfilm, but the prompts are totally
          wrong.  Stuck with thickness being Angstroms (or cm) only.

      * For SEMI-INFINITE, THICKFILM and BURIED, the default unit for the
          interface position has been changed to the angstrom.  However, the 
          code will recognize if a cm unit is entered (if less than 0.01, 
          then is assumed to be cm).

      * For semi-infinite, the direction has been reversed.  All species
          diffuse from the surface in -- so the species is at the top of
          the layer and the matrix is at the bottom.  See below for fix.

      * For ERFC, SEMI-INF and EXPONENTIAL, the profile can be moved to
          the lower edge by specifying a negative concentration for the
          species.

      * Added a new diffusion equation, BURIED which is for a thin buried
          layer.  It is equivalent to the GAUSSIAN but takes D and t 
          values instead of fwhm.

      * The sublayer thickness is no longer automatically set when an
        equation is specified.  Auto mode works instead with a default 
        subdivision set during creatr -- it is between 5 and 30 sublayers
        depending on the type of equation.  This may become more complex
        in the future to depend on the scale and overall thickness.

      * The equation USER is now much more powerful -- a different equation
        can be specified for each layer.  X is the distance in angstroms
        based on the matrix density!  If the diffusant becomes sufficiently
        concentrated to change that criteria, it will substantially modify
        the result.

      Formal definitions:

        x  -> distance (in cm) from "layer surface".  Conversion to cm based
              on given density or weighted atomic density of the matrix.

        CONSTANT    -  p0 -> fraction used
           frac = p0
           composition = matrix*(1-frac)+species*frac
        LINEAR      -  p0 -> Cs  atomic fraction at surface of layer
                       p1 -> C0  atomic fraction at bottom of layer
                             x0  thickness of layer
           frac = C0+(Cs-C0)*(x/x0)
        ERFC        -  p0 -> C0 -> maximum fraction
                       p1 -> D  -> diffusivity (cm^2/s)
                       p2 -> t  -> time (s)
           frac = C0 * erfc[x/sqrt(4DT)]
           composition = matrix*(1-frac)+species*frac
        EXPONENTIAL -  p0 -> C0 -> maximum fraction
                       p1 -> D  -> diffusivity (cm^2/s)
                       p2 -> v  -> velocity (cm/s)
           frac = C0 * exp[-x/(D/v)]
           composition = matrix*(1-frac)+species*frac
        SEMI_INF     - p0 -> C0 -> saturation fraction
                       p1 -> D  -> diffusivity (cm^2/s)
                       p2 -> t  -> time (s)
		       p3 -> x0 -> interface position (cm)
           frac = C0 * { erfc[ (x0-x)/sqrt(4Dt) ] / 2 }
           * Concentration shifts from pure matrix near start of layer (x->0)
             to fully added matrix+species at bottom of layer (x->maximum)
        THIN_FILM   - p0 -> Q  -> initial thickness
                       p1 -> D  -> diffusivity (cm^2/s)
                       p2 -> t  -> time (s)
           frac = 2*Q / sqrt(pi*D*t) * exp(-x^2/Dt)
           * Solution to diffusion of Q from surface into depth.  Old 
             versions of RUMP had a factor of 4 error in this equation.

MOT - Added commands
           XCOPY         XADD           XSUBtract
      which act like their counterparts except (1) they are hidden and
      (2) they do not attempt to update the simulation buffer before
      executing.  This can be useful on occassions.
      Users should also know that SPLOT simulations are stored in buffer -1.

MOT - Modified REGION command so much more flexible (similar to GENPLOT)
         REGION <lower channel> <upper channel>
         REGION [channel|bottom|x] <lower channel> <upper channel>
	 REGION [counts |left  |y] <lower counts> <upper counts>
         REGION [energy | top ]    <lower energy> <upper energy>
      The lower counts is still ignored, and the specification in
      energy is converted to a channel specification based on the current
      active buffer.  Old format is used if none of the keywords are
      used.  (key words are searched for unique characters only.)

MOT - Default RBS search path modified in the ginstall.exe program to
      include data directory (oops!)
          rump\os2\local;rump\os2\data;rump\os2;rump;

MOT - Modified density table handling so new values can be added to a
      file.  Specific compound densities are now loaded from a file
      "density.tab" from the RbsConfigPath.  This can have an arbitrary
      number of values.  Space for an extra SETDENS (currently=10) is
      provided for run-time definitions.  Need for more will require
      editing the file "density.tab".

MOT - Fixed so rump.ini is really obtained from the RUMP config path
      or the environment variable.  Prior, used default GENPLOT search
      path instead of the RUMP search path on initialization.  3/20/95.

MOT - Changed smoothing to use data outside the window for edge points.
      Old code (probably also in the F77 version) used a mixed method of
      constant extension on the left, but data extension on the right.

MOT - At same time as fixing elem /, also now allow Si29 as well as 29Si and
      Si+29 as valid isotope identifiers.

MOT - Fixed problem with "elem /" or "info /".  With dynamic memory alloc,
      previous request was forgotten so not meaningful to get last.  Made
      pointer static.

MOT - Per request of HAA group, inheritance on newly read files is based
      on the previously active spectrum.  This applies only to spectral
      parameters such as beam energy, MCA scaling, etc.  Defaults will be
      applied on RESET, but a read from a file without full specification
      will inherit parameters instead of inheriting defaults.

MOT - Fixed minor problem in RbsCopySpectrum having to do with overwrite
      on spectral data.

MOT - Per request of RDR, added -ASCII option to the READ command.  This
      reads data ONLY as ASCII float, arbitrary number per line ignoring
      each line at first error.  Up to CMAX (currently 2048) points will
      be read.  All other parameters in the buffer will be left as defaults.
         Usage:    read <filename> -ASCII
      Has not been extensively tested, but seems okay dokey.

MOT - Per request of WSB, added PERT GET/SAVE command pairs.  This works
      identical to the SIM GET/SAVE commands creating an ASCII command
      file containing the current settings of the PERT processor.  Let me
      know if there are any problems with this segment.  Default extension
      for PERT command files is .PERT.

MOT - Re-enabled the #pragma pack() structure in rdwr_old.c for GNU_C
      compiler.  This seems to be allowed for at least linux(tm) and permits
      reading of ancient file format.  Let me know if other machines using
      GNU_C fail.

MOT - Enabled the additional compression mode.  CONFIG WRITELEVEL 1 will
      now enable advanced compression on REWRITE (about 10-40%).  This is
      version 1.1 of the data file format.  Default initially and with
      CONFIG WRITELVEL 0 is the DOS compatability file format [1.0].  The
      advanced format is required for sparse 2D spectra characteristic of
      the TOF detector.

MOT - Started adding support for Time-of-Flight spectrum analyzers.  Use
      SIM TOF ON to enable.

      Ultimately, idea is that all SPECTRUM types must have a basic subset
      of parameters, but then how a collection of energy/yield trapezoids
      would be handled differently depending on the type of detection
      system.  SPECTRUM type would be subclassed for each type of detector.
      For now, it is just a simple kludge handling the output for Peter to
      write the TOF analysis portion.

MOT - Added PARTicle as alias to BEAM and ECONVert as alias to CONVersion

MOT - Fixed Y axis labels - novel concept to have them read properly

MOT - STOPP is now accessed form the CONFIG subprocess.

MOT - old change now documented.  Behavior of element
        ELement [-Height { / | <counts>}] <name>
      will draw a marker for element specified at appropriate surface
      channel and optionally at the specified height.  Any option will
      currently work, but preferred option is -Height.

MOT - Added NORMALIZE command to the SIM module
        NORMalize [-species] <sum>
      The sum of the composition will be set to "sum".  Any option
      causes the diffusing species sum to be done instead of normal comp.

MOT - Changed behavior of some commands which accept cursor settings.
      These commands will now first check for a command line value
      before going to the cursor.  Allows macros to set the values
      without having to do an "encursor no" command.  The / option 
      indicates desire to obtain values from the cursor (which may
      still be escaped for direct input).

      MAIN:
         background { / | <l1> <h1> <l2> <h2> } <fit_order> [-full | -noplot | -info]
            (1) This will behave differently if used to doing "background 2"
            (2) / selects cursor as in old mode
         expand { / | <low> <high> }
            (1) Expand around the specified channels
	 blowup { / | <low> <high> } <factor>
            (1) May create problems with old macros
            (2) Default expansion factor is now to abort rather than 2X

      PERT:
         normalize { / | <low_channel> <high_channel> }
            (1) the / causes the cursor to be displayed (also default)
            (2) Use of options -entry and -nocursor supported but unnecessary
         window -reset
            (1) Clears all currently defined error windows
	    (2) error is synonymous with window
         window [-add | -multiple] { / | <low_channel> <high_channel> }
            (1) the / causes the cursor to be displayed (also default)
            (2) Number of error windows currently limited to 10
            (3) If -add or -multiple is not specified, first window is
                modified.  The -add and -multiple add a new window to list.
            (4) Use of options -entry and -nocursor supported but unnecessary

MOT - Added configure command to create new stopping power tables with
      specific requirements. 
        CONFIGURE STOP_TABLE <particle> <emin> <emax> <cutoff> RETURN
           particle   - incident particle (4He, 14N, ...)
           emin, emax - range over which the Ziegler data is fit
           cutoff     - lower energy of simulation cutoffs
      You are welcome Jon.
MOT - Fixed so write and rewrite both clear the dirty and modify flags.
MOT - Fixed so read of old PC format files does not leave dirty and modify
      set.
MOT - Implemented the full smoothing functions
        smooth [-range { / | <lower> <upper>} ] [-SV | -CONV <ntimes> | -FFT <chans>]
      -SV is the default and is a normal 5 point smoothing algorithm
      -FFT uses the same FFT smoothing algorithm as GENPLOT
      -CONV attempts to smooth with a edge preserving convolution algorithm.
            Increasing the number of iterations sharpes the edge - one
            iteration is roughly equivalent to -FFT with FWHM/SCALE as width.
            Between 5 and 10 iterations is a reasonable compromise.
MOT - Numerous modifications from reports:
        (1) symbol and linetype now work like GENPLOT.  identify after a
            plot will work correctly.  ltype -2 always starts from linetype
            2 after a new axis - same with negative symbol types.
        (2) numerous message spellings errors and missing CRLF fixed.
        (3) message warning that feature does not exist for PROFILE
        (4) GENPLOT <buf> now defaults to passing the currently active buffer.

MOT - Well, background subtraction hadn't been implemented and there had
      been no complaints, so when someone mentioned it, I did not feel any
      need to remain compatible.  Use of the BACKGROUND subtract has changed
      significantly.  
         (1) Polynomial fits to order 8 allowed - weighting used on data with
             sqrt(counts) statistics
         (2) Default action is to create a new buffer containing only the
             region affected by the background subtract.  Rest of spectrum
             is simply deleted.  This buffer has filename equal to original
             but with extension replaced by ".cut".  This spectrum is
             "active" after the background operation.  Multiple "cut"
             spectra can be generated from one real spectra.
         (3) TMPBUF again gets the original spectrum with region of the
             background subtraction set to the fit background.
         (4) Option "-noplot" recognized and stops the automatic plotting
             following the background subtract.
         (5) Option "-full" causes the created buffer to contain the full
             spectrum with the background cut region sitting in the middle
             of the unmodified spectrum.
         (5) Option "-info" causes the fit parameters to be printed to the
             screen.  This uses scaled X coordinates, so may not be usful.
    
MOT - Fixed creatr function so pulse pileup shows up above the highest
      energy peak.  Stupid programming error.

MOT - Added a "calibration" clipboard to the buffer editing in XRUMP.  The
      parameters from one buffer may be copied to the clipboard, and then
      pasted into any other buffer.  Clipboard can be configured to copy
      a limited subset.

MOT - If QUIT from within GENPLOT in STOPP, program halted with an exception
      before.  Now gracefully exits with only a warning message (problem
      identified by Denis Endisch - why were you doing this??)

MOT - Corrected error in forward scattering cross-section. Had M2/M1 where
      M1/M2 should have been. (thanks again to Denis Endisch)

MOT - Incorporated changed from Denis Endisch fixing the COMPRESS error.

MOT - Fixed problem in pert when run twice.  Dumb programming error.

MOT - Error in stopping power make routine.  Never used the range specified
      in the table description - always defaulted to the fit range in the
      stopp subprocessor.  Need a table modified rewrite to STOPP.  For the
      time being, the table values always override so automatic generation
      of tables will work.

MOT - On exit, user is (by default) asked to confirm exit if any buffers
      have been modified but not saved.  There is a CONFIGURE subcommand
      to modify this behavior.

MOT - Modified "get <filename>" slightly.  First, see if fully qualified
      path is already in the buffers.  Second, try to read if the file 
      actually exists.  Third, check if an existing buffer has the same 
      filename, but different directory.  If all fail, return failure.
      Allows specification of RBS filenames while changing directories.

MOT - Edgeworth distribution limited to positive values only.  Negative
      concentrations truncated prior to error message.

MOT - New configuration options:
         PROMPT [NICE | ABUSIVE]  - nice always responds RUMP: 
         AUTORETURN [ON | OFF]    - autoreturn from SIM/PERT/CONFIGURE
	    
MOT - Added new configuration command, FRES_LIMIT, which sets the maximum
      Z that will be tracked on simulations in the forward recoil
      geometry.  Previously limited to H/D only - now can be set to any
      Z.  Stopping power tables should be automatically generated for any
      necessary interactions.

MOT - Rewrote/reorganized major sections, hope nothing broke.  Fully
      implemented the PERT with both METHOD DOOLITTLE and METHOD THOMPSON.
      There are slight differences, but not clear which is "better".

MOT - Added new command processor, CONFIGURE.  Has own help listing.  It
      will become the general interface for all global modifications of
      RUMP, such as loading resonance tables, atomic data bases.  Currently
      permits on-the-fly setting of the number of buffers.

MOT - Increased number of buffers to 20.  No reason to retain old limits.

MOT - Added PLX [ENERGY | CHANNEL] command and associated parms entry.  This
      is done under protest, but requested by some users.  Normally RUMP plots
      all data against the energy scale so disparous MCA calibrations can be
      compared.  Plotting against channel number effectively eliminates the
      energy calibrations.

MOT - Fixed error with LABEL [OFF BRIEF ON] command.

MOT - Fixed problem in SIM with isotopes.  Had printed out 1H as "1 H" which
      could not be parsed correctly again on SIM GET.  Also fixed problem in
      automatic simulation ID generation.

MOT - Added command "DESCription" to the SIM commands.  This DESCription
      command sets a description of the simulation which overrides the
      automatic generation.  SIM RESET clears this description.  Description
      also saved with the .LCM file.

MOT - Added true "IDENTIFY" command in APLOT.C code.  In response to Brunco
      request.
          IDs                --> Puts buffer description on plot as legend
          IDEn | IDENTIFIer  --> Sets description of buffer (old format)
          IDENTify           --> Similar to the GENPLOT identify command

MOT - Added "DESCription" to command list in BMANIP.C.  Replacement command
      for old IDENTIFIER command.  Use of IDENTIFIER discouraged now.

MOT - Compiler bug discovered in the ANLYZ.C code.  #ifdef CSET2 to patch
      for moment.

MOT -            ************ CHAMPAGNE DAY **************
                 ************ Dec 17, 1993  **************
      All SIM functions seem to reproduce their DOS counterparts with
      uncanny precision.  Normal RBS, FRES, RESONANCE and STRAGGLE
      simulations give results identical to parts in 10^5.  Only difference 
      is that first channel has a value in OS/2 whereas it's always 0 in 
      DOS. Major rewrite of large sections to simplify debugging work.  
      Much of the last vestige of F77 structure in the code has been squashed.

MOT - Routine tran_adt.c will convert the DOS .adt files into the new
      format for OS/2 version.  The new version is simple ASCII - they
      are neither large enough nor complex enough to warrent a binary
      storage format.  Just easier to keep it simple stupid.

MOT - Added logical names TMPBUF and SPLOT which with get or pointat will
      point to the temporary buffer where SPLOT leaves its results. 
      However, the temporary buffer will never be listed in BUFFERS command.

MOT - DOS RUMP always did simulations with FIRST set to 1.  RUMP will
      now do simulations with FIRST equal to the value in the comparison
      buffer.  Results are identical in energy, but this new mode allows
      channel by channel comparison of the original and simulated spectra
      in other modules (like GENPLOT).

MOT - Limits removed on number of knots in resonance files.  Arbitrary
      number and length of resonance files are now supported.

MOT - Limits removed on number of layers/sublayers in sample structure.

MOT - Scaling of individual stopping powers is not available.  Will have
      to be a new routine since they can by dynamically calculated, and
      old 3 table limit is removed.

MOT - (in docs) New pileup routine implemented based on Jon Custer thesis.  
      I added enough that it should now be quantitative as long as tau is 
      set  properly.  This is now the default mode, although the new command 
          SIM PILEUP [fast | accurate | none]
      can be used to select the old mode (fast).  Power-on default is
      accurate.  Setting tau (whose default is 5 uS) to zero will also
      select fast mode on a spectrum by spectrum basis.

MOT - Added new parameter to SPECTRUM structure, tau.  This is the shaping
      time constant prior to the MCA.  Used in quantitative determination of
      the pulse pileup. (See above).

MOT - New equation added to the SIM module.
          EDGEWORTH <dose> <mean> <sigma> <skew> <kurtosis>.
      Sigma is true standard deviation.  Kurtosis is specified with the
      definition where Gaussian is zero.

MOT - Modified prompts to GAUSSIAN equation to reflect that the mean
      and FWHM are specified in Angstroms independent of the user units.


REMOVED FEATURES:
==============================================================================
MOT - Reading of the real "old" format RBS files (such as the VAX ones) is
      no longer supported.   Old Ryan McFarland binary based files will be
      read (1.2 version successfully, 1.3 I think will work but I have no
      examples to test).


KNOWN LIMITATIONS:
==============================================================================
MOT - All filenames are considered case insensitive.  Don't try using
      FILE1.rbs as distinct from file1.rbs.  I don't like this feature 
      of UNIX and it won't be supported in RUMP - there!

</pre>
</body></html>