TensorEvaluator

`TensorEvaluator`

TensorEvaluator()

An optimized evaluator for symbolic tensor expressions.

This class provides efficient numerical evaluation of symbolic tensor expressions after optimization. It supports both real and complex-valued evaluations.

Create instances using the Tensor.evaluator() method rather than directly.

Examples

evaluator = my_tensor.evaluator(constants={}, funs={}, params=[x, y])
results = evaluator.evaluate([[1.0, 2.0], [3.0, 4.0]])

Methods

Name	Description
`compile`	Compile the evaluator to a shared library using C++ for maximum performance
`evaluate`	Evaluate the tensor expression for multiple real-valued parameter inputs.
`evaluate_complex`	Evaluate the expression for multiple inputs and return the results.

`compile`

TensorEvaluator.compile(
    function_name: builtins.str,
    filename: builtins.str,
    library_name: builtins.str,
    inline_asm: builtins.str = ...,
    optimization_level: builtins.int = ...,
    compiler_path: typing.Optional[builtins.str] = None,
    custom_header: typing.Optional[builtins.str] = None,
) -> CompiledTensorEvaluator

Compile the evaluator to a shared library using C++ for maximum performance.

Generates optimized C++ code with optional inline assembly and compiles it into a shared library that can be loaded for extremely fast evaluation.

Parameters

function_name (str) Name for the generated C++ function
filename (str) Path for the generated C++ source file
library_name (str) Name for the compiled shared library
inline_asm (str, optional) Type of inline assembly optimization (“default”, “x64”, “aarch64”, “none”)
optimization_level (int, optional) Compiler optimization level 0-3 (default: 3)
compiler_path (str, optional) Path to specific C++ compiler (default: system default)
custom_header (str, optional) Additional C++ header code to include

Returns

CompiledTensorEvaluator A compiled evaluator for maximum performance evaluation

Examples

compiled = evaluator.compile(
    function_name="fast_eval",
    filename="tensor_eval.cpp",
    library_name="tensor_lib",
    optimization_level=3,
)
results = compiled.evaluate_complex([[1.0, 2.0], [3.0, 4.0]])

`evaluate`

TensorEvaluator.evaluate(inputs: typing.Sequence[typing.Sequence[builtins.float]]) -> builtins.list[Tensor]

Evaluate the tensor expression for multiple real-valued parameter inputs.

Parameters

inputs (list of list of float) List of parameter value lists, where each inner list contains numerical values for all parameters in the same order as specified when creating the evaluator

Returns

list of Tensor List of evaluated tensors, one for each input parameter set

Raises

ValueError: If the evaluator contains complex coefficients

Examples

results = evaluator.evaluate([[1.0, 2.0], [3.0, 4.0]])

`evaluate_complex`

TensorEvaluator.evaluate_complex(inputs: typing.Sequence[typing.Sequence[builtins.complex]]) -> builtins.list[Tensor]

Evaluate the expression for multiple inputs and return the results.