Optimize sparse index build on partitioned table

The index build process is currently not using the information on a partitioned table when building the index:

We have a big table partitioned by range on a column.
When we build an sparse index on this table only including values in this column greater than a certain value, even though rows in a partition could never be included in the index (because of the range of values), the indexer goes through all the rows in the partition - wasting time and CPU on this.

It would be great if the indexer checked the range partition information and only processed the partitions where a hit is possible.

EDIT: Just discovered that even though no keys are in the index partition, it still occupies a lot of storage - here are the numbers from SYSINDEXSTAT for a partitioned EVI index:

PARTITION / NUMBER_KEYS / INDEX_SIZE
PART000001 / 0 / 11469369344
PART000002 / 2202143097 / 10781892608
PART000003 / 698188190 / 2835640320

So in addition to the wasted time and CPU, there is also a lot of wasted storage...

Imagine having a partitioned table TABLE1 with the columns
ID integer,
LINE integer,
STAT char(1) and other columns,
partitioned by RANGE on ID where
PART000001 is between MINVALUE and 5000 excluding,
PART000002 is between 5000 including and 10000 excluding
etc.

Creating a sparse index on TABLE1 where ID >= 7000, the indexer should not read PART000001 at all, since all values for ID is lower than 7000 and will not be included in the index anyway.