@gpestre , are you happy with the implementation? I expect this would be faster and we'd run less risk of memory leaks we our DataVector class had a setter for value() so we wouldn't have to go via a std::vector<double>. Or maybe it'd hardly make a difference?
The parallelisation seems to work.
@gpestre , are you happy with the implementation? I expect this would be faster and we'd run less risk of memory leaks we our
DataVector
class had a setter forvalue()
so we wouldn't have to go via astd::vector<double>
. Or maybe it'd hardly make a difference?