It performs a reduction operation on a datacube with respect to implicit dimensions.
- cube: name of the input datacube. The name must be in PID format.
- schedule: scheduling algorithm. The only possible value is 0, for a static linear block distribution of resources.
- group_size: size of the aggregation set. If set to “all”, the reduction will occur on all elements of each tuple.
- operation: reduction operation. Possible values are:
- “count”: to evaluate the actual values (not missing)
- “max”: to evaluate the maximum value
- “min”: to evaluate the minimum value
- “avg”: to evaluate the mean value
- “sum”: to evaluate the sum
- “std”: to evaluate the standard deviation
- “var”: to evaluate the variance
- “cmoment”: to evaluate the central moment
- “acmoment”: to evaluate the absolute central moment
- “rmoment” to evaluate the raw moment
- “armoment” to evaluate the absolute raw moment
- “quantile”: to evaluate the quantile
- “arg_max” to evaluate the index of the maximum value
- “arg_min” to evaluate the index of the minimum value
- order: order used in evaluation of the moments or value of the quantile in range [0, 1].
- missingvalue: value to be considered as missing value; by default it is NAN (for float and double).
- grid: optional argument used to identify the grid of dimensions to be used (if the grid already exists) or the one to be created (if the grid has a new name). If it isn’t specified, no grid will be used.
- container: name of the container to be used to store the output cube; by default, it is the input container.
- check_grid: optional flag to be enabled in case the values of grid have to be checked (valid only if the grid already exists).
- description: additional description to be associated with the output cube.
- exec_mode: operator execution mode. Possible values are async (default) for asynchronous mode, sync for synchronous mode with json-compliant output.
- ncores: number of parallel processes to be used (min. 1).
- nthreads: number of parallel threads per process to be used (min. 1).
- sessionid: session identifier used server-side to manage sessions and jobs. Usually, users don’t need to use/modify it, except when it is necessary to create a new session or switch to another one.
- objkey_filter: filter on the output of the operator written to file (default=all => no filter, none => no output, reduce => shows operator’s output PID as text).
Compute the maximum values of 10-element groups in the datacube identified by the PID “URL/1/1”:
[OPH_TERM] >> oph_reduce operation=max;group_size=10;cube=URL/1/1;grid=new_grid;