Entry point to compute kurtosis with the provided mean values.
Description and Assumptions
The oneapi::mkl::stats::kurtosis function is used to compute an array of kurtosis (kurtosis for each dataset's dimension) with the provided mean values.
:ref:`onemkl_stats_kurtosis_user_mean` supports the following precisions for data:
T float
double
Syntax
namespace oneapi::mkl::stats {
template<method Method = method::fast, typename Type,
layout ObservationsLayout>
void oneapi::mkl::stats::kurtosis(sycl::queue& queue,
sycl::buffer<Type, 1> mean,
const oneapi::mkl::stats::dataset<sycl::buffer<Type, 1>, ObservationsLayout>& data,
sycl::buffer<Type, 1> kurtosis);
}
Template Parameters
- Method
Method which is used for estimate computation. The specific values are as follows:
oneapi::mkl::stats::method::fast
- Type
- Data precision.
- ObservationsLayout
- Data layout. The specific values are described in :ref:`onemkl_stats_dataset`.
Input Parameters
- queue
- The queue where the routine should be executed.
- mean
- sycl::buffer to the array of provided mean values.
- data
- Dataset which is used for computation.
Output Parameters
- kurtosis
- sycl::buffer array of kurtosis values.
Throws
- oneapi::mkl::invalid_argument
- Exception is thrown when kurtosis.get_count() == 0, or mean.get_count() == 0, or dataset object is invalid
Syntax
namespace oneapi::mkl::stats {
template<method Method = fast, typename Type, layout ObservationsLayout>
sycl::event kurtosis(sycl::queue& queue,
Type* mean,
const dataset<ObservationsLayout, Type*>& data,
Type* kurtosis,
const std::vector<sycl::event> &dependencies = {});
}
Template Parameters
- Method
Method which is used for estimate computation. The specific values are as follows:
oneapi::mkl::stats::method::fast
- Type
- Data precision.
- ObservationsLayout
- Data layout. The specific values are described in :ref:`onemkl_stats_dataset`.
Input Parameters
- queue
- The queue where the routine should be executed.
- mean
- Pointer to the array of provided mean values.
- data
- Dataset which is used for computation.
- dependencies
- Optional parameter. List of events to wait for before starting computation, if any.
Output Parameters
- kurtosis
- Pointer to the array of the kurtosis values.
Throws
- oneapi::mkl::invalid_argument
- Exception is thrown when kurtosis == nullptr, or mean == nullptr, or dataset object is invalid
Return Value
Output event to wait on to ensure computation is complete.
Parent topic: :ref:`onemkl_stats_routines`