GPM model page

This package GPM_Model describes the specialized classed for GPM_Core package used in GPM_Gui packages.

This package is the core package for GUI interface.

It enables to generate graph from :

• the init graph file
• the pattern graph file
• the transformer graph file
• the rules graph file.

The main class of the package is GPM_ModelRun class.

To use the program (see Program's help ) , type

gpm_model.exe --help

somme examples:

• plant evolution (see GPM plant Example)
• refinment mesh (see GPM refinment Example)

Program's help

gpm_model.exe [options] make [args] where:

• options={–optionName=optionValue}
• args= [test | run | rewriting | isomorph ]

Options are :

• For help
  • help : print this help, no value
• Data args:
  • path : where to find the needed files
  • data for rewriting graph
    • init: file name of the initial graph (init.txt by default)
    • nRules : number of rules (1 by default)
    • pattern_i : file name of the pattern graph of the rule i (pattern.txt by default)
    • transform_i : file name of the transformer graph of the rule i (transform.txt by default)
    • rule_i : file name of rules description of the rule i (transform.txt by default)
    • iterationsNumber: number of iterations of the rewriting graph
  • data for testing isomorphism between graph
    • graph1 : file name of the graph 1
    • graph2 : file name of the graph 2
  • output args:
    • output : output path
    • prefix : prefix of the output files
    • verbose : if empty, print on the screen, otherwise print on indicated file
    • verbose-mode : (error,warning,debug)

Args are

• test : make the primary tests
• run : execute the rewriting graph : generate graph files with prefix in output path
• rewriting : execute the rewriting graph : generate graph files with prefix in output path
• isomorph test if the 2 graphs are isomorph

The graph file format is as follow:

#N vertices
idN  Type        variableName=value
#P edges
idE  sourceId        targetId EDGE

where :

• N is the number of vertices
• P is the number of edges
• idN is the identity the the vertex
• idE is the identity the the edge
• Type is the type of vertex either NODE or PORT
• variableName is the variable name mCoord, state for example
• value is the value of the variable:
  • for a string variable, put the value in "" for example "+"
  • for a real, integer, put the value as usual
  • for a boolean, set the value as true or false

For example,

#4 vertices
0       NODE    mCoord=[0;0;0] state=0
1       PORT    state=0.1
2       PORT    state=0.1
3       NODE    mCoord={[0;1;0]} state=1
# 3      edges
0 0       1       EDGE
1 1       2       EDGE
2 2       3       EDGE

The rules format is as follow:

#list of ids of vertices of pattern graph to remove
N
id_i
#list of ids of edges of pattern graph to remove
P
id_j
#list of edges between Pattern graph & Transform graph to add
Q
source_k  target_k
#group node ids rules
groupNodeIds=varName1[+varName2]
#group port ids rules
groupPortIds=varName1[+varName2]
#mapping rules for pattern vertex
R
IsUpToAutomorphism=true
boolean_condition_r(pattern graph variables)
#state update
Q
TvariableName_id=fct(pattern graph_variable,transform graph variables defined before the rules)
PvariableName_id=fct(pattern graph_variable)

where :

• N is the number of pattern vertex Ids to remove from the mapping pattern graph into large graph
• id_i is the id of the vertex to remove
• P is the number of pattern edge Ids to remove form the mapping pattern graph into large graph
• id_j is the id of the edge to remove
• Q is the number of adges to add form pattern graph vertex to transform graph vertex
• source_k is the source vertex from the pattern graph of the edge to add between pattern & transform graph
• target_k is the target vertex from the transform graph of the edge to add between pattern & transform graph
• groupeNodeIds is the variable to build the id of the node used for the maching to select & speed up the pattern graph mapping search. It consists in the adding of the value of the variable name.
• groupPortIds is the variable to build the id of the port used for the maching to select & speed up the pattern graph mapping search. It consists in the adding of the value of the variable name.
• R is the number of the mapping rules which can be either
  • IsUpToAutomorphism=true | false for dis/en-able the permuation of the same pattern graph mapping
  • a boolean expression depending on variables of pattern graph to be true to validate a mapping
• Q is the number of rules to update the state of pattern or transform vertices of the graph of the form:
  • varName_{T,P}_id=fct(varName_{T,P}_id) where
    • varName is a variable name of the vertex
    • T is the letter to indicate the vertex belongs to the transform graph
    • P is the letter to indicate the vertex belongs to the pattern graph
    • id is the id of the vertex.
    • fct i a function depinding on pattern vertices or transformer vertices.
  • note that the order of the rules in important to re-use a preceeding computing variable value.

For example,

#list of ids of vertices of pattern graph to remove
0 
#list of ids of edges of pattern graph to remove
0
#list of edges between Pattern graph & Transform graph to add
3
3,4
3,6
3,8
#group node ids rules
groupNodeIds=state
#group port ids rules
groupPortIds=
#mapping rules for pattern vertex
0
#state update 
5
P_3=2*P_3
T_0=rotation(P_3-P_0,-PI/3)+2*P_3
T_1=rotation(P_3-P_0,PI/3)+2*P_3
T_2=(P_3-P_0)+2*P_3
state_P_3=0

To compute the rewriting graph whith 3 rules & 3 iterations, execute the command:

gpm_model.exe --verbose-mode=debug  --path=./data  --output=./output --prefix=plant  --init=plant-init.txt --iterationsNumber=3 --verbose=./output/plant.log --nRules=3  --rule_1=plant-rule01.txt --pattern_1=plant-pattern01.txt --transform_1=plant-transform01.txt --rule_2=plant-rule02.txt --pattern_2=plant-pattern02.txt --transform_2=plant-transform02.txt --rule_3=plant-rule03.txt --pattern_3=plant-pattern03.txt --transform_3=plant-transform03.txt make run