Source code for csdl_alpha.src.operations.mult

from csdl_alpha.src.operations.operation_subclasses import ElementwiseOperation
import csdl_alpha.utils.testing_utils as csdl_tests
from csdl_alpha.utils.inputs import variablize, validate_and_variablize
from csdl_alpha.src.graph.operation import Operation, set_properties 
from csdl_alpha.utils.typing import VariableLike

class Mult(ElementwiseOperation):

    def __init__(self,x,y):
        super().__init__(x,y)
        self.name = 'mult'

    def compute_inline(self, x, y):
        return x*y

    def compute_jax(self, x, y):
        return self.compute_inline(x, y)
    
    def evaluate_vjp(self,cotangents, x, y, z):
        if cotangents.check(x):
            cotangents.accumulate(x, cotangents[z]*y)
        if cotangents.check(y):
            cotangents.accumulate(y, cotangents[z]*x)

class BroadcastMult(Operation):

    def __init__(self,x,y):
        super().__init__(x,y)
        self.name = 'broadcast_mult'
        out_shapes = (y.shape,)
        self.set_dense_outputs(out_shapes)

    def compute_inline(self, x, y):
        return x*y

    def compute_jax(self, x, y):
        return self.compute_inline(x, y)
    
    def evaluate_vjp(self, cotangents, x, y, z):
        if cotangents.check(x):
            cotangents.accumulate(x, cotangents[z].inner(y))
        if cotangents.check(y):
            cotangents.accumulate(y, x*cotangents[z])

[docs]def mult(x,y): """Elementwise multiplication of two tensors x and y. Parameters ---------- x : Variable y : Variable Returns ------- out: Variable Examples -------- >>> recorder = csdl.Recorder(inline = True) >>> recorder.start() >>> x = csdl.Variable(value = np.array([1.0, 2.0, 3.0])) >>> y = csdl.Variable(value = np.array([4.0, 5.0, 6.0])) >>> csdl.mult(x, y).value array([ 4., 10., 18.]) >>> (x * y).value # equivalent to the above array([ 4., 10., 18.]) >>> (x * 2.0).value # broadcasting is also supported array([2., 4., 6.]) """ x = validate_and_variablize(x, raise_on_sparse = False) y = validate_and_variablize(y, raise_on_sparse = False) if x.shape == y.shape: op = Mult(x,y) elif x.size == 1: op = BroadcastMult(x.flatten(),y) elif y.size == 1: op = BroadcastMult(y.flatten(),x) else: raise ValueError(f'Shapes {x.shape} and {y.shape} not compatible for mult operation.') return op.finalize_and_return_outputs()
class TestMult(csdl_tests.CSDLTest): def test_functionality(self,): self.prep() import csdl_alpha as csdl import numpy as np x_val = 3.0 y_val = 2.0 x = csdl.Variable(name = 'x', value = x_val) y = csdl.Variable(name = 'y', value = y_val) compare_values = [] # add scalar variables s1 = csdl.mult(x,y) t1 = np.array([x_val*y_val]) compare_values += [csdl_tests.TestingPair(s1, t1, tag = 's1')] s1 = x*y compare_values += [csdl_tests.TestingPair(s1, t1)] # add scalar constants s2 = csdl.mult(3.0, 2.0) compare_values += [csdl_tests.TestingPair(s2, t1, tag = 's2')] # add scalar constant and scalar variable s3 = csdl.mult(3.0, y) compare_values += [csdl_tests.TestingPair(s3, t1, tag = 's3')] s3 = 3.0*y compare_values += [csdl_tests.TestingPair(s3, t1, tag = 's3')] s3 = y*3.0 compare_values += [csdl_tests.TestingPair(s3, t1, tag = 's3')] # add tensor constants s4 = csdl.mult(3.0*np.ones((3,2)), 2.0*np.ones((3,2))) t2 = 6.0 * np.ones((3,2)) compare_values += [csdl_tests.TestingPair(s4, t2, tag = 's4')] # add scalar constant and tensor constant s5 = csdl.mult(3.0, 2.0*np.ones((3,2))) compare_values += [csdl_tests.TestingPair(s5, t2, tag = 's5')] # add scalar variable and tensor constant s6 = csdl.mult(x, 2.0*np.ones((3,2))) compare_values += [csdl_tests.TestingPair(s6, t2, tag = 's6')] s6 = x*2.0*np.ones((3,2)) compare_values += [csdl_tests.TestingPair(s6, t2, tag = 's6')] s6 = 2.0*np.ones((3,2))*x compare_values += [csdl_tests.TestingPair(s6, t2, tag = 's6')] z_val = 2.0*np.ones((3,2)) z = csdl.Variable(name = 'z', value = z_val) # add scalar variable and tensor variable s7 = csdl.mult(x, z) compare_values += [csdl_tests.TestingPair(s7, t2, tag = 's7')] s7 = x*z compare_values += [csdl_tests.TestingPair(s7, t2, tag = 's7')] s7 = z*x compare_values += [csdl_tests.TestingPair(s7, t2, tag = 's7')] # add scalar constant and tensor variable s8 = csdl.mult(3.0, z) compare_values += [csdl_tests.TestingPair(s8, t2, tag = 's8')] s8 = z*3.0 compare_values += [csdl_tests.TestingPair(s8, t2, tag = 's8')] s8 = 3.0*z compare_values += [csdl_tests.TestingPair(s8, t2, tag = 's8')] v11 = csdl.Variable(value = np.array([[2.0]])) v1 = csdl.Variable(value = np.array([3.0])) s9 = v1*v11 compare_values += [csdl_tests.TestingPair(s9, np.array([[6.0]]), tag = 's9')] s9 = v11*v1 compare_values += [csdl_tests.TestingPair(s9, np.array([6.0]), tag = 's9')] self.run_tests(compare_values = compare_values, verify_derivatives=True) def test_docstring(self): self.docstest(mult) if __name__ == '__main__': test = TestMult() test.overwrite_backend = 'inline' test.test_functionality() test.test_docstring()