1. A method executed by a computer, comprising:
generating, by the computer, performance characteristic metric vectors representing query execution metrics for a database;
given a point in a feature input space that corresponds to a first one of the performance characteristic metric vectors, identifying, by the computer, nearest neighbors of the point and corresponding points in a performance metric feature space;
determining, by the computer, a generated point that represents a cluster represented by the corresponding points; and
using, by the computer, the performance characteristic metric vectors that correspond to the generated point to predict performance of a query on a first database system having a first configuration of hardware and software and on a second database system having a second configuration of hardware and software, prediction occurring before the query executes on the first database system and before the query executes on the second database system, the second configuration being different than the first configuration.
2. The method of claim 1, wherein the generated point represents a center of the cluster.
3. The method of claim 1, further comprising storing a map mapping the performance characteristics metric vectors from an input space to a feature space.
4. The method of claim 1, further comprising using a kernel function to map the performance characteristics metric vectors from an input space to a feature space.
5. The method of claim 1, further comprising populating a feature space with a set of likely points from an input space.
6. The method of claim 1, further comprising performing a reverse mapping to identify coordinates in an input space to predict performance of the query.
7. A tangible computer readable storage disc or device comprising instructions which, when executed, cause a computer to at least:
obtain points for an input space for queries for a database;
map the points from the input space to a feature space; and
given locations in the feature space, use reverse mapping to identify corresponding locations in the input space for the locations in the feature space to predict execution performance of queries on a first database system having a first configuration of hardware and software and on a second database system having a second configuration of hardware and software, prediction occurring before the queries execute on the first database system and before the query executes on the second database system, the second configuration being different than the first configuration.
8. The tangible computer readable storage disc or device of claim 7, wherein the instructions cause the computer to perform the reverse mapping by performing regression on a known set of vectors in the feature space to identify parameters that predict corresponding vectors in the input space.
9. The tangible computer readable storage disc or device of claim 7, further comprising instructions which, when executed, cause the computer to at least:
impose boundaries for grouping similar points in the feature space; and
calculate ranges for vectors in the input space that define clusters of corresponding points in the feature space to generate a range of possible values.
10. The tangible computer readable storage disc or device of claim 7, further comprising instructions which, when executed, cause the computer to at least:
apply a correlation function to the feature space to generate a second feature space;
identify nearest neighbors to a point in the feature space; and
use the correlation function to identify nearest neighbors in the second feature space that correspond to the nearest neighbors in the feature space.
11. The tangible computer readable storage disc or device of claim 7, further comprising instructions which, when executed, cause the machine to at least store a map generated by the kernel function to map the points from the input space to the feature space.
12. A computer system, comprising:
a database;
a memory; and
a processor programmed to:
generate points from an input space obtained from a query for a database;
map the points from the input space to a feature space;
given a point in the feature space, perform a reverse mapping to identify coordinates in the input space for the point;
predict performance of the query on a first database system having a first configuration of hardware and software before the query is executed on the first database system; and
predict performance of the query on a second database system having a second configuration of hardware and software before the query is executed on the second database system, the second configuration being different than the first configuration.
13. The computer system of claim 12, wherein the processor is programmed to determine a nearest point in the input space when the point in the feature space does not exist in the input space.
14. The computer system of claim 12, wherein the processor is programmed to populate the feature space with all possible points from the input space.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. A laser-scanning fluoroscopy apparatus comprising:
a laser light source for emitting laser beams with a plurality of wavelengths;
a spectroscopic device for splitting the laser beams emitted from the laser light source according to wavelength;
a focusing lens for focusing the laser beams split by the spectroscopic device;
a mirror device including:
a plurality of first reflection sections, disposed near focal positions of the focusing lens, spaced out in a split direction of the spectroscopic device to reflect the laser beams with different wavelengths in a first direction; and
a second reflection section arranged adjacent to the first reflection sections to reflect incident light in a second reflection direction;
a diffraction grating for combining the laser beams reflected by the first reflection sections in the first direction;
a scanning section for two-dimensionally scanning the laser beams combined by the diffraction grating;
an objective optical system for focusing the laser beams scanned by the scanning section onto a tissue; and
a photodetector for detecting fluorescence emitted from the tissue, the fluorescence returning through the objective optical system, the scanning section, and the diffraction grating, and being reflected at the second reflection section of the mirror device, wherein
a width dimension, as measured along the split direction, of each first reflection section of the mirror device is smaller than the width dimension of the second reflection section.
2. The laser-scanning fluoroscopy apparatus according to claim 1, wherein the mirror device includes a plurality of movable mirrors such that the first reflection sections and the second reflection section can be relocated.