cumulate

Owned by Jason Kinchen
Jul 24, 2024

The cumulate operator produces a result array containing a running aggregate along a dimension of the input array.

Synopsis

cumulate(array, aggregate[,aggregate2,...] [, dimension]])

Summary

The cumulate operator calculates a running aggregate along a single dimension of the input array. The calculation happens along one particular dimension of the array. The aggregate call requires one of the SciDB or P4 aggregates, or a user-defined aggregate function. If you do not specify the dimension, cumulate uses the first declared dimension of the array.

Limitation

The cumulate operator cannot handle input arrays containing non-zero chunk overlaps.

Examples

Cumulative Sum Along the Second Dimension

To calculate the cumulative sum along the second dimension of a 2-dimensional, 1 -attribute array, do the following:

  1. Create a matrix and store values into it:

    AFL% store(build(<val:double>[row=0:2; col=0:3],row*4+col+1),A);


    The output is:

    [
[(1),(2),(3),(4)],
[(5),(6),(7),(8)],
[(9),(10),(11),(12)]
] 

  2. Calculate the cumulative sum along the first dimension:

    AFL% cumulate(A,sum(val),row);  


    The output is:

    [
[(1),(2),(3),(4)],
[(6),(8),(10),(12)],
[(15),(18),(21),(24)]
] 

  3. Calculate the cumulative product along the first dimension:

    AFL% cumulate(A,prod(val),row);  


    The output is:

    [
[(1),(2),(3),(4)],
[(5),(12),(21),(32)],
[(45),(120),(231),(384)]
]  

  4. Calculate the cumulative average along the second dimension:

    AFL% cumulate(A,avg(val),col); 


    The output is:

    [
[(1),(1.5),(2),(2.5)],
[(5),(5.5),(6),(6.5)],
[(9),(9.5),(10),(10.5)]
]  

  5. Calculate the the average, min, and max values along the first dimension:

    AFL% cumulate(A,avg(val),min(val),max(val),row); 


    The output is:

    [
[(1,1,1),(2,2,2),(3,3,3),(4,4,4)],
[(3,1,5),(4,2,6),(5,3,7),(6,4,8)],
[(5,1,9),(6,2,10),(7,3,11),(8,4,12)]
] 

Array Containing Empty Cells and Null Values

To calculate an array that contains empty cells and null values, do the following:

  1. Assume the following array, B, with dimensions i and j.

    AFL% scan(B) 


    The output is:

    [
[(1.25,null),(),(-7,-7)],
[(),(),(4.2,1)],
[(4.5,null),(11,9),(2.6,null)],
[(),(1.7,6.5),(6.2,null)]
] 

  2. Calculate the cumulative average along the second dimension, for both attributes.

    AFL% cumulate(B,avg(val1),avg(val2),j); 


    The output is:

    [
[(1.25,null),(),(-2.875,-7)],
[(),(),(4.2,1)],
[(4.5,null),(7.75,9),(6.03333,9)],
[(),(1.7,6.5),(3.95,6.5)]
] 

  3. Calculate the variance for the first attribute, along the first dimension.

    AFL% cumulate(B,var(val1));  


    The output is:

    [
[(null),(),(null)],
[(),(),(62.72)],
[(5.28125),(null),(36.6933)],
[(),(43.245),(34.28)]
]Â