Sparse Cholesky factorization algorithms use a fill-reducing
ordering to permute the columns of the Jacobian matrix. There
are two ways of doing this.
1. Compute the Jacobian matrix in some order and then have the
factorization algorithm permute the columns of the Jacobian.
2. Compute the Jacobian with its columns already permuted.
The first option incurs a significant memory penalty. The
factorization algorithm has to make a copy of the permuted
Jacobian matrix.
Starting with this change Ceres pre-permutes the columns of the
Jacobian matrix and generally speaking, there is no performance
penalty for doing so.
In some rare cases, it is worth using a more complicated
reordering algorithm which has slightly better runtime
performance at the expense of an extra copy of the Jacobian
matrix. Setting Solver::Options::use_postordering to true
enables this tradeoff.
This change also removes Solver::Options::use_block_amd
as an option. All matrices are ordered using their block
structure. The ability to order them by their scalar
sparsity structure has been removed.
Here is what performance on looks like on some BAL problems.
Memory
======
HEAD pre-ordering
16-22106 137957376.0 113516544.0
49-7776 56688640.0 46628864.0
245-198739 1718005760.0 1383550976.0
257-65132 387715072.0 319512576.0
356-226730 2014826496.0 1626087424.0
744-543562 4903358464.0 3957878784.0
1024-110968 968626176.0 822071296.0
Time
====
HEAD pre-ordering
16-22106 3.8 3.7
49-7776 1.9 1.8
245-198739 82.6 81.9
257-65132 14.0 13.4
356-226730 98.8 95.8
744-543562 325.2 301.6
1024-110968 42.1 37.1
Change-Id: I6b2e25f3fed7310f88905386a7898ac94d37467e
Add automatic recursive independent set decomposition.
Clean up the naming and the API for inner iterations.
Change-Id: I3d7d6babb9756842d7367e14b7279d2df98fb724
A non-linear generalization of Ruhe & Wedin's algorithm
for separable non-linear least squares problem. It is implemented
as coordinate descent on an independent subset of the parameter
blocks at the end of every successful Newton step. The resulting
algorithm has much improved convergence at the cost of some
execution time.
Change-Id: I8fdc5edbd0ba1e702c9658b98041b2c2ae705402
Remove the old ordering API, and modify solver_impl.cc
to use the new API everywhere.
In the process also clean up the linear solver instantion
logic in solver_impl.cc a bit too.
Change-Id: Ia66898abc7f622070b184b21fce8cc6140c4cebf
More flexible testing.
Read and parse the certified cost value from the data file.
Remove the ugly hack for computing the certified cost.
Refactored the flags parsing logic
Change-Id: I8f2e6be183b758b2453302fcdc6696bfa0db5eb8
For problems with a small number of variables, but a large
number of residuals, it is sometimes beneficial to use the
Cholesky factorization on the normal equations, instead of
the dense QR factorization of the Jacobian, even though it
is numerically the better thing to do.
Change-Id: I3506b006195754018deec964e6e190b7e8c9ac8f
Refactor some of the perturbation code so that the
normalization and perturbation code can share the
camera decomposition code.
Change-Id: I084064976804a92f9240d8f5e10d1bb23dcb5ff2
1. Add the ability to perturb the camera pose and the
point positions using user specified parameters.
2. Re-order the flags.
3. Minor name correction.
4. Added Box-Mueller generator to random.h
Change-Id: I2c9ce74c237f5bde9a7299cc71b205d1ca9bc742
Non-monotonic trust region algorithm based on the work of Phil Toint, as
described in
Non-monotone trust region algorithms for nonlinear
optimization subject to convex constraints.
Philippe L. Toint
Mathematical Programming 77 (1997), 69-94.
Change-Id: I199ecc644e8d1a8cb43666052aef66fb93e15569
1. A new dogleg trust region strategy.
2. Consistent naming of all variables taking and reporting
time. Also all are doubles now.
3. Enum to stringification routines.
4. bundle_adjuster.cc accepts max solver time and trust_region_strategy.
5. Time accounting is pushed into solver_impl.cc and there is now
postprocessing time accounted for explicitly.
6. IterationCallback now has cumulative time.
7. LoggingCallback logs per iteration and cumulative time.
8. TrustRegionStrategy now allows for Invalid steps to be indicated
explicitly.
9. Trust region minimizer actually terminates on max_solver_time.
Change-Id: I7e3b82c8beebc17b6b355ea46ddd280754a2d8b2
By virtue of the modeling layer in Ceres being block oriented,
all the matrices used by Ceres are also block oriented.
When doing sparse direct factorization of these matrices, the
fill-reducing ordering algorithms can either be run on the
block or the scalar form of these matrices. Running it on the
block form exposes more of the super-nodal structure of the
matrix to the Cholesky factorization routines. This leads to
substantial gains in factorization performance.
This changelist adds support for approximate minimium degree
orderings to be computed on the block structure of the
Schur complement matrix. This affects, SchurComplementSolver
and VisibilityBasedPreconditioner and SparseNormalCholesky
when using SuiteSparse.
A bool, use_block_amd has been added to Solver::Options and
bundle_adjuster.cc has been updated to allow testing with it.
When combined with a multithreaded Schur elimination, speed ups
can be seen quite uniformly across the board. For some problems
this can be dramatic, reducing the factorization time from 70
seconds down to 17 seconds.
Change-Id: I15ebb0afcbc85ada032ec8d179ee3a2f7c8d3e46
1. New TrustRegionMinimizer and basic tests for it.
2. New TrustRegionStrategy interface.
3. New LevenbergMarquardtStrategy and tests for it.
4. Updates to SolverImpl to reflect this.
5. Changes to Solver::Options and IterationSummary related to this.
6. Deleted levenberg_marquardt.cc/h/_test.cc
Change-Id: I6c1d1a7c774f014856f9f26263a830aa886e1400
- Rename BlockDiagonalPreconditioner to BlockJacobiPreconditioner
- Include the diagonal in the block jacobi preconditioner.
- Better flag help for eta.
- Enable test for CGNR
- Rename CONJUGATE_GRADIENTS to CGNR.
- etc.
This adds a new LinearOperator which implements symmetric
products of a matrix, and a new CGNR solver to leverage
CG to directly solve the normal equations. This also
includes a block diagonal preconditioner. In experiments
on problem-16, the non-preconditioned version is about
1/5 the speed of SPARSE_SCHUR, and the preconditioned
version using block cholesky is about 20% slower than
SPARSE_SCHUR.
Sameer Agarwal