source: java/main/src/test/resources/info/Neuro.info @ 1095

Last change on this file since 1095 was 86, checked in by psniegowski, 11 years ago

HIGHLIGHTS:

  • use java annotations to mark classes and fields to be used when:
    • using java classes with ReflectionAccess? to represent remote objects with FramsClass? description found by "info ..." requests
    • to build up FramsClass? representation of objects not present at remote server
  • allow using primitive types (instead of wraping counterparts) in reflected classes
  • rework FramsClass? creation process (add FramsClassBuilder?)
  • add more tests

CHANGELOG:
Prepare model.World class.

Minor change.

Use primitive types for Genotype and Creature classes.

Use primitive types in model.Neuro* classes.

Use primitive types in model.Joint* classes.

Use primitive types in model.Part* classes.

Fix primitive values.

Extract FramsClassBuilder?.

Add tests of Model classes.

More fixes.

Refactorize out ParamCandidate?.

Several fixes.

Fix all regressions after introducing annotations.

Use annotations throughout the project.

Add exception classes.

Improve creation of FramsClass?.

More changes.

Many changes regarding annotations.

Annotate classes in com.framsticks.model package.

Remove manual FramsClass? constructor.

Construct FramsClass? for Creature. Add test.

Add default values to the ParamAnnotation?.

Add ParamBuilderTest? and ParamAnnotation?.

Add FramsClassAnnotation?.

File size: 2.6 KB
Line 
1class:
2id:Neuro
3
4prop:
5name:Get input signal
6id:getInputState
7type:p
8flags:0
9
10prop:
11name:Get input weight
12id:getInputWeight
13type:p
14flags:0
15
16prop:
17name:Get weighted input signal
18id:getWeightedInputState
19type:p
20flags:0
21
22prop:
23name:Get signal sum
24id:getInputSum
25type:p
26flags:0
27
28prop:
29name:Get weighted signal sum
30id:getWeightedInputSum
31type:p
32help:Uses any number of inputs starting with the specified input. getWeightedInputSum(0)=weightedInputSum
33flags:0
34
35prop:
36name:Get input count
37id:getInputCount
38type:d
39flags:0
40
41prop:
42name:Full signal sum
43id:inputSum
44type:f
45flags:0
46
47prop:
48name:Full weighted signal sum
49id:weightedInputSum
50type:f
51flags:0
52
53prop:
54name:Get channel count for input
55id:getInputChannelCount
56type:p
57flags:0
58
59prop:
60name:Get input signal from channel
61id:getInputStateChannel
62type:p
63flags:0
64
65prop:
66name:Get weighted input signal from channel
67id:getWeightedInputStateChannel
68type:p
69flags:0
70
71prop:
72name:Neuron state (channel 0)
73id:state
74type:f
75help:~
76When read, returns the current neuron state.
77When written, sets the 'internal' neuron state that will become current in the next step.
78Typically you should use this field, and not currState.~
79flags:0
80
81prop:
82name:Number of output channels
83id:channelCount
84type:d
85flags:0
86
87prop:
88name:Get state for channel
89id:getStateChannel
90type:p
91flags:0
92
93prop:
94name:Set state for channel
95id:setStateChannel
96type:p
97flags:0
98
99prop:
100name:Hold state
101id:hold
102type:d 0 1
103help:"Holding" means keeping the neuron state as is, blocking the regular neuron operation. This is useful when your script needs to inject some control signals into the NN. Without "holding", live neurons would be constantly overwriting your changes, and the rest of the NN could see inconsistent states, depending on the connections. Setting hold=1 ensures the neuron state will be only set by you, and not by the neuron. The enforced signal value can be set using Neuro.currState before or after setting hold=1. Set hold=0 to resume normal operation.
104flags:0
105
106prop:
107name:Current neuron state (channel 0)
108id:currState
109type:f
110help:~
111When read, it behaves just like the 'state' field.
112When written, changes the current neuron state immediately, which disturbs the regular synchronous NN operation.
113This feature should only be used while controlling the neuron 'from outside' (like a neuro probe) and not in the neuron definition. See also: Neuro.hold~
114flags:0
115
116prop:
117name:Set current neuron state for channel
118id:setCurrStateChannel
119type:p
120help:Analogous to "currState".
121flags:0
122
123prop:
124name:Position x
125id:position_x
126type:f
127flags:0
128
129prop:
130name:Position y
131id:position_y
132type:f
133flags:0
134
135prop:
136name:Position z
137id:position_z
138type:f
139flags:0
140
Note: See TracBrowser for help on using the repository browser.