source: cpp/frams/genetics/f4/f4_general.h @ 1230

Last change on this file since 1230 was 1230, checked in by Maciej Komosinski, 11 months ago

Got rid of the (buggy) look-ahead function, made parsing stricter and simpler

  • Property svn:eol-style set to native
File size: 17.5 KB
Line 
1// This file is a part of Framsticks SDK.  http://www.framsticks.com/
2// Copyright (C) 1999-2023  Maciej Komosinski and Szymon Ulatowski.
3// See LICENSE.txt for details.
4
5// Copyright (C) 1999,2000  Adam Rotaru-Varga (adam_rotaru@yahoo.com), GNU LGPL
6
7#ifndef _F4_GENERAL_H_
8#define _F4_GENERAL_H_
9
10#include <frams/util/3d.h>
11#include <frams/util/sstring.h>
12#include <frams/util/multirange.h>
13#include <frams/genetics/geneprops.h>
14
15#ifdef DMALLOC
16#include <dmalloc.h>
17#endif
18
19/**
20 * Performs single rotation angle decrementation on a given value.
21 * @param v pointer to the decremented value
22 */
23void rolling_dec(double *v);
24
25/**
26 * Performs single rotation angle incrementation on a given value.
27 * @param v pointer to the incremented value
28 */
29void rolling_inc(double *v);
30
31class f4_Node;   // later
32class f4_Cell;   // later
33class f4_Cells;  // later
34
35
36/** @name Types of f4_Cell's */
37//@{
38#define CELL_UNDIFF 40 ///<undifferentiated cell
39#define CELL_STICK  41 ///<differentiated to stick, cannot divide
40#define CELL_NEURON 42 ///<differentiated to neuron, can divide
41//@}
42
43
44class f4_CellConn;
45
46/** @name Constraints of f4 genotype structures */
47//@{
48#define F4_MAX_CELL_INPUTS  10 ///<maximum number of neuron inputs in a developing organism
49#define F4_MAX_CELLS 100 ///<maximum number of f4 organism cells
50//@}
51
52/**
53 * Abstract cell type - the representation of a single component in the developmental
54 * encoding. In the beginning, each f4_Cell is undifferentiated. During the process
55 * of development it can divide or differentiate into a stick or a neuron. If it
56 * differentiates to a neuron, then it preserves the ability to divide, but divided
57 * cells will be the same type as the parent cell. If it is a stick, then it cannot
58 * be divided anymore.
59 *
60 * From f4_Cell array the final Model of a creature is created.
61 */
62class f4_Cell
63{
64public:
65        /**
66         * Represents the repetition marker. It holds information about the pointer
67         * to the repetition node and the count of repetitions.
68         */
69        class repeat_ptr
70        {
71        public:
72                repeat_ptr() : node(NULL), count(-1) { };
73
74                /**
75                 * A constructor that takes the pointer to the repetition node and the count of repetitions.
76                 * @param a pointer to f4_Node for repetition character
77                 * @param b the number of repetitions
78                 */
79                repeat_ptr(f4_Node *a, int b) : node(a), count(b) { };
80
81                inline void makeNull() { node = NULL; count = -1; };
82
83                inline bool isNull() const { return ((node == NULL) || (count <= 0)); };
84
85                inline void dec() { count--; };
86                f4_Node    *node; ///<pointer to the repetition code
87                int       count; ///<repetition counter
88        };
89
90        /**
91         * Represents the stack of repeat_ptr objects. The objects are
92         * pushed to the stack when '#' repetition symbol appears, and are popped when
93         * the end of the current cell definition, i.e. the '>' character, appears. After the
94         * '>' character, the cell is duplicated as many times as it is defined after the
95         * repetition marker.
96         */
97        class repeat_stack
98        {
99        public:
100                repeat_stack() { top = 0; }
101
102                inline void clear() { top = 0; }
103
104                /**
105                 * Pushes repeat_ptr object onto the stack. If the stack size is exceeded, then no
106                 * information is provided.
107                 * @param rn repetition node info
108                 */
109                inline void push(repeat_ptr rn) { if (top >= stackSize) return; ptr[top] = rn; top++; }
110
111                inline void pop() { if (top > 0) top--; }
112
113                /**
114                 * Gets the current top element.
115                 * @return pointer to the element on top of the repeat_stack object
116                 */
117                inline repeat_ptr* first() { return &(ptr[top - (top > 0)]); };
118                static const int stackSize = 4;  ///<max 4 nested levels
119                repeat_ptr ptr[stackSize]; ///<array holding pointers to repeat_ptr
120                int top;  ///<index of the top of the stack
121        };
122
123        /**
124         * Creates a new f4_Cell object.
125         * @param nnr number of the cell
126         * @param ndad pointer to the parent of the created cell
127         * @param nangle the amount of commas affecting branch angles
128         * @param newP genotype properties of a given cell
129         */
130        f4_Cell(int nnr, f4_Cell *ndad, int nangle, GeneProps newP);
131        /**
132         * Creates a new f4_Cell object.
133         * @param nO pointer to an organism containing the cell
134         * @param nnr number of the cell
135         * @param ngeno pointer to the root of the genotype tree
136         * @param ngcur pointer to the f4_Node representing the current cell in the genotype tree
137         * @param ndad pointer to the parent of the created cell
138         * @param nangle the number of commas affecting branch angles
139         * @param newP genotype properties of a given cell
140         */
141        f4_Cell(f4_Cells *nO, int nnr, f4_Node *ngeno, f4_Node *ngcur, f4_Cell *ndad, int nangle, GeneProps newP);
142
143        ~f4_Cell();
144
145        /**
146         * Performs a single step of cell development. This method requires a pointer to
147         * the f4_Cells object in org attribute. If the current node in genotype tree
148         * is the branching character '<', the cell divides into two cells, unless the
149         * cell was already differentiated into the stick cell. Otherwise, the current
150         * differentiation or modification is performed on the cell. If current node is
151         * creating a connection between two neuron nodes and the input node is not
152         * yet developed, the simulation of the development of the current cell waits until
153         * the input node is created. The onestep method is deployed for every cell
154         * at least once. If one cell requires another one to develop, onestep
155         * should be deployed again on this cell. This method, unlike genotype tree
156         * creation, checks semantics. This means that this function will fail if:
157         *  - the cell differentiated as a stick will have branching node '<',
158         *  - the undifferentiated cell will have termination node '>' (end of cell development without differentiation),
159         *  - the stack of repetition marker '#' will exceed maximum allowed value of repetition,
160         *  - the stick modifiers, like rotation, will be applied on neuron cell,
161         *  - the differentiated cell will be differentiated again,
162         *  - the connection between neurons cannot be established,
163         *  - the neuron class is not valid.
164         *
165         * @return 0 if development was successful, 1 if there was an error in genotype tree
166         */
167        int oneStep();
168
169        /**
170         * Adds a connection between this neuron cell and a given neuron cell in nfrom.
171         * @param nfrom input neuron cell
172         * @param nweight weight of connection
173         * @return 0 if connection is established, -1 otherwise
174         */
175        int   addConnection(f4_Cell *nfrom, double nweight);
176
177        /**
178         * Adjusts properties of stick objects.
179         */
180        void  adjustRec();
181
182        int        nr;                 ///<number of cell (seems to be used only in old f1 converter for neuron connections)
183        int        type;               ///<type
184        f4_Cell *dadlink;              ///<pointer to cell parent
185        f4_Cells  *org;                ///<uplink to organism
186
187        f4_Node *genot;                    ///<genotype tree
188        f4_Node *gcur;                 ///<current genotype execution pointer
189        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
190        repeat_stack repeat;           ///<stack holding repetition nodes and counters
191        int recProcessedFlag;          ///<used during recursive traverse
192        MultiRange genoRange;          ///<remember the genotype codes affecting this cell so far
193
194        GeneProps    P;                ///<properties
195        int          anglepos;         ///<number of position within dad's children (,)
196        int          childcount;       ///<number of children
197        int          commacount;       ///<number of postitions at lastend (>=childcount)
198        double       rolling;          ///<rolling angle ('R') (around x)
199        double       xrot;                         ///<rotation angle around x
200        double       zrot;             ///<horizontal rotation angle due to branching (around z)
201
202        double       mz;               ///<freedom in z
203        int          p2_refno;         ///<the number of the last end part object, used in f0
204        int          joint_refno;      ///<the number of the joint object, used in f0
205        int          neuro_refno;      ///<the number of the neuro object, used in f0
206
207        double       inertia;          ///<inertia of neuron
208        double       force;            ///<force of neuron
209        double       sigmo;            ///<sigmoid of neuron
210        f4_CellConn *conns[F4_MAX_CELL_INPUTS]; ///<array of neuron connections
211        int          conns_count;      ///<number of connections
212        NeuroClass *neuclass;          ///<pointer to neuron class
213};
214
215/**
216 * Class representing a connection between neuron cells.
217 */
218class f4_CellConn
219{
220public:
221        /**
222         * Constructor for f4_CellLink class. Parameter nfrom represents input
223         * neuron cell.
224         * @param nfrom pointer to input neuron cell
225         * @param nweight weight of connection
226         */
227        f4_CellConn(f4_Cell *nfrom, double nweight);
228
229        f4_Cell *from;  ///<pointer to input neuron cell
230        double weight;  ///<weight of connection
231};
232
233
234/**
235 * A class representing a collection of cells. It is equivalent to an organism.
236 */
237class f4_Cells
238{
239public:
240
241        /**
242         * Constructor taking genotype in a form of a tree.
243         * @param genome genotype tree
244         * @param nrepair 0 if nothing to repair
245         */
246        f4_Cells(f4_Node *genome, int nrepair);
247
248        /**
249         * Constructor taking genotype in a form of a string.
250         * @param genome genotype string
251         * @param nrepair 0 if nothing to repair
252         */
253        f4_Cells(SString &genome, int nrepair);
254
255        /**
256         * Destructor removing cells from memory.
257         */
258        ~f4_Cells();
259
260        /**
261         * Adds a new cell to organism.
262         * @param newcell cell to be added
263         */
264        void addCell(f4_Cell *newcell);
265
266        /**
267         * Creates an approximate genotype in the f1 encoding and stores it in a given parameter.
268         * @param out the string in which the approximate f1 genotype will be stored
269         */
270        void toF1Geno(SString &out);
271
272        /**
273         * Performs a single step of organism development. It runs each active cell in the organism.
274         * @return false if all cells are developed or there is an error, true otherwise
275         */
276        bool oneStep();
277
278        /**
279         * Performs the full development of organism and returns error code if something
280         * went wrong.
281         * @return 0 if organism developed successfully, error code if something went wrong
282         */
283        int simulate();
284
285        /**
286         * Prints the current state of the organism (for debugging purposes).
287         * @param description printout header
288         */
289        void print_cells(const char* description);
290
291        /**
292         * Returns error code of the last simulation.
293         * @return error code
294         */
295        int getErrorCode() { return errorcode; };
296
297        /**
298         * Returns position of an error in genotype.
299         * @return position of an error
300         */
301        int getErrorPos() { return errorpos; };
302
303        /**
304         * Sets error code GENOPER_OPFAIL for a simulation on a given position.
305         * @param nerrpos position of an error
306         */
307        void setError(int nerrpos);
308
309        /**
310         * Sets the element of genotype to be repaired by removal.
311         * @param nerrpos position of an error in genotype
312         * @param rem the f4_Node to be removed from the  genotype tree in order to repair
313         */
314        void setRepairRemove(int nerrpos, f4_Node *rem);
315
316        /**
317         * Sets repairing of a genotype by inserting a new node to the current genotype.
318         * @param nerrpos position of an error in genotype
319         * @param parent the parent of a new element
320         * @param insert the element to be inserted
321         * @return 0 if repair can be performed, or -1 otherwise because the repair flag wasn't set in the constructor
322         */
323        int setRepairInsert(int nerrpos, f4_Node *parent, f4_Node *insert);
324
325        /**
326         * Repairs the genotype according to setRepairRemove or setRepairInsert methods.
327         * @param geno pointer to the genotype tree
328         * @param whichchild 1 if first child, 2 otherwise
329         */
330        void repairGeno(f4_Node *geno, int whichchild);
331
332        // the cells
333        f4_Cell *C[F4_MAX_CELLS];  ///<Array of all cells of an organism
334        int     cell_count;        ///<Number of cells in an organism
335
336private:
337        // for error reporting / genotype fixing
338        int repair;
339        int errorcode;
340        int errorpos;
341        f4_Node *repair_remove;
342        f4_Node *repair_parent;
343        f4_Node *repair_insert;
344        void toF1GenoRec(int curc, SString &out);
345        f4_Cell *tmpcel;                // needed by toF1Geno
346        f4_Node *f4rootnode;          // used by constructor
347};
348
349
350/**
351 * A class to organize a f4 genotype in a tree structure.
352 */
353class f4_Node
354{
355public:
356        string name; ///<one-letter gene code or multiple characters for neuron classes (then neuclass != NULL)
357        f4_Node *parent; ///<parent link or NULL
358        f4_Node *child; ///<child or NULL
359        f4_Node *child2; ///<second child or NULL
360        int pos; ///<original position in the string
361
362        int reps; ///<repetition counter for the '#' gene
363        char prop_symbol; ///<old-style properties (force,intertia,sigmoid) of the N neuron: !=/
364        bool prop_increase; ///<false=decrease neuron property (force,intertia,sigmoid), true=increase it
365        int conn_from; ///<relative number of the neuron this neuron get an input from
366        double conn_weight; ///<neuron connection weight
367        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).
368
369        f4_Node();
370
371        /**
372         * Multiple-character name constructor.
373         * @param nname string from genotype representing node
374         * @param nparent pointer to parent of the node
375         * @param npos position of node substring in the genotype string
376         */
377        f4_Node(string nname, f4_Node *nparent, int npos);
378
379        /**
380         * Single-character name constructor.
381         * @param nname character from genotype representing node
382         * @param nparent pointer to parent of the node
383         * @param npos position of node character in the genotype string
384         */
385        f4_Node(char nname, f4_Node *nparent, int npos);
386
387        ~f4_Node();
388
389        /**
390         * Recursively print subtree (for debugging).
391         * @param root starting node
392         * @param indent initial indentation
393         */
394        static void print_tree(const f4_Node *root, int indent);
395
396        /**
397         * Adds the child to the node.
398         * @param nchi the child to be added to the node
399         * @return 0 if the child could be added, -1 otherwise
400         */
401        int addChild(f4_Node *nchi);
402
403        /**
404         * Removes the child from the node.
405         * @param nchi the child to be removed from the node
406         * @return 0 if child could be removed, -1 otherwise
407         */
408        int removeChild(f4_Node *nchi);
409
410        /**
411         * Returns the number of children.
412         * @return 0, 1 or 2
413         */
414        int childCount();
415
416        /**
417         * Returns the number of nodes coming from this node in a recursive way.
418         * @return the number of nodes from this node
419         */
420        int count() const;
421
422        /**
423         * Returns the nth subnode (0-)
424         * @param n index of the child to be found
425         * @return pointer to the nth subnode or NULL if not found
426         */
427        f4_Node* ordNode(int n);
428
429        /**
430         * Returns a random subnode.
431         * @return random subnode
432         */
433        f4_Node* randomNode();
434
435        /**
436         * Returns a random subnode with a given size.
437         * @param min minimum size
438         * @param max maximum size
439         * @return a random subnode with a given size or NULL
440         */
441        f4_Node* randomNodeWithSize(int min, int max);
442
443        /**
444         * Prints recursively the tree from a given node.
445         * @param buf variable to store printing result
446         */
447        void      sprintAdj(char *&buf);
448
449        /**
450         * Recursively copies the genotype tree from this node.
451         * @return pointer to a tree copy
452         */
453        f4_Node* duplicate();
454
455        /**
456         * Recursively releases memory from all node children.
457         */
458        void      destroy();
459private:
460        void     sprint(SString &out);  // print recursively
461};
462
463/**
464 * The main function for converting a string of f4 encoding to a tree structure. Prepares
465 * f4_Node root of tree and runs f4_processRecur function for it.
466 * @param geno the string representing an f4 genotype
467 * @return a pointer to the f4_Node object representing the f4 tree root
468 */
469//f4_Node* f4_processTree(const char *geno);
470
471/**
472 * Scans a genotype string starting from a given position. This recursive method creates
473 * a tree of f4_Node objects. This method extracts each potentially functional element
474 * of a genotype string to a separate f4_Nodes. When the branching character '<' occurs,
475 * f4_processRecur is deployed for the latest f4_Node element. This method does not
476 * analyse the genotype semantically, it only checks if the syntax is proper. The only
477 * semantic aspect is neuron class name extraction, where the GenoOperators
478 * class is used to parse the potential neuron class name.
479 * @param genot the string holding all the genotype
480 * @param pos0 the current position of processing in string
481 * @param parent current parent of the analysed branch of the genotype
482 * @return 0 if processing was successful, otherwise returns the position of an error in the genotype
483 */
484int f4_processRecur(const char *genot, int &pos_inout, f4_Node *parent);
485
486/**
487 * Parses notation of the neuron connection - takes the beginning of the connection
488 * definition, extracts the relative position of input neurons and the weight of the connection.
489 * After successful parsing, returns the pointer to the first character after the connection
490 * definition, or NULL if the connection definition was not valid due to the lack of [, :, ]
491 * characters or an invalid value of relfrom or weight.
492 * @param fragm the beginning of connection definition, should be the '[' character
493 * @param relfrom the reference to an int variable in which the relative position of the input neuron will be stored
494 * @param weight the reference to a double variable in which the weight of the connection will be stored
495 * @return the pointer to the first character in string after connection definition
496 */
497const char *parseConnection(const char *fragm, int &relfrom, double &weight);
498
499#endif
Note: See TracBrowser for help on using the repository browser.