void agent_init(const char* task_spec)
{
/*Struct to hold the parsed task spec*/
taskspec_t *ts=(taskspec_t*)malloc(sizeof(taskspec_t));
int decode_result = decode_taskspec( ts, task_spec );
if(decode_result!=0){
printf("Could not decode task spec, code: %d for task spec: %s\n",decode_result,task_spec);
exit(1);
}
/* Lots of assertions to make sure that we can handle this problem. */
assert(getNumIntObs(ts)==1);
assert(getNumDoubleObs(ts)==0);
assert(isIntObsMax_special(ts,0)==0);
assert(isIntObsMin_special(ts,0)==0);
numStates=getIntObsMax(ts,0)+1;
assert(getNumIntAct(ts)==1);
assert(getNumDoubleAct(ts)==0);
assert(isIntActMax_special(ts,0)==0);
assert(isIntActMin_special(ts,0)==0);
numActions=getIntActMax(ts,0)+1;
free_taskspec_struct(ts); /* Make the taskspec struct a "blank slate" */
free(ts); /* Free the structure itself */
/*Here is where you might allocate storage for parameters (value function or policy, last action, last observation, etc)*/
/*Here you would parse the task spec if you felt like it*/
/*Allocate memory for a one-dimensional integer action using utility functions from RLStruct_util*/
allocateRLStruct(&this_action,1,0,0);
allocateRLStruct(&last_action,1,0,0);
/* That is equivalent to:
this_action.numInts = 1;
this_action.intArray = (int*)calloc(1,sizeof(int));
this_action.numDoubles = 0;
this_action.doubleArray = 0;
this_action.numChars = 0;
this_action.charArray = 0;
*/
/*Allocate memory for a one-dimensional integer observation using utility functions from RLStruct_util*/
last_observation=allocateRLStructPointer(1,0,0);
/*Later we will parse this from the task spec, but for now*/
value_function=(double *)calloc(numActions*numStates,sizeof(double));
}
void agent_init(const char* task_spec)
{
srand(time(0));
/*Struct to hold the parsed task spec*/
taskspec_t *ts = (taskspec_t*)malloc(sizeof(taskspec_t));
int decode_result = decode_taskspec(ts, task_spec);
if (decode_result != 0){
printf("Could not decode task spec, code: %d for task spec: %s\n", decode_result, task_spec);
exit(1);
}
possibleActions = getIntActMax(ts,0)-1;
allocateRLStruct(¤tAction, 1, 0, 0);
//load();
}
void test_accessors( taskspec_t *ts )
{
int_range_t ir;
double_range_t dr;
int i, array_len;
if (ts == NULL)
return;
/* determine problem type */
printf( "Problem type: " );
if (isEpisodic( ts ) == 1) {
printf( "episodic\n" );
} else if (isContinuing( ts ) == 1) {
printf( "continuing\n" );
} else if (isOtherType( ts ) == 1) {
printf( "other\n" );
} else {
printf( "ERROR\n" );
}
/* observation types and ranges */
printf( "OBSERVATIONS:\n" );
array_len = getNumIntObs( ts );
printf( " INTS (%d total): ", array_len );
for (i = 0; i < array_len; i++) {
ir = getIntObs( ts, i );
printf( "(%d ", ir.repeat_count );
if (isIntObsMin_special( ts, i ) == 1) {
if (isIntObsMin_negInf( ts, i ) == 1) {
printf( "NEGINF " );
} else if (isIntObsMin_unspec( ts, i ) == 1) {
printf( "UNSPEC " );
} else {
printf( "ERROR " );
}
} else {
/* printf( "%d ", ir.min ); */
printf( "%d ", getIntObsMin( ts, i ) );
}
if (isIntObsMax_special( ts, i ) == 1) {
if (isIntObsMax_posInf( ts, i ) == 1) {
printf( "POSINF) " );
} else if (isIntObsMax_unspec( ts, i ) == 1) {
printf( "UNSPEC) " );
} else {
printf( "ERROR) " );
}
} else {
/* printf( "%d) ", ir.max ); */
printf( "%d) ", getIntObsMax( ts, i ) );
}
}
array_len = getNumDoubleObs( ts );
printf( "\n DOUBLES (%d total): ", array_len );
for (i = 0; i < array_len; i++) {
dr = getDoubleObs( ts, i );
printf( "(%d ", dr.repeat_count );
if (isDoubleObsMin_special( ts, i ) == 1) {
if (isDoubleObsMin_negInf( ts, i ) == 1) {
printf( "NEGINF " );
} else if (isDoubleObsMin_unspec( ts, i ) == 1) {
printf( "UNSPEC " );
} else {
printf( "ERROR " );
}
} else {
/* printf( "%g ", dr.min ); */
printf( "%g ", getDoubleObsMin( ts, i ) );
}
if (isDoubleObsMax_special( ts, i ) == 1) {
if (isDoubleObsMax_posInf( ts, i ) == 1) {
printf( "POSINF) " );
} else if (isDoubleObsMax_unspec( ts, i ) == 1) {
printf( "UNSPEC) " );
} else {
printf( "ERROR) " );
}
} else {
/* printf( "%g) ", dr.max ); */
printf( "%g) ", getDoubleObsMax( ts, i ) );
}
}
printf( "\n CHARCOUNT: %d", getCharcountObs( ts ) );
/* action types and ranges */
printf( "\nACTIONS:\n" );
array_len = getNumIntAct( ts );
printf( " INTS (%d total): ", array_len );
for (i = 0; i < array_len; i++) {
ir = getIntAct( ts, i );
printf( "(%d ", ir.repeat_count );
if (isIntActMin_special( ts, i ) == 1) {
if (isIntActMin_negInf( ts, i ) == 1) {
printf( "NEGINF " );
} else if (isIntActMin_unspec( ts, i ) == 1) {
printf( "UNSPEC " );
} else {
printf( "ERROR " );
}
} else {
/* printf( "%d ", ir.min ); */
printf( "%d ", getIntActMin( ts, i ) );
}
if (isIntActMax_special( ts, i ) == 1) {
if (isIntActMax_posInf( ts, i ) == 1) {
printf( "POSINF) " );
} else if (isIntActMax_unspec( ts, i ) == 1) {
printf( "UNSPEC) " );
} else {
printf( "ERROR) " );
}
} else {
/* printf( "%d) ", ir.max ); */
printf( "%d) ", getIntActMax( ts, i ) );
}
}
array_len = getNumDoubleAct( ts );
printf( "\n DOUBLES (%d total): ", array_len );
for (i = 0; i < array_len; i++) {
dr = getDoubleAct( ts, i );
printf( "(%d ", dr.repeat_count );
if (isDoubleActMin_special( ts, i ) == 1) {
if (isDoubleActMin_negInf( ts, i ) == 1) {
printf( "NEGINF " );
} else if (isDoubleActMin_unspec( ts, i ) == 1) {
printf( "UNSPEC " );
} else {
printf( "ERROR " );
}
} else {
/* printf( "%g ", dr.min ); */
printf( "%g ", getDoubleActMin( ts, i ) );
}
if (isDoubleActMax_special( ts, i ) == 1) {
if (isDoubleActMax_posInf( ts, i ) == 1) {
printf( "POSINF) " );
} else if (isDoubleActMax_unspec( ts, i ) == 1) {
printf( "UNSPEC) " );
} else {
printf( "ERROR) " );
}
} else {
/* printf( "%g) ", dr.max ); */
printf( "%g) ", getDoubleActMax( ts, i ) );
}
}
printf( "\n CHARCOUNT: %d", getCharcountAct( ts ) );
/* reward range */
printf( "\nREWARD: (" );
if (isRewardMin_special( ts ) == 1) {
if (isRewardMin_negInf( ts ) == 1) {
printf( "NEGINF " );
} else if (isRewardMin_unspec( ts ) == 1) {
printf( "UNSPEC " );
} else {
printf( "ERROR " );
}
} else {
printf( "%g ", getRewardMin( ts ) );
}
if (isRewardMax_special( ts ) == 1) {
if (isRewardMax_posInf( ts ) == 1) {
printf( "POSINF) " );
} else if (isRewardMax_unspec( ts ) == 1) {
printf( "UNSPEC) " );
} else {
printf( "ERROR) " );
}
} else {
printf( "%g) ", getRewardMax( ts ) );
}
printf( "\n" );
}