Changeset 1227 for cpp/frams/genetics/f4/f4_general.h

Timestamp:

04/27/23 04:04:06 (2 years ago)

Author:

Maciej Komosinski

Message:

Improvements in f4:

• fixed a bug where newly created cells in a given development step were not counted as in-active-development (overlooked), and if they were the only in-active-development cells, the development of an organism would stop
• added one extra development step (#ifdef EXTRA_STEP_CELL_DEVELOPMENT) so that cells that became not in-active-development ("halted" or yielding, usually due to waiting for neurons to develop in other cells) would get a chance to continue development (important when we don't want to ignore invalid neuron connections, #ifdef TREAT_BAD_CONNECTIONS_AS_INVALID_GENO)
• ensured that all connections in a cell are processed (earlier they could be skipped if the development of the cell was "halted" and all cells became not in-active-development)
• got rid of neuron connection syntax [sensor:weight], now all neuron classes are handled in a uniform way and their [connections] too; the only allowed syntax is [input_index:weight]
• unified handling of all neuroclasses during parsing, conversion and mutation
• more correct syntax coloring
• got rid of general-purpose fields (i1, i2, f1, s1) in class f4_node - now separate fields serve their individual purpose
• rewritten creating and modifying neuron connections - it is more deliberate to satisfy neuron input/output preferences
• some invalid neuron connections make a genotype invalid (previously invalid neuron connections were ignored and the genotype was considered valid)
• added (surprisingly missing) simple debug printout functions to see the f4_Node tree structure and the developing f4_Cells
• more informative variable and constant names
• improved performance

File:

• cpp/frams/genetics/f4/f4_general.h (modified) (32 diffs)

cpp/frams/genetics/f4/f4_general.h

@@ -1 +1 @@
 // This file is a part of Framsticks SDK.  http://www.framsticks.com/
-// Copyright (C) 1999-2015  Maciej Komosinski and Szymon Ulatowski.
+// Copyright (C) 1999-2023  Maciej Komosinski and Szymon Ulatowski.
 // See LICENSE.txt for details.
 
@@ -29 +29 @@
 void rolling_inc(double *v);
 
-class f4_node;   // later
+class f4_Node;   // later
 class f4_Cell;   // later
 class f4_Cells;  // later
@@ -36 +36 @@
 /** @name Types of f4_Cell's */
 //@{
-#define T_UNDIFF4 40 ///<undifferentiated cell
-#define T_STICK4  41 ///<differentiated to stick, cannot divide
-#define T_NEURON4 42 ///<differentiated to neuron, can divide
+#define CELL_UNDIFF 40 ///<undifferentiated cell
+#define CELL_STICK  41 ///<differentiated to stick, cannot divide
+#define CELL_NEURON 42 ///<differentiated to neuron, can divide
 //@}
 
@@ -54 +54 @@
 
 
-class f4_CellLink;
+class f4_CellConn;
 
 /** @name Constraints of f4 genotype structures */
 //@{
-#define MAXINPUTS 100 ///<maximum number of neuron inputs
-#define MAX4CELLS 100 ///<maximum number of f4 organism cells
+#define F4_MAX_CELL_INPUTS  10 ///<maximum number of neuron inputs in a developing organism
+#define F4_MAX_CELLS 100 ///<maximum number of f4 organism cells
 //@}
 
 /**
- * Abstract cell type - the representation of single component in the  developmental
+ * Abstract cell type - the representation of a single component in the developmental
  * encoding. In the beginning, each f4_Cell is undifferentiated. During the process
  * of development it can divide or differentiate into a stick or a neuron. If it
@@ -86 +86 @@
                 /**
                  * A constructor that takes the pointer to the repetition node and the count of repetitions.
-                 * @param a pointer to f4_node for repetition character
+                 * @param a pointer to f4_Node for repetition character
                  * @param b the number of repetitions
                  */
-                repeat_ptr(f4_node *a, int b) : node(a), count(b) { };
+                repeat_ptr(f4_Node *a, int b) : node(a), count(b) { };
 
                 inline void makeNull() { node = NULL; count = -1; };
@@ -96 +96 @@
 
                 inline void dec() { count--; };
-                f4_node    *node; ///<pointer to the repetition code
+                f4_Node    *node; ///<pointer to the repetition code
                 int       count; ///<repetition counter
         };
@@ -130 +130 @@
                 static const int stackSize = 4;  ///<max 4 nested levels
                 repeat_ptr ptr[stackSize]; ///<array holding pointers to repeat_ptr
-                short int top;  ///<index of the top of the stack
+                int top;  ///<index of the top of the stack
         };
 
         /**
          * Creates a new f4_Cell object.
-         * @param nname name of a cell, can be T_UNDIFF4, T_STICK4 or T_NEURON4
+         * @param nnr number of the cell
          * @param ndad pointer to the parent of the created cell
          * @param nangle the amount of commas affecting branch angles
          * @param newP genotype properties of a given cell
          */
-        f4_Cell(int nname, f4_Cell *ndad, int nangle, GeneProps newP);
+        f4_Cell(int nnr, f4_Cell *ndad, int nangle, GeneProps newP);
         /**
          * Creates a new f4_Cell object.
          * @param nO pointer to an organism containing the cell
-         * @param nname name of the cell, can be T_UNDIFF4, T_STICK4 or T_NEURON4
+         * @param nnr number of the cell
          * @param ngeno pointer to the root of the genotype tree
-         * @param ngcur pointer to the f4_node representing the current cell in the genotype tree
+         * @param ngcur pointer to the f4_Node representing the current cell in the genotype tree
          * @param ndad pointer to the parent of the created cell
          * @param nangle the number of commas affecting branch angles
          * @param newP genotype properties of a given cell
          */
-        f4_Cell(f4_Cells *nO, int nname, f4_node *ngeno, f4_node *ngcur, f4_Cell *ndad, int nangle, GeneProps newP);
+        f4_Cell(f4_Cells *nO, int nnr, f4_Node *ngeno, f4_Node *ngcur, f4_Cell *ndad, int nangle, GeneProps newP);
 
         ~f4_Cell();
@@ -172 +172 @@
          *  - the stick modifiers, like rotation, will be applied on neuron cell,
          *  - the differentiated cell will be differentiated again,
-         *  - the neuron class inside cell connection (i.e. N[G:5]) is not a sensor,
          *  - the connection between neurons cannot be established,
          *  - the neuron class is not valid.
@@ -178 +177 @@
          * @return 0 if development was successful, 1 if there was an error in genotype tree
          */
-        int onestep();
-
-        /**
-         * Adds a link between this neuron cell and a given neuron cell in nfrom. If the nfrom object
-         * is not given, neuron type in nt should be a sensor type.
-         * @param nfrom input neuron cell, or NULL if not given
-         * @param nw weight of connection
-         * @param nt empty string or name of sensor class
-         * @return 0 if link is established, -1 otherwise
-         */
-        int   addlink(f4_Cell *nfrom, double nw, string nt);
+        int oneStep();
+
+        /**
+         * Adds a connection between this neuron cell and a given neuron cell in nfrom.
+         * @param nfrom input neuron cell
+         * @param nweight weight of connection
+         * @return 0 if connection is established, -1 otherwise
+         */
+        int   addConnection(f4_Cell *nfrom, double nweight);
 
         /**
@@ -195 +192 @@
         void  adjustRec();
 
-        int        name;               ///<name of cell (number)
+        int        nr;                 ///<number of cell (seems to be used only in old f1 converter for neuron connections)
         int        type;               ///<type
         f4_Cell *dadlink;              ///<pointer to cell parent
         f4_Cells  *org;                ///<uplink to organism
 
-        f4_node *genot;                    ///<genotype tree
-        f4_node *gcur;                 ///<current genotype execution pointer
-        int active;                    ///<determines whether development is still active
+        f4_Node *genot;                    ///<genotype tree
+        f4_Node *gcur;                 ///<current genotype execution pointer
+        bool active;                   ///<determines whether development is still active; even if false, the cell may "yield" - may be halted (but still having its onStep() called) due to neural connections waiting for other cells to potentially develop neurons
         repeat_stack repeat;           ///<stack holding repetition nodes and counters
         int recProcessedFlag;          ///<used during recursive traverse
@@ -220 +217 @@
         int          neuro_refno;      ///<the number of the neuro object, used in f0
 
-        int          ctrl;             ///<neuron type
         double       inertia;          ///<inertia of neuron
         double       force;            ///<force of neuron
         double       sigmo;            ///<sigmoid of neuron
-        f4_CellLink *links[MAXINPUTS]; ///<array of neuron links
-        int          nolink;           ///<number of links
+        f4_CellConn *conns[F4_MAX_CELL_INPUTS]; ///<array of neuron connections
+        int          conns_count;      ///<number of connections
         NeuroClass *neuclass;          ///<pointer to neuron class
 };
 
 /**
- * Class representing link between neuron cells.
- */
-class f4_CellLink
+ * Class representing a connection between neuron cells.
+ */
+class f4_CellConn
 {
 public:
         /**
          * Constructor for f4_CellLink class. Parameter nfrom represents input
-         * neuron cell or NULL if connection has defined sensor type inside, like "[G:5]".
-         * The name of sensor class defined inside neuron connection is stored in the nt
-         * parameter.
-         * @param nfrom pointer to input neuron cell or NULL
-         * @param nw weight of connection
-         * @param nt name of neuron class or empty string
-         */
-        f4_CellLink(f4_Cell *nfrom, double nw, string nt);
-
-        f4_Cell *from; ///<pointer to input neuron cell
-        string t;      ///<empty if 'from' cell is given, NeuroClass name otherwise
-        double w;      ///<weight of connection
+         * neuron cell.
+         * @param nfrom pointer to input neuron cell
+         * @param nweight weight of connection
+         */
+        f4_CellConn(f4_Cell *nfrom, double nweight);
+
+        f4_Cell *from;  ///<pointer to input neuron cell
+        double weight;  ///<weight of connection
 };
 
@@ -264 +256 @@
          * @param nrepair 0 if nothing to repair
          */
-        f4_Cells(f4_node *genome, int nrepair);
+        f4_Cells(f4_Node *genome, int nrepair);
 
         /**
@@ -291 +283 @@
 
         /**
-         * Performs a single step of organism development. It runs each active cell
-         * in the organism.
-         * @return 0 if all cells are developed, or 1 otherwise
-         */
-        int  onestep();
+         * Performs a single step of organism development. It runs each active cell in the organism.
+         * @return false if all cells are developed or there is an error, true otherwise
+         */
+        bool oneStep();
 
         /**
@@ -302 +293 @@
          * @return 0 if organism developed successfully, error code if something went wrong
          */
-        int  simulate();
+        int simulate();
+
+        /**
+         * Prints the current state of the organism (for debugging purposes).
+         * @param description printout header
+         */
+        void print_cells(const char* description);
 
         /**
@@ -308 +305 @@
          * @return error code
          */
-        int  geterror() { return error; };
+        int getErrorCode() { return errorcode; };
 
         /**
@@ -314 +311 @@
          * @return position of an error
          */
-        int  geterrorpos() { return errorpos; };
+        int getErrorPos() { return errorpos; };
 
         /**
@@ -325 +322 @@
          * Sets the element of genotype to be repaired by removal.
          * @param nerrpos position of an error in genotype
-         * @param rem the f4_node to be removed from the  genotype tree in order to repair
-         */
-        void setRepairRemove(int nerrpos, f4_node *rem);
+         * @param rem the f4_Node to be removed from the  genotype tree in order to repair
+         */
+        void setRepairRemove(int nerrpos, f4_Node *rem);
 
         /**
@@ -336 +333 @@
          * @return 0 if repair can be performed, or -1 otherwise because the repair flag wasn't set in the constructor
          */
-        int  setRepairInsert(int nerrpos, f4_node *parent, f4_node *insert);
+        int setRepairInsert(int nerrpos, f4_Node *parent, f4_Node *insert);
 
         /**
@@ -343 +340 @@
          * @param whichchild 1 if first child, 2 otherwise
          */
-        void repairGeno(f4_node *geno, int whichchild);
+        void repairGeno(f4_Node *geno, int whichchild);
 
         // the cells
-        f4_Cell *C[MAX4CELLS];  ///<Array of all cells of an organism
-        int       nc;           ///<Number of cells in an organism
+        f4_Cell *C[F4_MAX_CELLS];  ///<Array of all cells of an organism
+        int     cell_count;        ///<Number of cells in an organism
 
 private:
         // for error reporting / genotype fixing
         int repair;
-        int error;
+        int errorcode;
         int errorpos;
-        f4_node *repair_remove;
-        f4_node *repair_parent;
-        f4_node *repair_insert;
+        f4_Node *repair_remove;
+        f4_Node *repair_parent;
+        f4_Node *repair_insert;
         void toF1GenoRec(int curc, SString &out);
         f4_Cell *tmpcel;                // needed by toF1Geno
-        f4_node *f4rootnode;          // used by constructor
+        f4_Node *f4rootnode;          // used by constructor
 };
 
@@ -366 +363 @@
  * A class to organize a f4 genotype in a tree structure.
  */
-class f4_node
+class f4_Node
 {
 public:
-        string name; ///<one-letter 'name', multiple characters for classes
-        f4_node *parent; ///<parent link or NULL
-        f4_node *child; ///<child or NULL
-        f4_node *child2; ///<second child or NULL
+        string name; ///<one-letter gene code or multiple characters for neuron classes (then neuclass != NULL)
+        f4_Node *parent; ///<parent link or NULL
+        f4_Node *child; ///<child or NULL
+        f4_Node *child2; ///<second child or NULL
         int pos; ///<original position in the string
-        int i1; ///<internal int parameter1
-        int l1; ///<internal long parameter1 (now also int, since long is not well specified and it is in our scenarios equivalent to int)
-        double f1; ///<internal double parameter1
-        string s1; ///<internal string parameter1
-
-        f4_node();
+
+        int reps; ///<repetition counter for the '#' gene
+        char prop_symbol; ///<old-style properties (force,intertia,sigmoid) of the N neuron: !=/
+        bool prop_increase; ///<false=decrease neuron property (force,intertia,sigmoid), true=increase it
+        int conn_from; ///<relative number of the neuron this neuron get an input from
+        double conn_weight; ///<neuron connection weight
+        NeuroClass *neuclass; ///< NULL or not if "name" is a neuroclass name with a proper genotype context ("N:neuroclassname"). New in 2023-04 - to fix fatal flaw with fundamental assumptions: it was impossible to distinguish between single-character neuron names such as S, D, G and single-character modifiers. They were all stored in the "name" field. Before 2018 this was never a problem because the only supported neuroclasses had distinctive symbols such as @|*GTS, and the set of supported modifiers was small and different from neuroclass letters (no G,D,S clash).
+
+        f4_Node();
 
         /**
@@ -387 +387 @@
          * @param npos position of node substring in the genotype string
          */
-        f4_node(string nname, f4_node *nparent, int npos);
+        f4_Node(string nname, f4_Node *nparent, int npos);
 
         /**
@@ -395 +395 @@
          * @param npos position of node character in the genotype string
          */
-        f4_node(char nname, f4_node *nparent, int npos);
-
-        ~f4_node();
+        f4_Node(char nname, f4_Node *nparent, int npos);
+
+        ~f4_Node();
+
+        /**
+         * Recursively print subtree (for debugging).
+         * @param root starting node
+         * @param indent initial indentation
+         */
+        static void print_tree(const f4_Node *root, int indent);
 
         /**
@@ -404 +411 @@
          * @return 0 if the child could be added, -1 otherwise
          */
-        int addChild(f4_node *nchi);
+        int addChild(f4_Node *nchi);
 
         /**
@@ -411 +418 @@
          * @return 0 if child could be removed, -1 otherwise
          */
-        int removeChild(f4_node *nchi);
+        int removeChild(f4_Node *nchi);
 
         /**
@@ -423 +430 @@
          * @return the number of nodes from this node
          */
-        int count();
+        int count() const;
 
         /**
@@ -430 +437 @@
          * @return pointer to the nth subnode or NULL if not found
          */
-        f4_node* ordNode(int n);
+        f4_Node* ordNode(int n);
 
         /**
@@ -436 +443 @@
          * @return random subnode
          */
-        f4_node* randomNode();
+        f4_Node* randomNode();
 
         /**
@@ -444 +451 @@
          * @return a random subnode with a given size or NULL
          */
-        f4_node* randomNodeWithSize(int min, int max);
+        f4_Node* randomNodeWithSize(int min, int max);
 
         /**
@@ -456 +463 @@
          * @return pointer to a tree copy
          */
-        f4_node* duplicate();
+        f4_Node* duplicate();
 
         /**
@@ -468 +475 @@
 /**
  * The main function for converting a string of f4 encoding to a tree structure. Prepares
- * f4_node root of tree and runs f4_processrec function for it.
+ * f4_Node root of tree and runs f4_processrec function for it.
  * @param geno the string representing an f4 genotype
- * @return a pointer to the f4_node object representing the f4 tree root
- */
-f4_node* f4_processtree(const char *geno);
+ * @return a pointer to the f4_Node object representing the f4 tree root
+ */
+f4_Node* f4_processtree(const char *geno);
 
 /**
  * Scans a genotype string starting from a given position. This recursive method creates
- * a tree of f4_node objects. This method extracts each potentially functional element
- * of a genotype string to a separate f4_nodes. When the branching character '<' occurs,
- * f4_processrec is deployed for the latest f4_node element. This method does not
+ * a tree of f4_Node objects. This method extracts each potentially functional element
+ * of a genotype string to a separate f4_Nodes. When the branching character '<' occurs,
+ * f4_processrec is deployed for the latest f4_Node element. This method does not
  * analyse the genotype semantically, it only checks if the syntax is proper. The only
  * semantic aspect is neuron class name extraction, where the GenoOperators
@@ -487 +494 @@
  * @return 0 if processing was successful, otherwise returns the position of an error in the genotype
  */
-int f4_processrec(const char *genot, unsigned pos0, f4_node *parent);
+int f4_processrec(const char *genot, unsigned pos0, f4_Node *parent);
 
 /**
@@ -502 +509 @@
 const char *parseConnection(const char *fragm, int &relfrom, double &weight);
 
-/**
- * Parses the notation of the neuron connection with neuron definition - takes
- * the beginning of the connection definition, extracts the name of neuron class
- * that will be the input for the current neuron and the weight of the connection.
- * After successful parsing, returns a pointer to the first character after the connection
- * definition, or NULL if the connection definition was not valid due to the lack of [, :, ]
- * characters, an invalid value of the weight or an invalid neuron class name.
- * @param fragm the beginning of the connection definition, should be the '[' character
- * @param neutype the reference to a string representing the input neuron class name. The name of the class is validated with GenoOperators::parseNeuroClass()
- * @param weight the reference to a double variable in which the weight of the connection will be stored
- * @return the pointer to the first character in string after the connection definition
- */
-const char *parseConnectionWithNeuron(const char *fragm, string &neutype, double &weight);
-
 #endif