I'm running MySQL 4.0.18-nt & accessing the database primarily through Java with MySQL/Connector J (similar vintage). I am administering this myself, and can upgrade as needed. I know I'm a bit behind the times.
I apologize for a bit of background, skip to the end for the question.... I am using a database to track large numbers of computer analysis runs. The analyses are treated abstractly as black boxes (tasks) with n-inputs and m-outputs. Inputs and outputs are both treated as quantities. At the time of the database design, we have no idea how many inputs or outputs any task will have. An output of one task may be an input to another task. So, there are tables for tasks (T) and quantities (Q), and then input T/Q and output T/Q tables to create the many-to-many pairing required. As cases are run, records are entered into a case (C) table, and the settings of the inputs and outputs from the run are stored in pairs in a Case/Quantity table. The general idea is to keep a record of many computer analysis runs and interpolate new results instead of running the code again. So, if the computer analysis is something like... f = sin(x)*cos(y) And we'll pretend that this is a very expensive analysis to perform, we want to get as much out of the runs we've already made as possible. Of course, in the real world, there may be a dozen or twenty input quantities and a similar number of outputs... I need to be able to perform nearest-neighbor type searches based on the Case/Quantity values for the inputs. My plan was to just request _all_ the data from the database (making the query easy), storing the data in a K-D Tree in my application, performing the more detailed searches there. Then, I read a random (ancient) blog entry about R-Trees in MySQL 4.1 (http://jeremy.zawodny.com/blog/archives/000418.html). Which got me thinking.... Is there some way to more effectively store this type of data in MySQL. I don't need R-Trees (though they'll work), my data are all points, not rectangles... So, is there some way to tell MySQL that I'll be doing these indirect searches on the values of the quantities, so that MySQL can do this much more efficiently? I'd rather not change my database schema, but will consider it if required. Recall that we don't know the input dimensionality of the problem at the time the database is created. Thanks in advance, Rob -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
