Daphne Lib - l1 norm calculation

[BacchusX1]

I would like to calculate the L1 norm in daphne-lib like that:

test_temp = np.array([0, 1, 2, 3, 4, 5])
support_temp = np.array([1, 2, 3, 4, 5, 6])

a = dc.from_numpy(test_temp)
b = dc.from_numpy(support_temp)

c = a - b

c_numpy = c.abs().sum().compute()

Output:

AttributeError: 'Matrix' object has no attribute 'abs'

For my understanding https://github.com/daphne-eu/daphne/blob/main/doc/DaphneLib/APIRef.md abs() should be available.

So I tried:

test_temp = np.array([0, 1, 2, 3, 4, 5])
support_temp = np.array([0, 1, 2, 3, 4, 5])

a = dc.from_numpy(test_temp)
b = dc.from_numpy(support_temp)

c = a - b

c = c**2
c_numpy = c.sqrt().sum().compute()

Output:

TypeError: unsupported operand type(s) for ** or pow(): 'Matrix' and 'int'

Any suggestions how to do that?

LG Bernhard

[pdamme]

Hi @BacchusX1, your scripts are valid. On the current main branch of DAPHNE they pass successfully, i.e., when doing print(c_numpy), the first script prints 6, the second one 0.0. Support for abs() on DaphneLib matrices was added in 2c1b7013dfea7e7ad839b40a93bbb10cb6edd9b2 in Sep 2023. Could it be that you use an outdated version of DAPHNE, e.g., the v0.2 release? If so, please try using the current state of the main branch, e.g., by following the quickstart guide for developers.

[BacchusX1]

Yes that was the issue