Polynomial evaluation by Horner rule?

1

Is there a polynomial evaluation function in PyTorch? Similar to https://www.tensorflow.org/api_docs/python/tf/math/polyval in Tensorflow, that automatically broadcasts over tensors?

2

You can have something like this

# POLYVAL in PYTORCH
coeffs=torch.FloatTensor([1.,2.,1.,1.])
x=torch.FloatTensor([3.0])

def getPolyVal(x,coeffs):
    curVal=0
    for curValIndex in range(len(coeffs)-1):
        curVal=(curVal+coeffs[curValIndex])*x[0]
    return(curVal+coeffs[len(coeffs)-1])

getPolyVal(x,coeffs)
1 Like
3

Thanks.

Any tricks I can do to make this as fast as possible?