It's in the works!

• MacOS X version
• Cocopan
• Windows version (preliminary tests)
  • Common source code compiled in a .DLL with CodeWarrior
  • User Interface with Delphi
• Integration of StoneTable to replace the List Manager
  • >16k cells displayed
  • Non-scrolling row/col headers
  • Easy in-cell editing
    • Row/col headers
    • Values within cells
    • Add/delete rows and columns (see Copy and Paste)
• The User's manual is being written as the program is being developed, and is distributed with the program

• Import of CANOCO files of any format
• Modules
  • SIMIL (Mard 09 mars 2004 at 23:14:37 by PhC)
    • Similarity coefficients (between objects) tested against version 3
      • S01 — (a+d)/(a+b+c+d) Simple matching coeff. (Sokal & Michener)
      • S02 — (a+d)/(a+2b+2c+d) (Rogers & Tanimoto)
      • S03 — (2a+2d)/(2a+b+c+2d)
      • S04 — (a+d)/(b+c)
      • S05 — (1/4) [a/(a+b) + a/(a+c) + d/(b+d) + d/(c+d)]
      • S06 — ad/√[(a+b)(a+c)(b+d)(c+d)]
      • S07 — a/(a+b+c) Coefficient of community (Jaccard)
      • S08 — 2a/(2a+b+c) (Sørensen, Dice)
      • S09 — 3a/(3a+b+c)
      • S10 — a/(a+2b+2c)
      • S11 — a/(a+b+c+d) (Russell & Rao)
      • S12 — a/(b+c) (Kulczynski)
      • S13 — (1/2) [a/(a+b)+a/(a+c)] (Kulczynski)
      • S14 — a/√[(a+b)(a+c)] (Ochiai)
      • S15 — Σ(w[i] s[i])/Σ(w[i]) (Gower, symmetrical)
      • S16 — Σ(w[i] s'[i])/Σ(w[i]) (Estabrook & Rogers)
      • S17 — 2W/(A+B) (Steinhaus)
      • S18 — (1/2) [(W/A) + (W/B)] (Kulczynski)
      • S19 — Σ(w[i] s[i])/Σ(w[i]) (Gower, asymmetrical)
      • S20 — Σ(w[i] s'[i])/Σ(w[i]) (Legendre & Chodorowski)
      • S21 — Chi-square similarity (Roux & Reyssac)
      • S22 — Chi-square probabilistic similarity
      • S23 — Goodall's probabilistic coefficient
      • S24 — [a/√((a+b)(a+c))] - 0.5√(a+c) (Fager & McGowan)
      • S25 — 1 - p(Chi-square) (Krylov)
      • S26 — [ a + (d/2) ]/(a+b+c+d) (Faith)
      • NEI — Nei's genetic similarity (bounded between 0 and 1)
    • Distance coefficients (between objects) tested against version 3
      • D01 — Euclidean distance
      • D02 — Taxonomic, or average distance
      • D03 — Chord distance
      • D04 — Geodesic metric
      • D05 — Mahalanobis generalized distance (among groups)
      • D06 — Minkowski metric
      • D07 — Manhattan metric
      • D08 — Mean character difference (Czekanowski)
      • D09 — Index of association (Whittaker)
      • D10 — Canberra metric (Lance & Williams)
      • D11 — Coefficient of divergence (Clark)
      • D13 — Nonmetric coefficient (Watson, Williams & Lance)
      • D14 — Percentage difference (Odum; Bray & Curtis)
    • R Mode coefficients (between variables) tested against version 3
      • TAU — Kendall's Tau (new algorithm)
      • RP — Pearson's r
      • RS — Spearman's r
      • CHI — G statistic (Wilks' chi-square)
      • HT — Tschuproff's contingency coefficient
      • HS0 — (reciprocal information; Estabrook)
      • HS1 — (Rajski's coherence coefficient)
      • HS2 — (symmetric uncertainty coefficient)
      • HD — (Rajski's metric)
    • New coefficients
      • S27 — 1 - p(a): Raup & Crick's modified probabilistic association coefficient (Raup & Crick)
      • D15 — Chi-square metric (introduced in Simil 3.02)
      • D16 — Chi-square distance (introduced in Simil 3.02)
      • D17 — Hellinger distance
    • Untested/Not yet implemented
      • D12 — Coefficient of racial likeness (among groups; Pearson)
  • Correspondance Analysis
  • Periodograph
  • Chrono
  • Biogeo (spatially-constrained clustering) (Vend 29 juin 2001 at 09:59:04 by PhC)
  • PnComp (Principal components) (Lund 07 mai 2001 at 14:25:48 by PhC)
    • Works on any-sized matrix, as long as you have enough memory
    • Still missing
      • Supplementary objects and descriptors
      • Pairwise deletion of missing values
  • K-Means (Lund 12 mars 2001 at 15:46:52 by PhC)
  • Mantel (Merc 14 févr 2001 at 16:18:21 by PhC)
    • Partial Mantel correlation rewritten
      • No longer uses Smouse, Long & Sokal 1986
      • Works with both Mantel's r and Spearman's r
    • Limited permutations with group selection
    • Current correlation coefficients:
      • Mantel's Z
      • Mantel's r
      • Mantel's t (permutation-free probability approximation, useful for large, > 30 objects datasets)
    • Added correlation coefficients:
      • Spearman's r
      • Kendall's Tau
    • Left to do
      • PROTEST: Procrustes statistic sensu Jackson, 1995.
    • Mantel's correlogram is now a part of module Autocor
  • Links (Merc 14 févr 2001 at 15:16:05 by PhC)
  • Autocor (Merc 29 mars 2000 at 11:29:57 by PhC)
  • Cluster (Merc 08 mars 2000 at 16:43:00 by PhC)
  • VerNorm (Jeud 08 avri 1999 at 12:20:36 by PhC)
    • Transpose any rectangular matrix (bugfix on Jeud 08 avri 1999)
    • Make data positive
      • Simple translation of the data to a user-defined (positive) minimum
      • Useful for Box-Cox and Taylor
    • Kolmogorov-Smirnov test of normality
    • Box-Cox and Box-Cox-Bartlett normalizations
    • Ranging and standardization
    • Your choice of transformation
      • Y' = a + bY
      • Y' = Y * exp(a)
      • Y' = exp(Y)
      • Y' = ln(a + bY)
    • Multinormality test
    • Left to do
      • Taylor transformation
  • Convert (Mard 23 févr 1999 at 15:09:34 by PhC)
    • Similarity <-> Distance matrix conversions
  • PCoord: Principal COORDinates (Lund 21 sept 1998 at 15:35:45 by PhC)
    • Standard reduced-space ordinations for distance matrices (similarity matrices must be converted to distances first)
    • Lingoes (1971) and Cailliez (1983) corrections for negative eigenvalues; see DistPCoA for more information
    • Left to do
      • Iterative eigenvalue algorithm, useful for large datasets with no correction for negative eigenvalues
  • GeoDistances (Jeud 02 avri 1998 at 11:27:13 by PhC)
    • Requires data already formatted as decimal degrees (conversion from other formats will happen when importing a file)
• Use of WASTE instead of TextEdit
  • >32k of text in Text windows (limited by available memory)
• Diagram types in the R Package
  • X-Y scatterplots with optional Z variable
    • Types of Z variable (all documented in the manual):
      • Group symbol, including "none" if you want no display of the points (e.g. you only want the ellipses)
        • Symbols are now highly contrasting and can be re-ordered
      • Label (number)
      • 95% ellipse-grouping criteria (draw ellipses around a group of points)
      • Proportional "bubble"
      • Arrow: draw an arrow from the origin [0, 0] to the point itself. Very useful in CCA/RDA biplots.
      • Color (8 colors to choose from)
    • You can also use row names (characters) as labels
    • Axes can be constrained to be proportional: if one centimeter equals 50 units horizontally, one centimeter will equal 50 units vertically. Very useful in ordination diagrams.
    • Bounds can be set on both axes
    • Settings are remembered when you double-click on a graph
  • Link map (see Links)
  • Autocorrelograms
    • Now including regular or progressive Bonferroni correction
  • Histograms
  • Growth diagrams are now functional
    • They are a special kind of histogram for population dynamics
  • Saved diagrams can be re-opened and re-sized at will
  • All diagram windows can be set to an arbitrary size regardless of your screen size (e.g. 1600x1600 on a Mac Plus)
  • PICT image: paste any PICT from the clipboard
  • Distance comparison diagrams
    • Useful in reduced-space ordination for reduced vs. original space distance comparison (called a Shepard diagram in this case)
    • One can also compare arbitrary distance matrices
• The User's manual is coming along nicely, over 115 pages!
  • I try to document every feature in the manual, e.g. "Press the option key while clicking to do this..."
  • Note: the self-reading (.srd) version was dropped because it was getting too big (4.2 MB vs. 247k for the pdf version)
• Printing has been implemented!
• Better import/export code
  • Bug fix when importing unfolded triangulars
  • Export matrices in many formats
• Change language from French to English and back on the fly
  • Now uses stand-alone "dictionary" files
• Import/Export text data file
  • Export of SIMIL-type files (resemblance matrices) is no longer restricted to square matrix. You can export as:
    • Lower triangular (with or without diagonal)
    • Upper triangular (with or without diagonal)
    • Unfolded lower triangular (without diagonal)
    • Unfolded upper triangular (without diagonal)
      (This is the format required by Permute! version 3.4 alpha)
    • Square
  • Import from text files
    • Import DD MM SS-type data (i.e. GeoDist)
    • Progress thermometers
    • Faster scanning
    • Can import via drag-and-drop (user preference), also works for multiple imports if option key is held down!
    • Direct importation of DOS or UNIX text files
    • Rectangular (nxp) matrices
    • Resemblance matrices
      • Square with diagonal
      • Unfolded upper triangular without diagonal
      • Unfolded lower triangular without diagonal
      • Other formats coming soon
    • Parser is now faster and more robust; also used for Copy/Paste
  • Text format is compatible with spreadsheet programs
  • Columns in files can be delimited with tabs, spaces or commas
  • Row and column headers
• Persistent user preferences
  • Import text files via drag-and-drop
  • Language
  • Missing value indicator
  • Matrix display font
  • Number of decimals in matrix display
  • Binary computations threshold (i.e. above this value, everything is 1 and below, 0)
• Multithreading using Thread Manager (System 7.5 and up, there is an extension for previous systems)
  • If the Thread Manager is present, one is able to run multiple, concurrent tasks and continue to use one's computer while computations are being made
  • One can also turn the multithreading off to use the full power of one's computer
• Open and save
  • Binary data files for accuracy and faster opening
    • Matrix data (rows/cols)
    • SIMIL-type file (internally stored as a one row matrix)
  • Text data files (computation results) in text display windows
• Copy and paste
  • To clipboard
    • Data matrix selection to clipboard, tab-delimited
      • Internally kept as binary (private clipboard) to prevent loss of precision when pasting between two data matrices
    • Text window to clipboard
    • Picture window to clipboard
  • From clipboard
    • Tab-delimited matrix text to a matrix (uses the import parser)
    • Append to text window from clipboard
    • Picture from clipboard (non-dynamic, though)
• On-screen display of rectangular matrices and SIMIL-type files
• Current data matrices limits:
  • Rectangular matrices: 32,000 rows x 32,000 columns
  • SIMIL matrices: 32,000 values, which means 256 objects (256*255/2 = 32640)
  • I use 32,000 because it makes debugging easier!
  • When the Package will be released, the limits will be raised to 2,147,483,647 * 2,147,483,647
  • Everything is stored internally as Long Integers
    • The theoretical limit is not 32,000 (~ 2¹⁵-1) but 2,147,483,647 (2³¹-1)
  • Due to a design constraint in early versions of the R package, SIMIL matrices will never be able to handle more than 32,767 objects
    • That is still 536,821,761 values!
    • And such a matrix would require over 5,110 Megabytes of memory and storage...
    • The theoretical limits of the internal format for SIMIL matrices is 46,341 objects, but to achieve it I would have to hack backward compatibility
  • Note that there is an internal limit to the List Manager
    • It cannot display more than 16,000 cells (e.g. a 400x40 matrix)
    • Larger matrices will only show the first 16,000 values
    • However, all values will be accessible for computation
• Error reporting
  • Use alert if Macsbug is not present
  • Distinguish between fatal and non-fatal errors
• Create empty files
  • Text
  • Data matrices
  • Diagrams (pictures)
    • Labels, arrows, ellipses, symbols, bubbles
    • Re-computed from original data upon window resize
• PowerPC-native (fat binary)
• Balloon help throughout
  • Though not completely translated...
• This list changes often... please bookmark it!

• Diagram types
  • Dendrogram or tree
  • Chronological clustering
• Import/Export text data file
  • Specify number of rows and columns
  • When exporting, allow the user to specify the number of columns to wrap the data
• Data processing
  • Fold/unfold SIMIL-type matrices (distance or similarity matrices) when copying and pasting
• Picture processing
  • Zoom in/out
• Help
  • AppleGuide (dependant on OSA support)
• OSA support
  • OSA-compliant: scriptable and recordable

Last updated on Saturday, March 30, 2013 by Philippe Casgrain
Created on Saturday, April 05, 1997