o .wi @szddlZddlmZddlmZddlmZdedededefd d Zd ededefd d ZdedededefddZ dS)N)Tensor)_check_same_shape)TorchMetricsUserErrorpredstargetspreturncCsNt||t|ttfr|dkstd|t||}tt||S)aUpdate and return variables required to compute Minkowski distance. Checks for same shape of input tensors. Args: preds: Predicted tensor targets: Ground truth tensor p: Non-negative number acting as the p to the errors z>Argument ``p`` must be a float or int greater than 1, but got ) r isinstancefloatintrtorchabssumpow)rrr differenceri/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/torchmetrics/functional/regression/minkowski.py_minkowski_distance_updates rdistancecCst|d|S)aCompute Minkowski Distance. Args: distance: Sum of the p-th powers of errors over all observations p: The non-negative numeric power the errors are to be raised to Example: >>> preds = torch.tensor([0., 1, 2, 3]) >>> target = torch.tensor([0., 2, 3, 1]) >>> distance_p_sum = _minkowski_distance_update(preds, target, 5) >>> _minkowski_distance_compute(distance_p_sum, 5) tensor(2.0244) g?)r r)rrrrr_minkowski_distance_compute)srcCst|||}t||S)a'Compute the `Minkowski distance`_. .. math:: d_{\text{Minkowski}} = \\sum_{i}^N (| y_i - \\hat{y_i} |^p)^\frac{1}{p} This metric can be seen as generalized version of the standard euclidean distance which corresponds to minkowski distance with p=2. Args: preds: estimated labels of type Tensor targets: ground truth labels of type Tensor p: int or float larger than 1, exponent to which the difference between preds and target is to be raised Return: Tensor with the Minkowski distance Example: >>> from torchmetrics.functional.regression import minkowski_distance >>> x = torch.tensor([1.0, 2.8, 3.5, 4.5]) >>> y = torch.tensor([6.1, 2.11, 3.1, 5.6]) >>> minkowski_distance(x, y, p=3) tensor(5.1220) )rr)rrrminkowski_dist_sumrrrminkowski_distance;s  r) r rtorchmetrics.utilities.checksr!torchmetrics.utilities.exceptionsrr rrrrrrrs