Add fast operator evaluation for tensor-product discretizations by a-alveyblanc · Pull Request #362 · inducer/grudge

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc · 2024-09-06T20:49:38Z

Supersedes #313, #354

Adds:

Fast operator evaluation for tensor-product discretizations
Metadata (tags) relevant to tensor-product discretizations used during (eager and lazy) compilation
WADG + Overintegration
Evaluation of operators using quadrature

~~Non-essential transformations will be added in a later PR. This is to keep this (already large) PR manageable.~~ The plan is to break this PR up into a sequence of smaller PRs to ease the review process. Keeping this up until that starts happening.

TODOs:

cc @MTCam

…product-operators

…o only one axis

inducer · 2024-09-07T19:51:36Z

Should this be marked draft given that there are pending TODOs?

…product-operators

…ation

…ated case

…sforms

inducer · 2024-11-25T17:23:07Z

+# {{{ BilinearForm base class
+
+@dataclass
+class _BilinearForm:


~~Classes~~

inducer · 2024-11-25T17:40:08Z

+ new_args = []
+ new_access_descrs = []
+ for iarg, arg in enumerate(expr.args):
+ # we assume that the 0th argument will be the one with this tag


So check that?
Generally, beef up the matching. (maybe come up with some tools, match statement)

inducer · 2024-11-25T17:42:25Z

+ included in the original einsum), and properly reshapes to and from
+ tensor-product form to apply the 1D mass operator.
+ """
+ def map_einsum(self, expr, *args, **kwargs):


Don't do this varargs-style.

inducer · 2024-11-25T17:50:09Z

+ return expr.copy(args=tuple(new_args))
+
+
+class InverseMassDistributor(CopyMapperWithExtraArgs):


Split into two mappers (with the inner only handling what's allowed as part of distribution)
Bonus points for generalizing to simplex

inducer · 2024-11-25T17:51:45Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


Split in two parts

inducer · 2024-11-25T18:02:01Z

+ return expr.copy(args=tuple(new_args))
+
+
+class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):


inducer · 2024-12-16T17:37:13Z

- in_element_group: InterpolatoryElementGroupBase):
+ in_element_group: InterpolatoryElementGroupBase) -> ArrayOrContainer:

 @keyed_memoize_in(


Let's have @keyed_memoize_in_first_arg.

inducer · 2024-12-16T17:41:47Z

+ else:
+ quadrature_rule = in_grp.quadrature_rule()
+
+ if isinstance(in_grp, SimplexElementGroupBase):


Maybe the TP and full versions want to be separate functions?

inducer · 2024-12-16T17:54:11Z


-def _strong_scalar_grad(dcoll, dd_in, vec):
- assert isinstance(dd_in.domain_tag, VolumeDomainTag)
+def _reference_stiffness_transpose_matrices(


Unify with above via basis_getter?

inducer · 2024-12-16T17:55:49Z


- discr = dcoll.discr_from_dd(dd_in)
- actx = vec.array_context
+ if in_grp == out_grp:


Suggested change
ifin_grp==out_grp:
ifisinstance(in_grp, Quadrature)
?
Suggested change
ifin_grp==out_grp:
ifnotisinstance(in_grp, Interpolatory)
?

inducer · 2024-12-16T17:58:09Z

+ out_grp.shape
+ )
+ else:
+ quadrature_rule = in_grp.quadrature_rule()


Check that quadrature_rule provides sufficient exactness.

inducer · 2024-12-16T18:01:09Z

+ return get_reference_stiffness_transpose_matrices(out_grp, in_grp)
+
+
+def reference_mass_matrix(actx: ArrayContext,


Rewire to use above machinery.

inducer · 2024-12-16T18:02:55Z

+ """
+
+ per_group_grads = []
+ for out_grp, in_grp, vec_i, ijm_i in zip(


cough comprehension cough

inducer · 2025-01-08T17:58:12Z

+ use_tensor_product_fast_eval=use_tensor_product_fast_eval
+ )
+
+ if input_group == output_group:


if isinstance(input_group, Interpolatory):

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

a-alveyblanc added 17 commits July 24, 2024 21:47

start updating grad
8bd3969

add TODOs
0afe732

add strong form tensor product gradient
9c4c150

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
c264cf9
…product-operators

start working on weak-overint case for TP elements
e105a64

start brainstorming better ways to handle weak overintegration
a3f7f7e

small changes
830c2e0

do not support overintegration (yet)
102f6e0

fix gradient tests
597fcc9

fix tensor product gradient
790ab13

do not compute stiffness matrix; apply mass to all axes and diff op t…
91e3121
…o only one axis

arg name change
9ff324b

add fast operator eval for weak and strong divergence
7b50e32

add tensor product inverse mass
230c1bd

tag mass operator as such
30f984e

add fast operator eval for mass operator application
92b54a2

toward overintegration support
34b1cd6

a-alveyblanc marked this pull request as draft September 7, 2024 21:27

a-alveyblanc added 11 commits September 8, 2024 09:53

minor changes to adjust for changes in pytato
b8e312e

adjust operators to accept 1D tensor product discretizations
a79a603

Merge branch 'main' ofhttps://github.com/inducer/grudgeinto tensor-…
51884a7
…product-operators

checkpoint before adding wadg + overintegration + fast operator evalu…
c8d0a92
…ation

tag axes of grad result
ad0d2e4

add WADG + overintegration for simplices
ca22480

add WADG + overintegration for simplices
092a6cc

start fixing up fast operator eval + overintegration + wadg
bea91ad

small changes
8182571

overintegration + fast operator evaluation
4cacc6b

first round of clean-ups
73ba29a

a-alveyblanc added 18 commits October 28, 2024 09:40

some ruff fixes
0a91985

fix failing MPI wave op test
53abda5

add redundant mass/inverse mass mappers
943f9ca

remove 2x refinedinducergh-339mesh
fa5b24e

toward transformations
5c74b6e

resolve merge conflicts
8898863

some dag rewriter work; restrict fast operator eval to non-overintegr…
19f7a49
…ated case

rewrite operators to be more predictable in the DAG
2dfd7ff

remove tagging for now
9359246

bypass WADG if base and quad discretizations are the same
da6dff7

initial algebraic dag xforms for tp
3f88f04

algebraic transforms v0.1
5bd67ed

basic (more like primitive) parallelization scheme; add more dag tran…
5a95360
…sforms

ruff fixes
8c5be5a

changes to make ruff and pylint happy
aad468a

ruff fix
7f8fb0e

remove ghost nodes left over after TP DAG xforms
654dd36

update some docs; fix dim from num faces computation
7819616

inducer reviewed Nov 25, 2024
View reviewed changes

a-alveyblanc mentioned this pull request Dec 14, 2024
Add generic nodal/modal bilinear form routines inducer/modepy#103
Merged

op refactor updates + transform updates
6a84f74

inducer reviewed Dec 16, 2024
View reviewed changes

a-alveyblanc added 7 commits December 20, 2024 12:52

move matrices to their own file and update operators accordingly
f6fd35d

all tests passing except mpi tests
5b23379

all tests passing; disable xforms for now
3b69323

attempt to resolve missing axis tags; minor transform changes
65cfe02

track down final culprit for untagged axes
d46246a

add name hints and propert tags to matrices
a70cdb3

re-implementation of tp transforms
567637c

inducer reviewed Jan 8, 2025
View reviewed changes

		return expr.copy(args=tuple(new_args))


		class InverseMassDistributor(CopyMapperWithExtraArgs):

		return expr.copy(args=tuple(new_args))


		class RedundantMassTimesMassInverseRemover(CopyMapperWithExtraArgs):

		return get_reference_stiffness_transpose_matrices(out_grp, in_grp)


		def reference_mass_matrix(actx: ArrayContext,

Conversation

a-alveyblanc commented Sep 6, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

inducer commented Sep 7, 2024

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

a-alveyblanc commented Sep 6, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

inducer commented Sep 7, 2024

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

a-alveyblanc commented Sep 6, 2024•
edited
Loading

a-alveyblanc commented Sep 6, 2024•
edited
Loading

a-alveyblanc commented Sep 6, 2024•
edited
Loading