o yi@sddlmZmZmZddlZddlmZddlmZmZddl m Z dededed ed ed ed eeeeeeeffd dZ Gddde Z dS))AnyListTupleN)Tensor)_pearson_corrcoef_compute_pearson_corrcoef_update)Metricmeans_xmeans_yvars_xvars_ycorrs_xynbsreturncCs<t|dkr|d|d|d|d|d|dfS|d|d|d|d|d|df\}}}} } } tdt|D]} || || || || || || f\} }}}}}| |}| ||| |}| ||||}| d|| |}|||||||d7}|d||| }||| ||||d7}||}| d|| |}| ||||||d7} |d|||}|||||||d7}| |}| ||||||||7} ||| ||||||7}| |}||||||f\}}}} } } q;||||||fS)zAggregate the statistics from multiple devices. Formula taken from here: `Aggregate the statistics from multiple devices`_ r)lenrange)r r r r r rmx1my1vx1vy1cxy1n1imx2my2vx2vy2cxy2n2nbmean_xmean_y element_x1 element_x2var_x element_y1 element_y2var_ycorr_xyr+S/home/ubuntu/.local/lib/python3.10/site-packages/torchmetrics/regression/pearson.py_final_aggregations. (44    $$r-cseZdZUdZdZdZdZeed<e e ed<e e ed<e ed<e ed<e ed <e ed <e ed <e ed < dde de ddffdd Z de de ddfddZde fddZZS)PearsonCorrCoefaComputes `Pearson Correlation Coefficient`_: .. math:: P_{corr}(x,y) = \frac{cov(x,y)}{\sigma_x \sigma_y} Where :math:`y` is a tensor of target values, and :math:`x` is a tensor of predictions. As input to ``forward`` and ``update`` the metric accepts the following input: - ``preds`` (:class:`~torch.Tensor`): either single output float tensor with shape ``(N,)`` or multioutput float tensor of shape ``(N,d)`` - ``target`` (:class:`~torch.Tensor`): either single output tensor with shape ``(N,)`` or multioutput tensor of shape ``(N,d)`` As output of ``forward`` and ``compute`` the metric returns the following output: - ``pearson`` (:class:`~torch.Tensor`): A tensor with the Pearson Correlation Coefficient Args: num_outputs: Number of outputs in multioutput setting kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info. Example (single output regression): >>> from torchmetrics import PearsonCorrCoef >>> target = torch.tensor([3, -0.5, 2, 7]) >>> preds = torch.tensor([2.5, 0.0, 2, 8]) >>> pearson = PearsonCorrCoef() >>> pearson(preds, target) tensor(0.9849) Example (multi output regression): >>> from torchmetrics import PearsonCorrCoef >>> target = torch.tensor([[3, -0.5], [2, 7]]) >>> preds = torch.tensor([[2.5, 0.0], [2, 8]]) >>> pearson = PearsonCorrCoef(num_outputs=2) >>> pearson(preds, target) tensor([1., 1.]) TNfull_state_updatepredstargetr"r#r&r)r*n_totalr num_outputskwargsrc stjd i|t|ts|dkrtd||_|jdt|jdd|jdt|jdd|jdt|jdd|jdt|jdd|jdt|jdd|jd t|jdddS) NrzQExpected argument `num_outputs` to be an int larger than 0, but got {num_outputs}r")defaultdist_reduce_fxr#r&r)r*r2r+) super__init__ isinstanceint ValueErrorr3 add_statetorchzeros)selfr3r4 __class__r+r,r8vszPearsonCorrCoef.__init__c CsBt|||j|j|j|j|j|j|j \|_|_|_|_|_|_dS)z*Update state with predictions and targets.N)rr"r#r&r)r*r2r3)r?r0r1r+r+r,updates zPearsonCorrCoef.updatecCs|jdkr |jdks|jdkr-|jjdkr-t|j|j|j|j|j|j \}}}}}}n |j}|j}|j}|j }t ||||S)z4Computes pearson correlation coefficient over state.r) r3r"numelndimr-r#r&r)r*r2r)r?_r&r)r*r2r+r+r,computes.zPearsonCorrCoef.compute)r)__name__ __module__ __qualname____doc__is_differentiablehigher_is_betterr/bool__annotations__rrr:rr8rBrF __classcell__r+r+r@r,r.Cs. &   r.) typingrrrr=r*torchmetrics.functional.regression.pearsonrrtorchmetrics.metricrr-r.r+r+r+r,s*    ,