Source code for pyhdc.components.similarity.cosine

from typing import Optional

import numpy as np

try:
    import torch

    TORCH_AVAILABLE = True
except ImportError:
    TORCH_AVAILABLE = False
    torch = None

from pyhdc.components.input_formatting import _normalize_similarity
from pyhdc.types import ArrayLike


[docs] def CosineSimilarity( *hypervectors: ArrayLike, axis: Optional[int] = None, mode: str = "pairwise" ): """CosineSimilarity Cosine Similarity of hypervectors cos(theta) = ( A dot B ) / ( norm(A) * norm(B) ) Hypervectors are dimension-first (axis 0 is always the dimension ``D``). Supports these calling conventions:: (a, b) where a and b are 1D: returns a scalar in [-1, 1] (a, b) batches: per-pair scores (trailing axes broadcast) (arr,) where arr is (D, N): sim(col_0, col_i) for i in 1..N-1 (arr,) where arr is (D, N, M, ...): requires ``axis`` (split index 0 vs the rest along that batch axis) With ``mode="cross"`` and two batches ``A=(D, P)``, ``B=(D, M)``, returns the full ``(P, M)`` cross-similarity matrix via a single matmul of the normalized operands (an all-zero column is treated as orthogonal, scoring 0 rather than nan). Args: *hypervectors: Two hypervectors, or a single batch array axis: For a single ``(D, N, M, ...)`` batch, the batch axis to split on mode: ``"pairwise"`` (default) or ``"cross"`` Returns: Scalar similarity, or an array of similarities over the trailing axes """ a, b, is_torch, scalar = _normalize_similarity(*hypervectors, axis=axis, mode=mode) if mode == "cross": if is_torch: assert torch is not None a_t = torch.as_tensor(a).float() b_t = torch.as_tensor(b).float() na = torch.linalg.norm(a_t, dim=0, keepdim=True) nb = torch.linalg.norm(b_t, dim=0, keepdim=True) na = torch.where(na == 0, torch.ones_like(na), na) nb = torch.where(nb == 0, torch.ones_like(nb), nb) return (a_t / na).T @ (b_t / nb) a_n = np.asarray(a, dtype=np.float32) b_n = np.asarray(b, dtype=np.float32) na = np.linalg.norm(a_n, axis=0, keepdims=True) nb = np.linalg.norm(b_n, axis=0, keepdims=True) na[na == 0] = 1.0 nb[nb == 0] = 1.0 return (a_n / na).T @ (b_n / nb) if is_torch: assert torch is not None a_t = torch.as_tensor(a).float() b_t = torch.as_tensor(b).float() dots = (a_t * b_t).sum(dim=0) norms = torch.linalg.norm(a_t, dim=0) * torch.linalg.norm(b_t, dim=0) sims = dots / norms return sims.item() if scalar else sims a_n = np.asarray(a, dtype=np.float32) b_n = np.asarray(b, dtype=np.float32) dots = np.sum(a_n * b_n, axis=0) norms = np.linalg.norm(a_n, axis=0) * np.linalg.norm(b_n, axis=0) sims = dots / norms return sims.item() if scalar else sims