syrte
commented
1 year ago My apology! I found the explanation in solvers/_hodlr.cpp.
min_size (Optional[int]): The block size where the solver switches to a
general direct factorization algorithm. This can be tuned for platform
and problem specific performance and accuracy. As a general rule,
larger values will be more accurate and slower, but there is some
overhead for very small values, so we recommend choosing values in the
hundreds. (default: ``100``)
tol (Optional[float]): The precision tolerance for the low-rank
approximation. This value is used as an approximate limit on the
Frobenius norm between the low-rank approximation and the true matrix
when reconstructing the off-diagonal blocks. Smaller values of ``tol``
will generally give more accurate results with higher computational
cost. (default: ``0.1``)
seed (Optional[int]): The low-rank approximation method within the HODLR
algorithm is not deterministic and, without a fixed seed, the method
can give different results for the same matrix. Therefore, we require
that the user provide a seed for the random number generator.
(default: ``42``, obviously)Nevertheless, it might be nice to put these into the python docstring as well.
dfm
Hi Dan, thanks for the superb algorithm and package! I am confused about the meaning of the
HODLRSolverparametersmin_size=100, tol=0.1, seed=42. As far as I noticed, tol=0.1 may lead to some difficulties in optimizing the parameters sometimes, and a smaller tol e.g., 1e-3 can help.I confess that I didn't go through the original algorithm paper (I skimmed quickly, but the answer is not obvious to me). But I did search through the repo, issues, and document. What do they mean and when should I change the default numbers? Maybe it would be helpful for the users if adding some more explanation in the documentation. Thanks again!