maybe I just have don't get it, ---could I get some help?
RLLib::RLProblem<T>
has: Base::discreteActions && Base::continuousActions
almost all models uses Base::discreteActions,
there is Helicopter.h that uses Base::continuousActions but is incomplete.
I've been trying to model a problem with two continuous actions,
I've try this:
Base::continuousActions->push_back(0, 0.0); Base::continuousActions->push_back(1, 0.0);
this would create something like:[[0],[0]]
but this crashes the distribution in ASSERT((phi->dimension() == 1) && (actions->dimension() == 1));
based on Helicopter.h I've also try:
Base::continuousActions->push_back(0, 0.0); Base::continuousActions->push_back(0, 0.0);
this would create something like:[[0, 0]]
here the code runs, but the distribution ignores the second dimension and uses the first always: actions->getEntry(defaultAction);
before I attempt to modify the distributions, it would be a blessing if someone knows how may I approach to solve a continuous two dimensional action space RLLib::RLProblem<T>.
maybe I just have don't get it, ---could I get some help?
RLLib::RLProblem<T>has:
Base::discreteActions && Base::continuousActionsalmost all models uses
Base::discreteActions,there is
Helicopter.hthat usesBase::continuousActionsbut is incomplete.I've been trying to model a problem with two continuous actions,
I've try this:
Base::continuousActions->push_back(0, 0.0); Base::continuousActions->push_back(1, 0.0);this would create something like:
[[0],[0]]but this crashes the distribution in
ASSERT((phi->dimension() == 1) && (actions->dimension() == 1));based on
Helicopter.hI've also try:Base::continuousActions->push_back(0, 0.0); Base::continuousActions->push_back(0, 0.0);this would create something like:
[[0, 0]]here the code runs, but the distribution ignores the second dimension and uses the first always:
actions->getEntry(defaultAction);before I attempt to modify the distributions, it would be a blessing if someone knows how may I approach to solve a continuous two dimensional action space
RLLib::RLProblem<T>.