[ILNumerics Core Module]
Namespace: ILNumerics
Assembly: ILNumerics.Core (in ILNumerics.Core.dll) Version: 5.5.0.0 (5.5.7503.3146)
Parameters
- d0
- Type: SystemInt64
Index into the first dimension. - d1
- Type: SystemInt64
Index into the second dimension. - d2
- Type: SystemInt64
Index into the third dimension.
Return Value
Type: Int64Sequential index into the array when stored as 1D array of arbitrary storage order.
Exception | Condition |
---|---|
IndexOutOfRangeException | if the leading dimension specifier d0 is equal or greater than the length of the first dimension. |
IndexOutOfRangeException | if the leading dimension specifier d1 is equal or greater than the length of the second dimension. |
IndexOutOfRangeException | if d2 is equal or greater than the length of the third dimension and the resulting index points to an non-existing element. |
For arrays of matrix shape d0 references the index of the row and d1 references the index of the column of the sequential index to return.
More general, for n-dimensional arrays the first index corresponds to the position in the first (index: 0) dimension, the second index to the position of the element in the second dimension (index: 1), a.s.o.
If the array addressed by this size has less dimensions than addressed by this function, trailing indices (i.e.: such indices which deal with a dimension whose index is higher than existing in the array) correspond to 'virtual' or singleton dimensions. Those indices must address the 0-th element of the virtual dimension. Hence its value must be 0.
Conversely, if the array addressed by this size stores more dimensions than addressed by this function the last index parameter d2 may exceeds the limit of its corresponding dimension. In this case the sequential index to be returned is computed by subsequently merging trailing dimensions and applying the superflous modulus of the value of d2 to the next dimension, correspondingly. This process is repeated until either the number of dimensions is reached or the modified new value of d2 reaches 0.
Any index parameter may be negative, in which case the resulting index is computed by adding the corresponding dimension length to the parameters value. Hence, a value of -1 addresses the last element in the dimension. Note, that for the last specified dimension a value of -1 references the last element in the _virtual_ dimension, merged with any non specified trailing dimensions.
This function recognizes arbitrarily strided size objects. It expects all sizes (dimension lengths, strides, the base offset and the number of elements) to fit into MaxValue. Consider using one of the overloads with UInt32 parameters if you are not dealing with very big arrays and don't require negative indexing. This may bring better performance in tight loops.
Performance hint: this function (and all corresponding overloads) are optimized for the case where the number of index parameters provided corresponds to the number of dimensions in the array and all provided index parameters are within the range of their corresponding dimension. The functions, however, handle arbitrary cases, including addressing, merging and ommitting trailing dimensions. Any of those extended features may introduce a performance penalty, though.
[ILNumerics Core Module]