NVIDIA Modulus Sym v1.3.0
v1.3.0

deeplearning/modulus/modulus-sym-v130/_modules/modulus/sym/utils/sympy/numpy_printer.html

Source code for modulus.sym.utils.sympy.numpy_printer

# Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""
Helper functions for converting sympy equations to numpy
"""

import types
import inspect
import numpy as np
import symengine as se
import sympy as sp

NP_LAMBDA_STORE = {}


[docs]def np_lambdify(f, r): """ generates a numpy function from a sympy equation Parameters ---------- f : Sympy Exp, float, int, bool or list of the previous the equation to convert to a numpy function. If float, int, or bool this gets converted to a constant function of value `f`. If f is a list then output for each element in list is is concatenated on axis -1. r : list, dict A list of the arguments for `f`. If dict then the keys of the dict are used. Returns ------- np_f : numpy function """ # possibly lambdify list of f if not isinstance(f, list): f = [f] # convert r to a list if dictionary # break up any tuples to elements in list if isinstance(r, dict): r = list(r.keys()) no_tuple_r = [] for key in r: if isinstance(key, tuple): for k in key: no_tuple_r.append(k) else: no_tuple_r.append(key) # lambidfy all functions in list lambdify_f = [] for f_i in f: # check if already a numpy function if isinstance(f_i, types.FunctionType): # add r inputs to function args = inspect.getargspec(f_i).args def lambdify_f_i(**x): return f_i(**{key: x[key] for key in args}) else: # check if already lambdified equation if (f_i, tuple(no_tuple_r)) in NP_LAMBDA_STORE.keys(): lambdify_f_i = NP_LAMBDA_STORE[(f_i, tuple(no_tuple_r))] else: # if not lambdify it try: if not isinstance(f_i, bool): f_i = float(f_i) except: pass if isinstance(f_i, (float, int)): # constant function def loop_lambda(constant): return ( lambda **x: np.zeros_like(next(iter(x.items()))[1]) + constant ) lambdify_f_i = loop_lambda(f_i) elif type(f_i) in [ type((se.Symbol("x") > 0).subs(se.Symbol("x"), 1)), type((se.Symbol("x") > 0).subs(se.Symbol("x"), -1)), bool, ]: # TODO hacky sympy boolian check def loop_lambda(constant): if constant: return lambda **x: np.ones_like( next(iter(x.items()))[1], dtype=bool ) else: return lambda **x: np.zeros_like( next(iter(x.items()))[1], dtype=bool ) lambdify_f_i = loop_lambda(f_i) else: try: # first try to compile with Symengine kk = [] for k in no_tuple_r: if isinstance(k, str): kk.append(se.Symbol(k)) else: kk.append(k) kk = [se.Symbol(name) for name in sorted([x.name for x in kk])] se_lambdify_f_i = se.lambdify(kk, [f_i], backend="llvm") def lambdify_f_i(**x): if len(x) == 1: v = list(x.values())[0] else: v = np.stack( [v for v in dict(sorted(x.items())).values()], axis=-1, ) out = se_lambdify_f_i(v) if isinstance(out, list): out = np.concatenate(out, axis=-1) return out except: # fall back on older SymPy compile sp_lambdify_f_i = sp.lambdify( [k for k in no_tuple_r], f_i, [NP_SYMPY_PRINTER, "numpy"] ) def lambdify_f_i(**x): v = sp_lambdify_f_i(**x) if isinstance(v, list): v = np.concatenate(v, axis=-1) return v # add new lambdified function to dictionary NP_LAMBDA_STORE[(f_i, tuple(no_tuple_r))] = lambdify_f_i # add new list of lambda functions lambdify_f.append(lambdify_f_i) # construct master lambda function for all def loop_grouped_lambda(lambdify_f): def grouped_lambda(**invar): output = [] for lambdify_f_i in lambdify_f: output.append(lambdify_f_i(**invar)) return np.concatenate(output, axis=-1) return grouped_lambda return loop_grouped_lambda(lambdify_f)

def _xor_np(x): return np.logical_xor(x) def _min_np(x, axis=None): return_value = x[0] for value in x: return_value = np.minimum(return_value, value) return return_value def _max_np(x, axis=None): return_value = x[0] for value in x: return_value = np.maximum(return_value, value) return return_value def _heaviside_np(x, x2=0): # force x2 to 0 x2 = 0 return np.heaviside(x, x2) def _equal_np(x, y): return np.isclose(x, y) NP_SYMPY_PRINTER = { "amin": _min_np, "amax": _max_np, "Heaviside": _heaviside_np, "equal": _equal_np, "Xor": _xor_np, } SYMENGINE_BLACKLIST = [sp.Heaviside, sp.DiracDelta]

© Copyright 2023, NVIDIA Modulus Team. Last updated on Jan 25, 2024.