How does numpy.float64 extract values from ufl functions?

General
1

Hi everyone,
I would have a curiosity about numpy.float64. Along my code I need to check the value of some constant quantities, i.e.:

if R > 0:
   c -= 16 * t**3 / (1 + 2 * t)**4

R can be either a “simple” number or let’s say a fem.Constant(), so I can’t use directly this condition or something like if R.values.item():. In order to overcome this problem I edit my code as follow:

from dolifinx import default_scalar_type as scalar


if scalar(R) > 0:
   c -= 16 * t**3 / (1 + 2 * t)**4

Now I known that default_scalar_type is just numpy.float64 and I’m curious to understand how numpy.float64 can extract the value of my fem.Constant(). Surprisingly I could see that this also works in other cases, for example I can pass as argument ufl.conditional(gt(1, 0), 2, 4) to numpy.float64() and then I get 2. But really I don’t understand how it’s possible, some of you could kindly give me some explanation?

thanks in advance :slight_smile:

2

The code you have presented here doesn’t really make sense.

is a statement that always evals to True.
Calling np.float64 on a ufl expression will cause ufl to try to convert all values in the ufl-expression to floats. In your case this will work, as there are no function or trial/test-functions in the condition, only scalar constants.

1 Like
3

Hi Jorgen, thank you for your prompt reply. Yes, you are right this condition doesn’t make sense, I made it up just to say that if I pass it as a numpy.float argument I will obtain the numerical value (in this case since it’s always True I get 2.0).
My question was about the relationship between numpy and ufl, maybe I misexplained myself when I asked the question. What I would like to understand is how does numpy.float64 extract values from ufl functions?

4

It calls float recursively on each operand in the ufl expression.

If the float conversion function is implemented for all operands, then you will be able to display it as a float

5

Okay, now I understand. I went to look at the source code and found what I was looking for. Specifically by inspecting the class Constant(), in fem module, I found the __float__() method. I guess it’s just through this method that values can be extracted.
https://docs.fenicsproject.org/dolfinx/v0.7.1/python/_modules/dolfinx/fem/function.html#Constant

I hope I got it right, thanks for your help :slight_smile: