Misc ToolsΒΆ
Misc toolsΒΆ
- class sumpy.tools.BlockIndexRanges(cl_context, indices, ranges)[source]ΒΆ
Convenience class for working with blocks of a global array.
- indicesΒΆ
A list of not necessarily continuous or increasing integers representing the indices of a global array. The individual blocks are delimited using
ranges
.
- class sumpy.tools.MatrixBlockIndexRanges(cl_context, row, col)[source]ΒΆ
Keep track of different ways to index into matrix blocks.
- rowΒΆ
A
BlockIndexRanges
encapsulating row block indices.
- colΒΆ
A
BlockIndexRanges
encapsulating column block indices.
- block_take(x, i)[source]ΒΆ
Retrieve a block from a linear representation of the matrix blocks. A linear representation of the matrix blocks can be obtained, or should be consistent with
i = index.linear_row_indices() j = index.linear_col_indices() linear_blks = global_mat[i, j] for k in range(index.nblocks): assert np.allclose(index.block_take(linear_blks, k), index.take(global_mat, k))
- Parameters
x β a 1D
numpy.ndarray
.i β block index.
- Returns
requested block, reshaped into a 2D array.
- get(queue=None)[source]ΒΆ
Transfer data to the host. Only the initial given data is transfered, not the arrays returned by
linear_row_indices()
and friends.- Returns
a copy of self in which all data lives on the host, i.e. all
pyopencl.array.Array
instances are replaces bynumpy.ndarray
instances.
- linear_row_indicesΒΆ
- take(x, i)[source]ΒΆ
Retrieve a block from a global matrix.
- Parameters
x β a 2D
numpy.ndarray
.i β block index.
- Returns
requested block from the matrix.
- class sumpy.tools.ExprDerivativeTaker(expr, var_list, rscale=1, sac=None)[source]ΒΆ
Facilitates the efficient computation of (potentially) high-order derivatives of a given
sympy
expression expr while attempting to maximize the number of common subexpressions generated.This class defines the interface and realizes a baseline implementation. More specialized implementations may offer better efficiency for special cases.
- diff(mi)[source]ΒΆ
Take the derivative of the expression represented by
ExprDerivativeTaker
.- Parameters
mi β multi-index representing the derivative
- class sumpy.tools.LaplaceDerivativeTaker(expr, var_list, rscale=1, sac=None)[source]ΒΆ
Specialized derivative taker for Laplace potential.
- class sumpy.tools.RadialDerivativeTaker(expr, var_list, rscale=1, sac=None)[source]ΒΆ
Specialized derivative taker for radial expressions.
- class sumpy.tools.HelmholtzDerivativeTaker(expr, var_list, rscale=1, sac=None)[source]ΒΆ
Specialized derivative taker for Helmholtz potential.
- class sumpy.tools.DifferentiatedExprDerivativeTaker(taker, derivative_transformation)[source]ΒΆ
Implements the
ExprDerivativeTaker
interface for an expression that is itself a linear combination of derivatives of a base expression. To take the actual derivatives, it makes use of an underlying derivative taker taker.- Parameters
taker (sumpy.tools.ExprDerivativeTaker) β
derivative_transformation (dict) β
- Return type
- takerΒΆ
- A :class:`ExprDerivativeTaker` for the base expression.
- derivative_transformationΒΆ
- A dictionary mapping a derivative multi-index to a coefficient.
- The expression represented by this derivative taker is the linear
- combination of the derivatives of the expression for the
- base expression.
InstallationΒΆ
This command should install sumpy
:
pip install sumpy
You may need to run this with sudo. If you donβt already have pip, run this beforehand:
curl -O https://raw.github.com/pypa/pip/master/contrib/get-pip.py
python get-pip.py
For a more manual installation, download the source, unpack it, and say:
python setup.py install
In addition, you need to have numpy
installed.
UsageΒΆ
Environment variablesΒΆ
Name |
Purpose |
---|---|
SUMPY_FORCE_SYMBOLIC_BACKEND |
Symbolic backend control, see Symbolic backends |
SUMPY_NO_CACHE |
If set, disables the on-disk cache |
SUMPY_NO_OPT |
If set, disables performance-oriented |
Symbolic backendsΒΆ
sumpy
supports two symbolic backends: sympy and SymEngine. To use the
SymEngine backend, ensure that the SymEngine library and the SymEngine Python bindings are installed.
By default, sumpy
prefers using SymEngine but falls back to sympy if it
detects that SymEngine is not installed. To force the use of a particular
backend, set the environment variable SUMPY_FORCE_SYMBOLIC_BACKEND to
symengine or sympy.
User-visible ChangesΒΆ
Version 2016.1ΒΆ
Note
This version is currently under development. You can get snapshots from sumpyβs git repository
Initial release.
LicenseΒΆ
sumpy
is licensed to you under the MIT/X Consortium license:
Copyright (c) 2012-16 Andreas KlΓΆckner
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the βSoftwareβ), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED βAS ISβ, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Frequently Asked QuestionsΒΆ
The FAQ is maintained collaboratively on the Wiki FAQ page.
AcknowledgmentsΒΆ
Work on meshmode was supported in part by
the US National Science Foundation under grant numbers DMS-1418961, DMS-1654756, SHF-1911019, and OAC-1931577.
AK also gratefully acknowledges a hardware gift from Nvidia Corporation.
The views and opinions expressed herein do not necessarily reflect those of the funding agencies.