Post-compression residual data object processing
A system includes a logic that generates a primary data object and an associated residual data object for each of a plurality of uncompressed data objects. The logic determines to remove at least some of the generated residual data objects based on a determination of a constrained adjustment that maintains a net effect of the removed residual data objects. The logic removes the at least some of the generated residual data objects determined for removal and applies the constrained adjustment to at least one of the generated primary data objects.
1 . A computer-implemented data compression method including:
obtaining, by a processor, a plurality of uncompressed data objects; and
compressing, by the processor, the uncompressed data objects by:
generating, by the processor, a plurality of residual data objects by decomposing a plurality of precision data objects into the plurality of residual data objects and a plurality of primary data objects, the plurality of precision data objects generated via combination of groups of the uncompressed data objects, each of the primary data objects having a data size smaller than one of the precision data objects from which that primary data object was generated;
generating, by the processor, a primary data object and an associated residual data object for each of the precision compression data objects;
determining, by the processor, to remove at least a first portion of the plurality of residual objects;
determining an adjustment to a first one of the plurality of primary data objects, the adjustment having a compensating effect maintaining a net effect of the first portion of the plurality of residual objects; and
applying, by the processor, the adjustment to the first one of the plurality of primary data objects to effect removal of the first portion of the plurality of residual objects thereby reducing the total data size, relative to that of the uncompressed data objects, of the plurality of primary data objects and any remaining ones the plurality of residual data objects.
2 . The computer-implemented data compression method of claim 1 , wherein the compensating effect maintains flow neutrality.
3 . The computer-implemented data compression method of claim 1 , wherein the compensating effect maintains a fixed rate magnitude.
4 . The computer-implemented data compression method of claim 1 , wherein the first one of the plurality primary data objects includes fewer decimal-precision bits than a corresponding precision compression data object.
5 . The computer-implemented data compression method of claim 1 , further including determining the adjustment to the first one of the plurality of primary data objects by performing a local search using a minimization of a penalty parameter among a set of candidate adjustments.
6 . The computer-implemented data compression method of claim 5 , wherein the penalty parameter is based on:
a change to a notional magnitude;
a change to a rate magnitude; and/or
a change to a flow magnitude.
7 . The computer-implemented data compression method of claim 5 , wherein performing the local search includes applying a clean-up procedure by accounting for an account effect from a previously applied adjustment.
8 . The computer-implemented data compression method of claim 5 , wherein performing the local search includes sorting the plurality of residual data objects into a particular order.
9 . The computer-implemented data compression method of claim 8 , wherein sorting the plurality of residual data objects into the particular order includes ordering the plurality of residual data objects based on a size parameter.
10 . The computer-implemented data compression method of claim 1 , wherein applying the adjustment to the first one of the plurality of primary data objects includes applying a risk-free netting operation to the plurality of primary data objects and the plurality of residual data objects.
11 . Non-transitory computer-readable media configured to storage instructions thereon, the instructions configured to, when executed, cause a processor to:
obtain a plurality of uncompressed data objects; and
compress the uncompressed data objects by:
generate a plurality of residual data objects by decomposing a plurality of precision data objects into the plurality of residual data objects and a plurality of primary data objects, the plurality of precision data objects generated via combination of groups of the uncompressed data objects, each of the primary data objects having a data size smaller than one of the precision data objects from which that primary data object was generated;
generate a primary data object and an associated residual data object for each of the precision compression data objects;
determine to remove at least a first portion of the plurality of residual objects;
determine an adjustment to a first one of the plurality of primary data objects, the adjustment having a compensating effect maintaining a net effect of the first portion of the plurality of residual objects; and
apply the adjustment to the first one of the plurality of primary data objects to effect removal of the first portion of the plurality of residual objects thereby reducing the total data size, relative to that of the uncompressed data objects, of the plurality of primary data objects and any remaining ones the plurality of residual data objects.
12 . The non-transitory computer-readable media of claim 11 , wherein the compensating effect maintains flow neutrality.
13 . The non-transitory computer-readable media of claim 11 , wherein the compensating effect maintains a fixed rate magnitude.
14 . The non-transitory computer-readable media of claim 11 , wherein the first one of the plurality primary data objects includes fewer decimal-precision bits than a corresponding precision compression data object.
15 . The non-transitory computer-readable media of claim 11 , wherein the instructions are further configured to cause the processor to determine the adjustment to the first one of the plurality of primary data objects by performing a local search using a minimization of a penalty parameter among a set of candidate adjustments.
16 . The non-transitory computer-readable media of claim 15 , wherein the penalty parameter is based on:
a change to a notional magnitude;
a change to a rate magnitude; and/or
a change to a flow magnitude.
17 . The non-transitory computer-readable media of claim 15 , wherein the instructions are further configured to cause the processor to perform the local search by applying a clean-up procedure by accounting for an account effect from a previously applied adjustment.
18 . The non-transitory computer-readable media of claim 15 , wherein the instructions are further configured to cause the processor to perform the local search by sorting the plurality of residual data objects into a particular order.
19 . The non-transitory computer-readable media of claim 18 , wherein the instructions are further configured to cause the processor to sort the plurality of residual data objects into the particular order by ordering the plurality of residual data objects based on a size parameter.
20 . A data compression system including:
means for obtaining a plurality of uncompressed data objects; and
means for compressing the uncompressed data objects including:
means for generating a plurality of residual data objects by decomposing a plurality of precision data objects into the plurality of residual data objects and a plurality of primary data objects, the plurality of precision data objects generated via combination of groups of the uncompressed data objects, each of the primary data objects having a data size smaller than one of the precision data objects from which that primary data object was generated;
means for generating a primary data object and an associated residual data object for each of the precision compression data objects;
means for determining to remove at least a first portion of the plurality of residual objects;
means for determining an adjustment to a first one of the plurality of primary data objects, the adjustment having a compensating effect maintaining a net effect of the first portion of the plurality of residual objects; and
means for applying the adjustment to the first one of the plurality of primary data objects to effect removal of the first portion of the plurality of residual objects thereby reducing the total data size, relative to that of the uncompressed data objects, of the plurality of primary data objects and any remaining ones the plurality of residual data objects.