Regarding: building a 1000000 row x 2000 column matrix
If by chance these are sparse matrices (i.e., the majority of the values
are empty) then conceivably you could store only the available values.
The schema might then be something like:
CREATE TABLE myTable (
matrixRow INTEGER,
matrixCol INTEGER,
sample REAL );
I imagine you'd want to create an index on matrixRow after import.
If the matrices are NOT sparse, then you'd need 2 thousand million
inserts per matrix (2 billion in USA parlance) and more disk space than
your original schema -- probably not fun.
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
