Games of Daze/Infomagic - Games of Daze (Summer 1995) (Disc 1 of 2).iso/x2ftp/msdos/ai/tierra40/doc/diverse.doc

Text File | 1992-09-10 | 13KB | 313 lines

====== DIVERSE DOCUMENTATION =================================================

Usage:  The diverse tool will prompt you for input parameters.
        Start by typing "diverse" to the prompt.
        To accept default values, hit Enter.

What Diverse Does:

     Diverse reads the files of birth and death records (break.X) that are
written to disk by the Tierra simulator, and prepares new files that
contain several indices of diversity.  These indices may be recomputed and
output after each birth or death, or they may be averaged over an interval
of your choice and output at the end of each interval (this keeps the volume
of output down to reasonable levels).

Diverse reads the break.1 ... break.X files that are written to the \td
directory, and outputs files whose default names are averages, divdat.X,
divrange and brkrange.  The files divdat.X contain eight parameters
related to diversity and turn over of size classes and genotypes in the soup.

The file divrange contains the ranges and averages of the eight variables for
the entire run.  The file brkrange contains the ranges of the variables for
each of the divdat.X files.  Diverse takes a long time to run, so it prints a
statement after every million Tierran instructions processed, just to let you
know it is still ticking.  Diverse does a lot of floating point calculations,
so it will be very slow if it is run on any machine without a math
co-processor.

To compile diverse on a Unix system: cc diverse.c -lm -o diverse

There are two basic dialog scenarios:

input file (default: break.1) =
threshold (default: 2) =
Frequency for Average Output (default: 1000000) =
output birth death records: y or n (default: n) =

and:

input file (default: break.1) =
threshold (default: 2) =
Frequency for Average Output (default: 1000000) =
output birth death records: y or n (default: n) = y
output file (default: divdat.1) =
output format: binary or ascii (default: binary) =
break size (default: 4096) =

In both scenarios, you must respond to the first four questions:

input file (default: break.1) =
threshold (default: 2) =
Frequency for Average Output (default: 1000000) =
output birth death records: y or n (default: n) = y

If you choose to output diversity indices for each birth and death, then
you must answer three more questions:

output file (default: divdat.1) =
output format: binary or ascii (default: binary) =
break size (default: 4096) =

====== DIVERSE ===============================================================

     The following is an explanation of each of the lines of the dialog:

input file (default: break.1) =

     The name of the file containing the birth and death records, which
will be read by the diverse program.  The default is break.1, which is
the default name of the file output by the Tierra program.

threshold (default: 2) =

     The threshold for including size or genotype classes in the calculations
of the diversity indices.  The default is 2, which means that genotypes for
which there is only a single individual (probably mostly inviable mutants)
will not be included in the calculations.

Frequency for Average Output (default: 1000000) =

     If this parameter is non-zero, averages for the diversity indices will
be calculated over the interval indicated, and output only at then end of
each interval.  The default is once every million instructions.

output birth death records: y or n (default: n) =

     This line asks if diversity indices should be output for each record of
birth and death.  The default is no because this generates a huge volume
of data.  The file produced by this option will be about four times the
size of the break files used as input.

output file (default: divdat.1) =

     If one chooses to output diversity indices for each birth and death
record, this option allows you to specify the name of the output file(s).
The default is divdat.1.

output format: binary or ascii (default: binary) =

     There are two possible file formats for the file containing diversity
indices for each birth and death record: binary and ascii.  The default is
binary, because the binary format is more compact than the ascii format.  This
is important because the even the binary output file (divdat.X) will be about
four times as large as the input file (break.X).

break size (default: 4096) =

     If diversity indices are output for each birth and death, they will be
broken into a series of files, and this variable specifies the size of the
files in K (1024 bytes).  The default size is 4 Mb.  If there is more than
4 Mb of output, the file will be broken into pieces.

====== DIVERSE OUTPUT FORMATS ================================================

from divdat.1:

11 0.0
0 1 1 0 0 1 0 0
33a 2 1 0 33a 1 0 33a
62c 3 1 0 966 1 0 966
ac 4 1 0 a12 1 0 a12
b6e 5 1 0 1580 1 0 1580
57 6 1 0 15d7 1 0 15d7
...
34f 341 32 1.98219 20795c 163 4.43486 72ed46
216 342 32 1.99079 207b72 164 4.44187 724167
70 343 32 1.98692 207be2 165 4.44885 72aa9b
2 342 32 1.99079 207be4 165 4.44592 72aa9d
21e 343 32 1.98692 207e02 165 4.43891 72acbb
2 342 32 1.98663 207e04 165 4.445 72acbd
41e 341 32 1.99052 208222 164 4.438 724837

from divdat.2:

11 12472882.0
56 342 32 1.98663 208278 164 4.44095 72488d
1e3 341 32 1.97801 20845b 163 4.43394 7299d8
d26 342 32 1.97415 209181 164 4.44095 730fc0
2 341 32 1.96972 209183 163 4.43394 739114
355 340 32 1.9736 2094d8 162 4.42689 73252f
1a1 339 32 1.97749 209679 162 4.42544 7326d0
...
50 362 27 2.17167 3e4a3d 156 4.35437 a084a7
6a 363 27 2.16879 3e4aa7 156 4.34815 a08511
376 362 27 2.16406 3e4e1d 155 4.34112 a09769
0 363 27 2.16879 3e4e1d 156 4.34815 a08882
76 364 27 2.16591 3e4e93 157 4.35515 9fd231
2 363 27 2.16118 3e4e95 156 4.34815 a0a811
5a 362 27 2.16406 3e4eef 156 4.34803 a0a86b

     Note that each divdat.X file begins with a line that specifies the
format and starting time of that file.  The first byte specifies if the
file is binary (0) or ascii (1).  The second byte specifies if the file
contains data on genotype diversity (1) or size diversity only (0).  After
the two format bytes, there is a double precision floating point number
which specifies the start time, in Tierran instructions of this file
(actually the end time of the previous divdat.X file, or zero for the first
file).

     The following code writes the format line of the divdat.X files:

    if(format) /* ASCII format */
    {   ouf = fopen(ofile,"w");
        if(genotypes)
            fprintf(ouf,"11 %.1lf\n", totals.time);
        else
            fprintf(ouf,"10 %.1lf\n", totals.time);
    }
    else /* binary format */
    {   ouf = fopen(ofile,"wb");
        c = (char) format;
        fwrite(&c,sizeof(char),1,ouf);
        c = (char) genotypes;
        fwrite(&c,sizeof(char),1,ouf);
        fwrite(&totals.time,sizeof(double),1,ouf);
    }

     After the format line, the divdat.X files contain an eight column data
stream (if genotypes are included), or a five column output if only size
class data is included.  A typical output line from an ASCII file looks like
the following:

34f 341 32 1.98219 20795c 163 4.43486 72ed46

     The first column (Time) is the time increment since the last birth or
death in hexadecimal format.  In order to determine the actual time associated
with a particular output line, the values in the first column must be summed.

     The second column (NumCell) is the total population, the number of cells,
in decimal format.  The third column (NumSize) is the number of size classes,
in decimal format.  The fourth column (SizeDiv) is the size diversity as a
floating point number.  Size diversity is the negative sum of p log p, where
p is the proportion of the population occupied by a particular size class,
summed over all size classes.  The fifth column (AgeSize) is the average age
of the size classes, as a hexidecimal number.  When a new size class appears,
or an extinct size class reappears, its age is set to zero.  The age then
increments with time.  At each time increment, the age of all size classes
is summed, and divided by the number of size classes to compute the average
age of size classes.

     If genotype data is displayed, the the sixth column (NumGeno) contains
the number of genotypes as a decimal number.  The seventh column (GenoDiv)
contains the genotype diversity as a floating point number.  Genotype
diversity is the negative sum of p log p, where p is the proportion of the
population occupied by a particular genotype class, summed over all genotype
classes.  The eighth column (AgeGeno) contains the average age of the
genotypes as a hexidecimal number.  When a new genotype class appears, or an
extinct genotype class reappears, its age is set to zero.  The age then
increments with time.  At each time increment, the age of all genotype classes
is summed, and divided by the number of genotype classes to compute the
average age of genotype classes.

     The code that produces the data lines in the divdat.X files follows:

struct siz {
    long   Time;
    long   NumCell;
    long   NumSize;
    float  SizeDiv;
    long   AgeSize;
    } si;

struct gen {
    long   NumGeno;
    float  GenoDiv;
    long   AgeGeno;
    } ge;

    if(format) /* ASCII format */
    {   if(genotypes)
            fsize += 1 + fprintf(ouf,"%lx %ld %ld %g %lx %ld %g %lx\n",
            rlo.itime, totals.TotPop, divdat.NumSize, divdat.SizeDiv,
            (Ulong) AgeSize, divdat.NumGeno, divdat.GenoDiv,
            (Ulong) AgeGeno);
        else fsize += 1 + fprintf(ouf,"%lx %ld %ld %g %lx\n",
            rlo.itime, totals.TotPop, divdat.NumSize, divdat.SizeDiv,
            (Ulong) AgeSize);
    }
    else /* binary format */
    {   si.Time = rlo.itime;
        si.NumCell = totals.TotPop;
        si.NumSize = divdat.NumSize;
        si.SizeDiv = divdat.SizeDiv;
        si.AgeSize = (Ulong) AgeSize;
        fwrite(&si, sizeof(struct siz), 1, ouf);
        fsize += sizeof(struct siz);
        if(genotypes)
        {   ge.NumGeno = divdat.NumGeno;
            ge.GenoDiv = divdat.GenoDiv;
            ge.AgeGeno = (Ulong) AgeGeno;
            fwrite(&ge, sizeof(struct gen), 1, ouf);
            fsize += sizeof(struct gen);
        }
    }

from brkrange:

     divdat.1 0 0 0 0 0 0 0 0
     divdat.1 12472882 351 32 2.01064 3456383 166 4.45098 7989488
     divdat.2 23007672 456 36 2.27044 4583906 212 4.78146 10726890
     divdat.3 32218752 456 43 2.71956 5252442 240 5.19617 14721618
     divdat.4 41767554 465 51 2.95793 7448691 240 5.19617 20849185
     divdat.5 48183196 467 51 2.95793 7722602 242 5.19617 22182141

     The brkrange file contains lines with the same eight numeric columns
as the lines in the divdat.X files.  However, in the brkrange file, all
columns are expressed in decimal format (no hexidecimal), and each entry
represents the cumulative maximum value of the corresponding entry at the
end of the divdat.X file named in the first column.  An exception is that
there are two lines for the divdat.1 file, the first line represents the
values of the eight variables at the beginning of the file, while the
second line represents the maximum values of each variable at the end
of the file.  Another exception is that the first numeric column represents
the cumulative time at the end of the file (not the maximum time increment).
The data is provides so that individual divdat.X files can be plotted, and
the user can know the range of the variables when configuring the plot.

from divrange:

           Minimum      Maximum      Average Number

Time             0    103717758              226162
NumCell          0          374        262.2 226162
NumSize          0           21          9.9 226162
SizeDiv   0.000000     2.503943     1.177503 226162
AgeSize          0     26892940   12170879.1 226162
NumGeno          0           92         56.8 226162
GenoDiv   0.000000     4.346522     3.372861 226162
AgeGeno          0      6710709    3274868.8 226162

     The divrange file is fairly self explanatory.  It lists the minimum,
maximum and average values for each of the eight variables computer by
the diversity tool.  The last column shows the sample size (the number of
births and deaths that went into the computation).

from averages:

       Time NumCell NumSize SizeDiv    AgeSize NumGeno GenoDiv    AgeGeno

    1001395   230.7     1.7    0.05     436052    13.9    0.86     321276
    2000507   279.0     5.9    0.27     925677    32.4    1.63     823193
    3000940   274.8     6.8    0.35    1354576    36.8    1.83    1220080
    4001155   276.5     5.3    0.34    2059764    38.8    1.96    1524246
    5000097   274.8     7.4    0.52    2147820    35.4    1.86    1681769
    6000879   274.8     7.3    0.79    2476165    36.6    2.08    1841812
    7000084   277.1     6.8    0.96    3315557    39.7    2.30    1713698
    8001588   304.5     8.1    0.97    3324491    40.6    2.32    1676695

     The averages file is also self-explanatory.  It lists the average
value of each parameter over a user defined interval.  The time at the
end of the interval is shown in the first column.  All numbers are expressed
in decimal format.