Next: , Previous: , Up: test   [Contents][Index]


Performing the Interaction

Performing the Interaction

A test of algorithm efficiency that consists of num_passes passes with a new random deterministic finite automaton, which has defined numbers of input, output signals, and states, being chosen at each pass, is performed by command

$ ./test -t num_passes [ -C VAL ] [ additional options ]  \
  num_in_sig num_out_sig num_states

If the option -C VAL with non-zero VAL is specified, then best cycles will be found in automaton state graphs which will be provided to be connected ones. The command prints a test log, an example of which is shown below.

$ ./test -t10 -Ccs 20 20 20
    DFA inputs: 20
   DFA outputs: 20
    DFA states: 20
 Input signals: dfa-state
        Passes: 10
Steps per pass: 10000
         Large: off
 R. prob. type: 1
 N-gram length: 1
 Algorithms v.: 0
       K*temp.: 1.000000000000000E+00
   Max. cycles: -2
   Random seed: 0

pass       earned     random      maximal cl    % efr   % efa
---- ------------ ---------- ------------ -- -------- -------
   1     5821.000    331.000     7500.000  4 1758.610  76.580
   2     5826.000    441.000    10000.000  2 1321.088  56.334
   3     5656.000    359.000     6666.667  3 1575.487  83.977
   4     6443.000    414.000    10000.000  2 1556.280  62.894
   5     7075.000    462.000    10000.000  2 1531.385  69.333
   6     5949.000    329.000     7142.857  7 1808.207  82.479
   7     4272.000    281.000     6666.667  3 1520.285  62.499
   8     8781.000    411.000    10000.000  2 2136.496  87.288
   9     8144.000    648.000    10000.000  2 1256.790  80.154
  10     4671.000    270.000     6000.000  5 1730.000  76.806

TOTL    62638.000   3946.000    83976.190  3 1587.380  73.337

stddev efr:  253.554
stddev efa:   10.486

The log begins with a description of test modes used. Then there goes a table, where each row corresponds to a test pass, with a summary row. The columns of the table have the following meaning.

pass

A test pass index. For the summary row ‘TOTL’ is printed here.

earned

The amount of spur the system has received during interaction with the automaton according to the algorithm. In the summary row—the sum of values in the column.

random

The amount of spur the system has received during random interaction with the automaton when output signals are equiprobably chosen from the set of allowed signals. In the summary row—the sum of values in the column.

maximal

The amount of spur the system would receive during the most optimal interaction with the automaton. Will be determined if the option -C VAL with non-zero VAL is specified in the command line, otherwise will be equal to the number of steps at a test pass. In the summary row—the sum of values in the column.

cl

The length of a cycle, continuous repeating of which would give the maximum amount of spur to the system. Will be determined if the option -C VAL with non-zero VAL is specified in the command line, otherwise will be equal to 0. In the summary row—the average length of cycles for all test passes.

% efr

Relative efficiency equal to (earned/random)*100%. For the summary row is calculated using corresponding values in that row.

% efa

Actual efficiency equal to (earned-random)/(maximal-random)*100%. For the summary row is calculated using corresponding values in that row.

A standard deviation of values in ‘% efr’ and ‘% efa’ columns is printed at the end of the log.


Next: , Previous: , Up: test   [Contents][Index]