File format for submitting job to the test platform of POPAN.

The theory is described in
    Schwarz, C.J. and Arnason, A.N. (1996)
    A general methodology for the analysis of capture-recapture experiments in
      open populations.
    Biometrics, 52, 860-873

Running the program:
    Type carlan.
    The program will prompt you for the name of a data file and an output file.

Note that this standalone program does not allow all of the constraints
that are possible in POPAN, e.g. gross births equal over time, unless the
user writes a special subroutine.

Several input sets may be stacked together; the program will process each
one in turn. This is useful for simulations where another program can generate
the input sets. CARLAN will analyze each simulation, and then another program can
reads the results from CARLAN and produce summary statistics about the simulations.

Error checking is MINIMAL. In particular, you must ensure that
   - all m(i) are > 0 in an unrestricted model (i>1).
   - all r(i) are > 0 in an unrestricted model (i<number of samples).
     *** However, see section below on zero summary statistics on
         work arounds ***
   - you don't exceed the limits for number of groups or sample times.

Current limits: (taken from sysconst.inc file). If you exceed these
   limit, it will usually be caught by a subscript out of range error. 
      parameter (maxsam=30)      ! maximum number of sample times
      parameter (maxgrp=8)       ! maximum number of groups
      parameter (maxrest=200)    ! maximum number of restrictions
      parameter (maxparm=3*maxsam*maxgrp+maxrest)
      parameter (maxuconstr=500) ! maximum user constraints




------------------------- Input format ---------------------------

Data file format (all lines are free format). Each set of input has the following format

      Comments are denoted by an exclamation mark (!) in column 1. These lines will be ignored completely

      Line 1: title (up to 80 characters)
      Line 2: ngroup nsamp trace_type data_type model_type design_type init_type where
              
         ngroup = number of groups
         nsamp  = number of sample times within each group. Each group must have
                  the same number of sample times.
         trace_type = notrace   - no tracing done 
                      trace     - trace flags follow 

         data_type  = stat      - summary statistics will be used
                      hist      - capture histories will be used
    
         model_type = mod       - the model will be specified using keywords (see below)
                      pim       - the model will be specified using PIMs similar to program MARK
 
         design_type = design     - additional constraints will be specified
                       nodesign   - no additional constraints will be specified.
                       Note that if one of the model keywords is ufit, a design matrix is almost certainly required
   
         init_type   = init     - specify initial values for starting iterations
                       noinit   - no initial values specified for starting the iterations
 
      Following the second line is a series of records depending on the value of the *-type flags. 

      Trace_type = notrace - nothing 
                 = trace If you specify that trace_type='trace' then 
           you can specify a series of trace flags.
           These are usually only of interest if convergence problems occurs
           or other unusal things occur. Each trace flag occurs in column k,
           and is T for setting trace flag on, and F for turning trace flag
           off. The meanings of each trace flag is found in the trace.inc file.

           trace(1) - normal output for analysis
           trace(2) - brief output suitable for simulations
           trace(3) - print initial parameter estimates
           trace(4) - print score vector components during construction
           trace(5) - print final score vector
           trace(6) - print final hessian matrix
           trace(7) - print restriction matrices 
           trace(8) - print extracted score, parms, hessian
           trace(9) - print out estimates after every iteration
           trace(10) - print out var-cov after every iteration
           trace(11) - turn off cycling, i.e. start at stage="find all" when begining iterations
           trace(12) - print the expected values of the statistics
           trace(13) - print out variances and covariances
                       for net births, pop size, gross births, and gamma estimates.
           trace(14) - if adjustments per unit time for phi are NOT to be made
           trace(15) - print hessian and full covariance matrix
                       when unconditioning the estimates
           trace(16) - uncondition the variances to account for
                       random numbers of births. This only works
                       when there are no constraints (other than
                       keeping to 0 or 1) on the births.
           trace(17) - print out details on singular values for every iteration
           trace(20) - print out details of parsing of MARK input file (not applicable to all versions)

    Data_type = stat   - summary statistics
                hist   - individual capture histories
        Summary statistics:
            Data consists of the summary statistics for each group.
            abstime(i) - absolute sample time for sample i.
            n(i) - number captured at time i
            m(i) - number marked in n(i)
            RR(i)- number released at time i. It may differ from n(i) because of
                   losses on capture and injections
            r(i) - number of RR(i) that are ever seen again
            z(i) - number seen before i, not at i, and after i.
 
            The statistics for the second group follow immediately after that of the first group.
        Capture Histories
            The first line gives the absolute sample times for times 1 ... nsamp
            This is followed by a series of histories and counts for each group in the following format
               hhhhhhhhhhhhh  c1 c2 ....
            where h=0 or 1 indicating if captured or not
                  c1 c2 ... are the counts for group 1, 2 ....
            If a count is negative, this indictes a loss on capture. It is not possible to
            have injected animals using this format - you will have to create the
            summary statistics and use the data_type=stat option.
 
            This histories MUST appear in columns 1:nsamp of the record

            The end of the histories is specified by a history of 0000000000000 0 0 ....
            with the null history also appearing only in columns 1:nsamp of the record.
            
            This is very similar to format used by RELEASE and MARK except do NOT put semi-colons at
            the end of each record, comments are not allowed, and the terminating record of all zeroes.
    
    Model_type = mod       - the model will be specified using keywords (see below)
                 pim       - the model will be specified using PIMs similar to program MARK
        Model keyword method
          The model to be fit is specified by a triplet of codes , one for
          the model for p, the second for the model for phi, and the third 
          for the model for the entrance probabilities. The syntax is similar
          to an ANOVA format. Each of the codes may be one of (taken from the models.inc file):
              g*t     no restrictions on the parameter (other than admissibility) 
              g       varies over groups and equal over time period
              t       varies over time but equal over groups
              same    equal over groups and time periods
              nodeath (only for phi) all survivals set to 100%
              nobirth (only for entry) no additonal births after time 0
              ufit    user specified constraints
              uprog   user specified subprograms - the user needs to supply the subprograms
          For example, using a 'surge' type of notation we get:
             g*t  g*t  g*t  = all sets can vary by time and group
             g    g*t  g*t  = P changes over groups, but equal across times.
                              PHI varies by time and group;
                              ENTRY varies by time and group.
             g*t  g    g*t  = P varies by time and group;
                              PHI varies by group only;
                              ENTRY varies by time and group.
             g*t nodeath g*t =P varies by time and across groups;
                              PHI = 100%, i.e. no deaths,
                              ENTRY varies by time and group.
             g*t g*t nobirth =P varies by time and across groups;
                              PHI varies by group and time;
                              No ENTRY, ie. a death only model.
          If you specify a 'ufit' for one or all of the model values, you must specify
          a set of constraints as outlined below. You may specify a 'ufit' for one or more of the
          triplets and select one of the possible models for the other parameters.
          If you specify a 'uprog' for at least one of the sets, you must write special
          subroutines to return the restrictions that you want. Contact cschwarz@cs.sfu.ca
          for more details.

        PIM Method
          A series of PIMs will be specified. 
          This will be series of index numbers for p, phi, pent for group 1, then for group 2, etc.
          If is not necessary that all index numbers be present, i.e. the index numbers don't have
          to start at 1 and have no gaps. If you specify further user constraints, these must use
          the same index numbers as in the PIMS.
 
              p1 p2 .....   p(nsamp)   - index numbers for group 1 capture probs
              phi1 ....   phi(nsamp-1) - index numbers for group 1 survival probs
              pent0 ...  pent(nsamp-1) - index numbers for group 1 entrance probabilities
              Then group2 index number follow etc.
 
          If the same index number is used more than once, then it specifies that these
          corresponding parameters are equal.



    Design_type = design     - design matrix specification follows
                  nodesign  

       The most common usage for the design matrix specification is to fix non-identifiable parameters
       such a p(1)=p(nsamp)=1.

       User specified design matrix
          The design matrix is specified by a header record specifying the number of explicit beta
          terms to be fit.
          This is followed by a series of ndrows of the form
              pim constant b1 b2 b3... b(nbeta)
          followed by a row of 
              0 0 0 0 etc
          to indicate the end of the design matrix specifications.
  
          Each row in the design matrix specification consists of
              pim - refers to a pim number either implicitly (using the internal numbering system)
                    or explicitly from the PIM matrices above
              constant - if a constant is to be fit
              b1 ... b(nbeta) coefficients of the linear model


         NOTE THAT THERE IS ONE MORE COLUMN THAN THE NUMBER OF BETA TERMS AND THAT
         THE FIRST COLUMN IS USED TO SET parameters to CONSTANTS ON THE LOGIT SCALE

          For example, consider the following specifications: 
              2        2 beta terms, but three columns
              1 10  0 0 
              2  0  1 1 
              3  0  1 2 
              4  0  1 3 
              5  0  1 4 
          specifies that pim=1 has the value of 10 (on the logit scale = 1 on ordinary scale)
          while pim=2, pim=3, pim=4, pim=5 are on a linear slope with intercept and slope terms.

          The default numbering for parameters is as follows. [If you use PIM, you must use the PIM numbers.]
             1   ... k     the prob of capture                 p(i),    i=1...k
             k+1 ... 2k-1  the prob of survival from i to i+1, phi(i),  i=1...k-1
             2k  ... 3k-1  the prob of entry from i to i+1,    pent(i), i=0,...,k-1. 
          The next groups parms then continue in the same sequence.
 
          All linear constraints can be fit using the design matrix.

          THE MOST COMMON ERROR is to forget that the constant for a parameter specified in the
          design matrix is on the logit scale. Hence to constrain a parameter to the value 1, use the
          logit value 10 (as above). To constrain a parameter to the value 0, use the logit value -10.
          Refer to some of the examples below.


    Init_type   = init     - specify initial values for starting iterations
                  noinit   - no initial values specified for starting the iterations
 
       Finally, initial values to be used when starting to iterate.
          pc(1,1)   ...   pc(1, nsamp)
          phi(1,1)  ...  phi(1, nsamp-1)
          pent(1,0) ...  pent(1,nsamp-1)
          then initial values for group 2, group 3 etc.
       Only those initial values >0 will replace those set by the program automatically. Hence
       use a negative number to let the program choose an initial estimates.




------------------------- Other issues ---------------------------

       PARAMETER NON-IDENTIFIABILITY:
       The completely unrestricted model (p(g*t), phi(g*t), b(g*t)) has
       a non-identifiability problem as outlined in the paper cited above:
            p(nsamp)*phi(nsamp-1)      cannot be separated
            p(1)*b(0)                  cannot be separated
            function(b(1),p(1),phi(1)) cannot be separated.
       This has implications for counting the number of parameters and for model
       fitting. In model 'g*t g*t g*t', the restrictions p(1)=p(nsamp)=1 are automatically
       set which implies that phi(nsamp-1), b(0), and b(1) estimate some confounded
       parameters (see paper cited above). If reduced models are fit, you should
       pay attention to some of the restrictions. E.g. if you fit a model where
       phi is equal over groups, then phi(nsamp-1) is not restricted because it is
       unlikely that phi(nsamp-1)*p(nsamp) is the same over all groups. 
       Use a design matrix to fit simpler models or to fix parameters at some value.
 
       ZERO SUMMARY STATISTICS:
       It can happen by accident or design that certain summary statistics are zero.
       This can cause identifiability problems. The most common occurances
       and what to do are listed below:
      
       m(i)=0. m(i)>0. No marked animals captured in recapture sample. 
	This could be a problem because the MLE for the population size could
        be infinite. However, the u(i) does provide some information about
        the true population size and in many cases the program does converge
        properly. If the program seems to be driving pent(i-1) to 1.000
        then you need to constrain pent(i-1) to some value.
 
       n(i)=0, m(i)=0, R(i)=0, r(i)=0, z(i)>0
        Usually caused by no samples taken at a particular sample time.
        Very common in age/cohort models where animals cannot be captured
        before released in a multiyear study. Refer to Age directory for
        example programs.
 
        phi(i-1)phi(i) can be estimated.    Constrain phi(i)=1 and use phi(i-1) as the product.
        pent(i-1)+pent(i) can be estimated. Constrain pent(i)=0 and use pent(i-1) as the sum.
        p(i)=0.                             Constrain p(i)=0.
 
      z(i)=0, r(i)>0
        Usually caused by poor sample sizes. Program should converge
        properly when restricting p(i)=1 using the design matrix..
 
      z(i)=0, r(i)=0
        Usually caused by poor sample sizes. Program should converge
        when restricting phi(i)=0, but this causes problems in estimating the gross
        births.
    
!------------------- SAMPLE INPUT FILES showing various features
!These files are available in the Example subdirectory
!
!These can be run as is, all lines with the comment indicator (!) in column 1 are ignored.
!
!
!Example:   Grey Seals
!     This has two groups (males and females) with 6 sample times.
!     We will fit a model where
!           p(i) equal over groups, but changes over time
!           phi(i) equal over groups but changes over time
!           births(i) change over time and groups.
!     
!     This will use the model keywords to fit the model
!        
Grey Seals Band.year = 1969.70 Sighting year=91. p(t), phi(t), b(g*t) - model keywords
2 6 notrace stat mod design noinit   
1 8 0 8 8 0        Female summary statistics
2 43 5 43 30 3
3 65 28 65 45 5
4 72 44 72 39 6
5 60 44 60 6 1
6 10 7 10 0 0
1 49 0 49 44 0       Male summary statistics
2 75 28 75 59 16
3 58 42 58 49 33
4 75 56 75 51 26
5 80 67 79 25 10
6 39 35 39 0 0
 t t g*t             Note the way the model keywords are used
0                    Note that 0 beta terms, but 1 column appears below to fix parameters
1  10   p(1) = 1     Note that you must know the internals to be able to fix this without using a PIM 
6  10   p(6) = 1
0 0 0 0 0 0 0       end of design matrix specification



!-------------------------------------------------------------------
!Example:   Grey Seals- same as above but using a PIM
!     This has two groups (males and females) with 6 sample times.
!     We will fit a model where
!           p(i) equal over groups, but changes over time
!           phi(i) equal over groups but changes over time
!           births(i) change over time and groups.
!
!     Note the use of non-consecutive PIM index elements, i.e.
!     when you use PIM numbers, you can use any coding that is consistent with your model
!        
Grey Seals Band.year = 1969.70 Sighting year=91. p(t), phi(t), b(g*t) - PIM
2 6 notrace stat pim design noinit   
1 8 0 8 8 0        Female summary statistics
2 43 5 43 30 3
3 65 28 65 45 5
4 72 44 72 39 6
5 60 44 60 6 1
6 10 7 10 0 0
1 49 0 49 44 0       Male summary statistics
2 75 28 75 59 16
3 58 42 58 49 33
4 75 56 75 51 26
5 80 67 79 25 10
6 39 35 39 0 0
!   1  2  3  4  5  6   <- sample times
    1  2  3  4  5  6   capture rates  for group 1 vary over time
   10 11 12 13 14      survival rates for group 1 vary over time
20 21 22 23 24 25      entrance probabilities for group 1 vary over time
    1  2  3  4  5  6   capture rates  for group 2 same as first group
   10 11 12 13 14      survival rates for group 2 same as first group
30 31 32 33 34 35      entrance probabilities for group 2 differ from group 12
 0                  Note 0 beta terms specified, but one column appears below to fix parameters
1  10    p(1) = 1 
6  10    p(6) = 1
0 0 0 0 0 0 0       end of design matrix specification





!-------------------------------------------------------------------
!Example:   Grey Seals- same as above but using a mixture of PIM and DESIGN matrix
!     This has two groups (males and females) with 6 sample times.
!     We will fit a model where
!           p(i) equal over groups, but changes over time
!           phi(i) equal over groups but changes over time
!           births(i) change over time and groups.
!
!     The PIM matrix is used to constrain the capture prob between the two groups
!     while the design matrix is used to constrain the survival prob between the two groups
!        
Grey Seals Band.year = 1969.70 Sighting year=91. p(t), phi(t), b(g*t)  - PIM and design
2 6 notrace stat pim design noinit   
1 8 0 8 8 0        Female summary statistics
2 43 5 43 30 3
3 65 28 65 45 5
4 72 44 72 39 6
5 60 44 60 6 1
6 10 7 10 0 0
1 49 0 49 44 0       Male summary statistics
2 75 28 75 59 16
3 58 42 58 49 33
4 75 56 75 51 26
5 80 67 79 25 10
6 39 35 39 0 0
!   1  2  3  4  5  6   <- sample times
    1  2  3  4  5  6   capture rates  for group 1 vary over time
   10 11 12 13 14      survival rates for group 1 vary over time
20 21 22 23 24 25      entrance probabilities for group 1 vary over time
    1  2  3  4  5  6   capture rates  for group 2 same as first group
   15 16 17 18 19      survival rates for group 2 
30 31 32 33 34 35      entrance probabilities for group 2 differ from group 12
5                      There are 5  beta parameters
 1 10   0 0 0 0 0
 6 10   0 0 0 0 0
10  0   1 0 0 0 0      Group 1, phi 1
11  0   0 1 0 0 0      Group 1, phi 2
12  0   0 0 1 0 0      ...
13  0   0 0 0 1 0 
14  0   0 0 0 0 1
15  0   1 0 0 0 0      Group 2, phi 1. Note same beta paramater as group1 phi 1
16  0   0 1 0 0 0      Group 2, phi 2
17  0   0 0 1 0 0      ...
18  0   0 0 0 1 0 
19  0   0 0 0 0 1
0 0 0 0 0 0 0       end of design matrix specification






!-------------------------------------------------------------------
!Example:   Coho salmon
!     This has two groups (adults and jacks) with 8 sample times
!
!         p(i) constant over time but changes among groups, 
!         phi(i) changes over time, but equal among groups
!         birth(i) changes over time and among groups.
!
!         This examples shows how to use trace flags, and a design matrix
! 
!         [This could also have been specified with the triplet g t g*t 
!         Other ways of fitting this model are found in the Examples/Chase directory]
!  
!         
!

Chase89 Males and Jacks  1&2, 8&9 pooled - p(g), phi(t), b(g*t)
2 8 trace stat pim design noinit ( 2 groups, 8 sample times, trace flags, summary statist, no models, design, no initvals)
tffffffffffffffffffffffffffff      ( trace flags)
   1.5  85        0       85       28        0    Adults sum stats
   3    35       12       34       19       16  
   4    97       14       72       31       21  
   5    84       25       78       34       27  
   6    67       39       56       14       22  
   7    51       28       37        5        8  
   8.5  39        6       29        7        7  
   10   18       14        0        0        0
   1.5  67        0       62       21        0    Jacks sum stats
   3    28        9       25        7       12   
   4    46        6       44        9       13   
   5    47       12       45        5       10   
   6    25        9       24        3        6   
   7    16        6       12        1        3   
   8.5   7        1        5        1        3   
   10    7        4        0        0        0  
!   1  2  3  4  5  6  7  8  <- sample times
    1  2  3  4  5  6  7  8
    9 10 11 12 13 14 15
16 17 18 19 20 21 22 23
   24 25 26 27 28 29 30 31
   32 33 34 35 36 37 38
39 40 41 42 43 44 45 46
9           There are 9 beta parms
 1  0  1 0 0 0 0 0 0 0 0        p(group1,1)
 2  0  1 0 0 0 0 0 0 0 0        p(group1,2) is same a p(group1,1)
 3  0  1 0 0 0 0 0 0 0 0
 4  0  1 0 0 0 0 0 0 0 0 
 5  0  1 0 0 0 0 0 0 0 0 
 6  0  1 0 0 0 0 0 0 0 0 
 7  0  1 0 0 0 0 0 0 0 0 
 8  0  1 0 0 0 0 0 0 0 0  
24  0  0 1 0 0 0 0 0 0 0        now constrain the group 2 p's to be all equal
25  0  0 1 0 0 0 0 0 0 0
26  0  0 1 0 0 0 0 0 0 0
27  0  0 1 0 0 0 0 0 0 0 
28  0  0 1 0 0 0 0 0 0 0 
29  0  0 1 0 0 0 0 0 0 0
30  0  0 1 0 0 0 0 0 0 0 
31  0  0 1 0 0 0 0 0 0 0
 9  0  0 0 1 0 0 0 0 0 0       phi(group1,1)
32  0  0 0 1 0 0 0 0 0 0       phi(group2,1) is the same as phi(group1,1)
10  0  0 0 0 1 0 0 0 0 0
33  0  0 0 0 1 0 0 0 0 0
11  0  0 0 0 0 1 0 0 0 0  
34  0  0 0 0 0 1 0 0 0 0  
12  0  0 0 0 0 0 1 0 0 0
35  0  0 0 0 0 0 1 0 0 0
13  0  0 0 0 0 0 0 1 0 0
36  0  0 0 0 0 0 0 1 0 0
14  0  0 0 0 0 0 0 0 1 0
37  0  0 0 0 0 0 0 0 1 0
15  0  0 0 0 0 0 0 0 0 1
38  0  0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 0 0 0 0 0       end of design matrix specification





!-------------------------------------------------------------------
!Example: Age/Cohort model with constraints applied:
!   Data taken from Evan Cooch's Surge Manual - 3 age classes.
!   Refer to paper in Euring 97 proceedings for more details.
!
!   This illustrates setting some parameters to constants using the design matrix
!   and a joint use of PIM and design matrices
!
!   Note that the constant term is specified on the LOGIT scale where -10 = 0, 10 = 1
 
Example for Euring - Three age classes
3 6 notrace stat pim design noinit
    1     185       0            185      59       0  Juveniles released in yr 1
    2      46      46             46      34      13 
    3      35      35             35      26      12 
    4      22      22             22      15      16 
    5      21      21             21      16      10 
    6      26      26             26       0       0 
    1       0       0              0       0       0  Juveniles released in yr 2
    2     185       0            185      40       0 
    3      33      33             33      22       7 
    4      17      17             17       9      12 
    5      19      19             19      14       2 
    6      16      16             16       0       0 
    1       0       0              0       0       0  Juveniles released in yr 3
    2       0       0              0       0       0 
    3     185       0            185      33       0 
    4      13      13             13      13      20 
    5      26      26             26      12       7 
    6      19      19             19       0       0 

!   1  2  3  4  5  6     <- sample times
    1  2  3  4  5  6     Group 1  p's
    7  8  8  8  8        group 1  phis are equal after age 2 onwards
12 13 13 13 13 13        group 1  pents

   18 19  3  4  5  6      group 2  p's
   24 25 26 26 26        group 2  phis are equal after age 2
12 13 13 13 13 13        group 2  pents

   35 36 37  4  5  6     group 3  p's
   24 24 43 44 44        group 3  phi's are equal after age 2
12 13 13 13 13 13        group 3  pents

0         There are 4 beta parameters
 1  10        p  ( group 1, 1) = 1
18 -10        p  ( group 2, 1) is zero because of 0 sample size
19  10        p  ( group 1, 2) = 1
24  10        phi( group 2, 1) is 1 because of 0 sample size
35 -10        p  ( group 3, 1) is zero because of 0 sample size
36 -10        p  ( group 3, 2) is zero because of 0 sample size
37  10        p  ( group 1, 3) = 1

41  10        phi( group 3, 1) is 1 because of 0 sample size
42  10        phi( group 3, 2) is 1 because of 0 sample size
12  10        pent(*,1) = 1
13 -10        pent(8,rest) = 0
0 0 0 0 0 0 0       end of design matrix specification




!-------------------------------------------------------------------
!Example: European Dippers
! 
!   This example shows how to read in the capture histories instead of summary statistics,
!   the specification of absolute sample times
!
EUROPEAN DIPPER, MALES AND FEMALES
2 7 notrace hist mod design noinit
1 2 3 4 5 6 7    absolute sample times
1111110  1 0     histories must start in column 1
1111100  0 1 
1111000  1 0 
1111000  0 1 
1101110  0 1 
1100000  1 0 
1100000  1 0 
1100000  1 0 
1100000  1 0 
1100000  0 1 
1100000  0 1 
1010000  1 0 
1010000  0 1 
1000000  1 0 
1000000  1 0 
1000000  1 0 
1000000  1 0 
1000000  1 0 
1000000  0 1 
1000000  0 1 
1000000  0 1 
1000000  0 1 
0111111  0 1 
0111111  0 1 
0111110  0 1 
0111100  1 0 
0111100  0 1 
0111100  0 1 
0111000  1 0 
0111000  0 1 
0110110  0 1 
0110000  1 0 
0110000  1 0 
0110000  1 0 
0110000  1 0 
0110000  1 0 
0110000  1 0 
0110000  1 0 
0110000  0 1 
0110000  0 1 
0110000  0 1 
0110000  0 1 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  1 0 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0100000  0 1 
0011111  0 1 
0011111  0 1 
0011110  1 0 
0011110  0 1 
0011100  1 0 
0011100  1 0 
0011100  1 0 
0011100  1 0 
0011100  0 1 
0011100  0 1 
0011000  1 0 
0011000  1 0 
0011000  1 0 
0011000  1 0 
0011000  1 0 
0011000  1 0 
0011000  1 0 
0011000  1 0 
0011000  0 1 
0011000  0 1 
0011000  0 1 
0011000  0 1 
0010110  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  1 0 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0010000  0 1 
0001111  1 0 
0001111  1 0 
0001111  1 0 
0001111  1 0 
0001111  1 0 
0001111  1 0 
0001111  0 1 
0001111  0 1 
0001110  1 0 
0001110  1 0 
0001110  1 0 
0001110  0 1 
0001110  0 1 
0001110  0 1 
0001110  0 1 
0001100  1 0 
0001100  1 0 
0001100  1 0 
0001100  1 0 
0001100  1 0 
0001100  1 0 
0001100  0 1 
0001100  0 1 
0001100  0 1 
0001100  0 1 
0001100  0 1 
0001011  0 1 
0001001  1 0 
0001001  0 1 
0001000  1 0 
0001000  1 0 
0001000  1 0 
0001000  1 0 
0001000  1 0 
0001000  1 0 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0001000  0 1 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  1 0 
0000111  0 1 
0000111  0 1 
0000111  0 1 
0000111  0 1 
0000111  0 1 
0000111  0 1 
0000110  1 0 
0000110  1 0 
0000110  1 0 
0000110  0 1 
0000110  0 1 
0000110  0 1 
0000110  0 1 
0000110  0 1 
0000110  0 1 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  1 0 
0000100  0 1 
0000100  0 1 
0000100  0 1 
0000100  0 1 
0000100  0 1 
0000100  0 1 
0000100  0 1 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  1 0 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000011  0 1 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  1 0 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000010  0 1 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  1 0 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000001  0 1 
0000000  0 0     ****** Notice the 0000000 record to indicate the end of the histories *******
g*t g*t g*t
0        There are 0 beta terms
 1  10    p(1,1) = 1   Because PIM not specified, you must know the internal PIM numbers
 7  10    p(1,7) = 1
21  10    p(2,1) = 1
27  10    p(2,7) = 1
0 0 0 0 0 0 0       end of design matrix specification



 ----------------------------------------------------------------------------------------------------

Revision History

31 Aug     00 - implemented cycling, but this can be turned off with trace(11)
                fixed bug on reading initial values

10 July    00 - implemented full design matrix specifications
                dropped all user constraint coding (as this is superseded by design matrix method)
                added feature for comment lines in input deck
                adjustments per unit time for phi added. Trace(14) turns this off

28 Feb     00 - converted all source to f90 using a "simple" conversion program from the web

13 Jan     00 - added PIM. Cleaned up code for restrictions

6 Oct      99 - Checked and added code for AUC estimes, stream residence time etc
                Added code to allow user to specify initial values

15 June    99 - added code to compute the phi_0 in the age-specific breeding proportion
                models. 

1 June     99 - added code to compute lambda estimates and to fit models where
                lambda is constrained to be equal across time values

1  Dec     98 - added code to compute gamma estimates of Pradel and their standard
                errors and to compute age specific breeding proportions and their
                standard errors.

25 May     97 - Fixed bug that prevented data with n(i)=0 from being 
                fit. 

19 June    96 - Bug fix to uncond and modres for nobirth models.

31 August  95 - Added the three code model selection replacing old method
                of specifying a single model number.
                Use the SVD decomposition to remove redundant constraints
                to avoid the singular matrix problem in general.  

11 April   95 - Fixed a bug that gave a singular matrix when you  specify a constraint 
                that pent(j)=1. Routine modified was res_pent01.

 3 April   95 - Added trace(16) flag for adjusting variances by conditioning on
                B. This would work if there are no constraints on
                the beta (excluding constraints at zero) but there
                doesn't seem to be any way of implementing the 
                correction when there are constraints on betas
                (e.g., beta(2)=beta(3), etc).

28 January 95 - change all include pointers to Include library
              - change all print statements to write(6, ) statements

9 December 94 - README file created.