ChangeLog

2015-October-12
- fixed a serious bug that resulted in wrong results for
  p-polarized transmittances. Thank you to Matt Witzman
  who noticed it.
- fixed a bug in tf_*_sub.m functions that affected the
  calculation of transmittances when substrate effects are 
  included (thanks to Petar Atanackovic for pointing it out).
- rename tf_ellip_fit.m --> tf_ellip_d.m
- (temporarily) remove tf_ellip_model.m because it does 
  not work
- set up framework for compiling mex files.
- create Octave compatible version of fastBSpline class
  for ellipsometric modeling.
- set of auxiliary functions in directory ./aux
- make tf_nk.m much more flexible
- new function tf_ellip_sub.m to calculate the refractive
  index of a material (substrate) from the reflection at a
  single interface.
- bug fix in tf_ellip.m for substrates only.

2014-December-14, Ulf Griesmann
- begin eliminating 'levmar' library because it is too
  difficult to compile on all required platforms.
- improved 'tf_fitn.m'; no longer uses 'levmar'. Much better 
  convergence than with Octave 'leasqr'.
- new functions 'tf_ellip_fit.m' and 'tf_ellip_model.m' to 
  extract layer thicknesses, refractive index n(lambda),
  and extinction coefficient k(lambda) from ellipsometric 
  measurement data.

2014-December-07, Ulf Griesmann
- change 'tf_plotpd' to plot in different modes 'std'
  for plotting Psi, Delta, and 'mag' (magnitude) for
  plotting tan(Psi) and cos(Delta). Also add a 'pol' (polar)
  mode that plots Psi(Delta). Remove pointless 'xrange'
  argument.
- new function 'tf_ellip_match.m' to calculate or plot
  the rms difference of ellipsometric tan(Psi) and measured
  tan(Psi) = |rho| as a function of layer thickness
- new function 'tf_ellip_fit.m' to calculate layer thicknesses
  from ellipsometric data when the optical properties of the
  layer materials are known.
- new function 'tf_listnk.m' to list the materials in a 
  refractive index data collection.

2014-October-5, Ulf Griesmann
- version 62
- swapped last two arguments in 'tf_write.m', because
  the function is mostly used with the 'matlab' default
  option.
- renamed tf_optimize* functions to indicate the optimization
  algorithm in the function name. Unconstrained optimization is
  now the default in 'tf_optimize_lm.m' (Levenberg-Marquardt).
- new 'tf_optimize2.m' based on adaptive simulated annealing.
- better axis labeling in 'tf_plotnk.m'
- much improved 'tf_write.m'; now can write a nice HTML table
  that can be used by fabricators.
- better uncertainty handling in 'tf_stoney.m'. Not all 
  uncertainties need to be supplied.

2014-January-27, Ulf Griesmann
- version 60b
- correct error (missing ,) in 'tf_swingcurve.m'.
- constrain parameters to values >= 0 in 'tf_optimize3.m'.
- improvements to 'tf_plotN.m'.

2014-January-26, Ulf Griesmann
- version 60
- simplify 'tf_rootdir.m'. Now uses 'which', which is much faster.
- fixed a serious bug in 'tf_searchnk.m'.
- index data collection 'nist' is renamed to 'litho'.
- 'tf_swingcurve.m' and 'tf_rayfan.m' now accept vectors
  with wavelengths.
- 'tf_plotN.m' indicates layer boundaries.

2014-January-23, Ulf Griesmann
- version 59
- corrected the initialization of bounds in 'tf_optimize.m'.

2014-Januray-22, Ulf Griesmann
- version 58
- corrected an error in 'tf_rayfan_sub.m'.

2014-January-21, Ulf Griesmann
- version 57
- tolerance argument for 'tf_optimze.m'
- add a grid option to 'tf_plot.m'
- new function 'tf_plotGD.m' for plotting
  of group delay and group delay dispersion.
- new function 'tf_int_sub.m' calculates reflectance
  and transmittance for a substrate that has multilayers
  on front- and back-surfaces.
- new functions 'tf_spectrum_sub.m' and 'tf_rayfan_sub.m'
  calculate reflectance and transmittance spectra (and ray
  fans) with the inclusion of (incoherent) substrate effects. 

2014-January-20, Ulf Griesmann
- version 56
- calculation of GD and GDD is now correct.

2014-January-19, Ulf Griesmann
- version 55
- labels for the y-axes in 'tf_plot.m' can now
  be changed, which makes the plotting function
  more general.
- fixed title of admittance diagrams. Units are um, not nm.
- new functions 'tf_efield.m' and 'tf_plotE.m' to calculate
  and plot the electric field strength through a thin film
  stack.

2014-January-15, Ulf Griesmann
- version 54
- 'tf_write.m' now can write out a thin film stack in form
  of a callable MATLAB/Octave function which returns the
  thin film stack structure array. 

2014-January-15, Ulf Griesmann
- version 53
- tweaked parameters of 'tf_optimize3.m'.
- vectorized another loop in 'devec3.m'.
- add new Octave error code for sqp.m to 'tf_optimize.m'.

2014-January-12, Ulf Griesmann
- version 52
- new function 'tf_optimize3.m' which implements a global
  differential evolution algorithm for minimizing a 
  cost/merit function.

2014-January-6, Ulf Griesmann
- version 51
- fixed a bug in 'private/lm_opts_check.m'
- fixed a couple bugs in 'tf_needle.m', and new option
  nic.swid to set the needle width in the output film stack.
- several fixes in 'tf_plot.m' to correct bugs related
  to plotting of multiple datasets.
- slightly enhance 'tf_layer.m'; default thickness is 0, which 
  is useful for defining entrance and exit media.

2014-January-3, Ulf Griesmann
- version 49,50
- more plotting fixes and improvements. RAD and Y
  diagrams are finally plotted with square aspect
  ratio. Circle diagrams are now actual circles.
- fixed a bug with legends in 'tf_plot.m'
- fixed a recently introduced bad bug in 'tf_plotnk.m'

2014-January-2, Ulf Griesmann
- version 48
- further improvements to plotting functions, make plots
  look identical in Ocatave and MATLAB.
- correctly print '_' in plot legends.

2014-January-1, Ulf Griesmann
- version 47
- improved all plotting functions and corrected several
  errors in the error message texts. 

2013-December-30, Ulf Griesmann
- version 46
- 'tf_stack.m' also prints the total thickness
  of the thin film stack.
- improved description of 'tf_plotrad.m'.
- improved function 'tf_plot.m'. Now can plot more
  than one dataset in one plot for easier comparison.

2013-December-26, Ulf Griesmann
- version 45
- replaced 'tf_plotphase.m' with 'tf_phase.m', which 
  calculates the phase as function of the wavelength.
- new function 'tf_gd.m' for computing group delay (GD) 
  and group delay dispersion (GDD).
- new function 'tf_layer.m' makes it easier to set up 
  film stacks and to define lambda/2, lambda/4, etc. layers.

2013-December-23, Ulf Griesmann
- version 44
- fixed a bad bug in 'private/getnames.m' which
  could result in allocating far too much memory
  for even modest size film stacks.

2013-December-22, Ulf Griesmann
- version 43
- fixed spelling of message in 'tf_optimize.m'
- make unconstrained optimization an option in 'tf_optimize2.m'
- new function 'tf_prune.m' which removes layers with thickness
  outside a specified range.
- in 'tf_plotrad.m' changed the order of arguments 'radtype' and
  'nila'.
- new function 'tf_plotphase.m', which complements the amplitude plot
  by showing only the phase of the electric field through the stack.

2013-December-16, Ulf Griesmann
- version 42
- fix help text of 'tf_optimize.m'
- in 'tf_int.m' bracket the computation of t*conj(t)
  to avoid a small imaginary part due to roundoff errors in MATLAB. 
- changed matrix XYZ --> RGB in 'tf_color.m'.
- corrected the calculation of the CIE tristimulus values;
  they are now correctly normalized. The sRGB color
  space is used.

2013-December-11, Ulf Griesmann
- version 41
- 'tf_needle.m' implements the needle method for thin
  film stack synthesis.
  
2013-December-10, Ulf Griesmann
- version 40
- 'tf_optimize.m' can optimize for objectives (merit function
  target values other than 0. This enables the calculation 
  e.g. of beam splitters. 
- additional optimization function 'tf_lmoptimize2.m', which is based
  on constrained Levenberg-Marquardt optimization; easier to use for
  some applications, and it is good to have more than one nonlinear
  optimization algorithm. The new function also enables the use of 
  function objectives, which, e.g., would be needed for the design
  of dichroic mirrors.
- renamed functions tf_BC.m --> tf_bc.m and tf_BCinc.m --> tf_bcinc.m
  for reasons of style.
- cleaned up 'tf_color.m' a bit.
- fixed a bug in 'tf_maxgarn.m'. Now it beautifully reproduces the 
  color of ruby glass.
- 'tf_stack.m' now prints the optical thickness in units of the
  wavelength.
- function 'tf_write.m' now writes the data of a thin film material
  stack to a file either as a table or as Octave/MATLAB code.

2013-December-6, Ulf Griesmann
- version 39
- fixed bug in 'tf_stack.m', now prints correct sign of k.
- improvements to 'n_sellmeier.m'.
- changed 'tolower' --> 'lower' because the former only 
  exists in Octave.
- made 'tf_searchnk.m' simpler and better.

2013-December-1, Ulf Griesmann
- version 38
- better error messages when index files cannot be found/opened.
- new function 'tf_maxgarn.m' calculates the effective refractive
  index of an isotropic Maxwell Garnet model.
- 'tf_searchnk.m' can search multiple collections with a single 
  invocation.
- new 'tf_fitn.m' function to fit Cauchy and Sellmeier models
  to experimental refractive index data. Requires the LEVMAR
  library for Levenberg-Marquardt least-squares minimization.
- needle method is still work in progress ...

2013-November-16, Ulf Griesmann
- version 37
- fixed small bug in description of 'tf_plotrad.m'
- new function 'tf_plotY.m' for plotting of admittance
  diagrams.
- new functions 'tf_admit.m' and 'tf_admitinc.m' for 
  calculating admittances.
- moved 'make_arc.m' function into private directory.
- new function 'tf_stoney.m' for calculating film stresses,
  and their uncertainty, with Stoney's formula.
- plot admittance diagrams and circle diagrams with grid.

2013-November-14, Ulf Griesmann
- version 36
- 'tf_searchnk.m' no longer aborts when a wavelength is outside
  the wavelength range of a data file.
- rewrite 'nkb_file_exists.m' to not use 'dir' because that turned
  out to be incredibly slow on Octave.
- clean up 'tf_stack.m'.
- fixed generic merit functions 'tf_rmin.m' and 'tf_tmin.m'. Now they 
  work for more than one wavelength. 

2013-November-11, Ulf Griesmann
- version 35
- convert refractive index files to binary files upon first reading
  to speed up subsequent file accesses. Binary files have the file
  extension .nkb
- new function 'tf_searchnk.m' that searches the refractive index
  database for materials with a specified refractive index or
  absorption coefficient.
- correct bug in range checking in 'tf_nk.m'.
- new function 'tf_insert.m' for inserting new material layers in
  a thin film stack.
- 'tf_stack.m' also prints out the layer materials and the stack
  can be evaluated at more than one wavelength

2013-November-08, Ulf Griesmann
- version 34
- fixed bugs in general purpose merit functions 'tf_rmin.m' 
  and'tf_tmin.m'. Averaging over angles had been done wrong.
- 'tf_optimize.m' now works on MATLAB but requires 'fmincon' 
  from the Optimization Toolbox.
- fixed bugs in 'tf_plotnk.m'; plots with more than 2 refractive 
  index spectra now work correctly.
- enable plotting of refractive index spectra that are defined
  through functions with 'tf_plotnk.m'.

2013-November-07, Ulf Griesmann
- version 32, 33
- use constrained optimization in 'tf_optimize.m' to ensure
  that thicknesses remain positive.

2013-November-03, Ulf Griesmann
- version 31
- function 'tf_plotnk.m' can now plot several refractive
  index files together in one plot for comparison.
- function 'tf_readnk.m' returns the name of the index file
  and the name of the collection in ri.name

2013-October-22, Ulf Griesmann
- version 30
- arg(Z) --> atan2(imag(Z),real(Z)) because MATLAB doesn't have arg.
- tf_plotrad.m : bugfix - tf_amplinc.m has only 4 arguments.
- tf_color.m : bugfix - lnm must be defined outside if block.
- blue_mirror.m example : fixed - MATLAB imwrite can't write .eps files.

2013-October-17, Ulf Griesmann
- version 29
- changed argument order in 'tf_optimize.m' to make it 
  consistent with other functions.
- general overhaul of 'tf_optimize.m'

2013-October-15, Ulf Griesmann
- version 28
- Better handling of data from the palik collection.
  Still a mess; unclear how to handle missing data, which
  cause problems with interpolation.
- finally fixed 'tf_color.m' - I hope ...
- new function 'tf_plotrgb.m' that plots arrays
  of color swatches with titles.

2013-October-14, Ulf Griesmann
- version 27
- improved 'tf_plotnk.m': uses log x-axis when range is large,
  plot title can be added.
- fixed another bug in 'tf_color.m'.

2013-October-12, Ulf Griesmann
- version 26b
- add more CIE illuminants (A, F1-F4, D50, D65, ID50, ID65);
  should be more than enough. Especially F4 is useful for the 
  fluorescent lighting in typical laboratories.
- read the CIE standard observer functions from files
- different CIE illuminants in 'tf_color.m'.
- move all Octave/MATLAB files out of the nk directory.
- much simplified 'tf_readnk.m'
- several fixes to help texts.

2013-October-11, Ulf Griesmann
- version 25
- changed 'tf_optimize.m' such that user defined merit functions 
  can be used in addition to the built-in merit functions. 

2013-October-11, Ulf Griesmann
- version 24
- new function 'tf_version.m' that displays the toolbox version.
- revert to previous way of calculating ellipsometric angles Psi
  and Delta, which seems to be correct.
- fixed bug in help text of 'tf_plotN.m'.  

2013-October-10, Ulf Griesmann
- version 23
- new function 'tf_plotrad.m' which plots reflection amplitude 
  and transmission amplitude diagrams (aka. circle diagrams). 

2013-October-09, Ulf Griesmann
- version 22
- use atan2 in 'tf_psi.m' for the calculation of Psi.
- renamed 'tf_index.m' --> 'tf_plotN.m'.

2013-October-09, Ulf Griesmann
- version 21
- Use the same formulae as OpenFilters. [See S. Larouche and
  L. Martinu, "OpenFilters: open-source software for the design,
  optimization, and synthesis of optical filters", Appl. Opt. 47,
  C219-C230 (2008)]. They are more in line with conventions used in
  the thin film optics community. This also implies a change in
  the definition of the complex refractive index: N = n - i*k  
- new function 'tf_charmat.m' that calculates the characteristic
  matrix for each layer and returns the stack of matrices.
- change internal calculations to the 
- calculate Psi, Delta according to S. Larouche and L. Martinu, 
  Appl. Opt. 47, C219-C230 (2008).
- No default unwrapping in tf_ellip.m 

2013-October-07, Ulf Griesmann
- version 20
- added function 'tf_index.m' which plots the refractive
  index profile of a thin film stack.
- fixed help texts of several functions.

2013-October-07, Ulf Griesmann
- version 19
- make color swatches a bit smaller
- tf_plot makes a new plot window by default.
- fixed bug in tf_optimize: did not correctly return
  the optimized film stack.
- more flexible file name specification in 'tf_color.m' that permits
  writing of color swatches in different formats.

2013-October-07, Ulf Griesmann
- version 18
- bug fix in XYZ --> RGB transformation of tf_color.m. Now uses the
  CIE D65 illuminant. Colors are same as in FreeSnell now

2013-October-06, Ulf Griesmann
- version 17
- added function 'tf_color' which calculates of color swatches
  for reflection and transmission.

2013-October-05, Ulf Griesmann
- version 16
- added functions 'tf_ellip.m' and 'tf_psi.m' for calculation
  of ellipsometric Psi & Delta.

2013-October-04, Ulf Griesmann
- version 15
- move refractive index collections into a separate archive
- add index data from Edward Palik's "Handbook of Optical Constants I
& II"

2013-October-04, Ulf Griesmann
- version 14
- made tf_ampl.m simpler and faster; no extra loop to calculate M.
- mex function tf_ampl.mex to speed up inner loops of calculations
  (not ready, also not much faster. hm ...).
- added data for AZ MiR-701 and Dow Ultra-i123 i-line photoresists.
- function 'tf_optmize.m' to optimize film thicknesses.
- Contents.m file added (real documentation still missing).

2013-March-03, Ulf Griesmann
- version 12
- fixed the last bug in 'tf_ampl.m'. Now I can
  replicate the FreeSnell examples.

2013-February-24, Ulf Griesmann
- version 7
- seems to work now.

2013-February-21, Ulf Griesmann
- basic functionality for film analysis is implemented.