对于具有高斯先验和高斯观测噪声的线性反问题，后验分布为高斯分布，其均值和协方差由条件公式确定。利用Feldman-Hajek定理，我们分析了无限维Hilbert参数空间和有限维观测下的先验到后验更新及其低秩近似。我们证明后验分布在有限维子空间上与先验分布不同，并在保持均值不变的情况下构建后验协方差的低秩近似。由于在无限维情况下，并非所有低秩协方差近似都使得近似的后验分布与先验和后验分布等价，我们刻画了能够实现这种等价性的低秩协方差近似及其对应的逆（即“精度”）。对于这些近似，通过识别同时对多种损失函数最优的低秩近似，解决了一类测度逼近问题。这些损失函数包括Rényi散度族、Amari $\alpha$-散度（$\alpha\in(0,1)$）、Hellinger度量和Kullback-Leibler散度。我们的结果将Spantini等人（SIAM J. Sci. Comput. 2015）的工作扩展到了Hilbert空间，并为构造离散化版本的无限维反问题的低秩近似提供了理论支持，通过提出独立于离散化的结果。

For linear inverse problems with Gaussian priors and Gaussian observation noise, the posterior is Gaussian, with mean and covariance determined by the conditioning formula. Using the Feldman-Hajek theorem, we analyse the prior-to-posterior update and its low-rank approximation for infinite-dimensional Hilbert parameter spaces and finite-dimensional observations. We show that the posterior distribution differs from the prior on a finite-dimensional subspace, and construct low-rank approximations to the posterior covariance, while keeping the mean fixed. Since in infinite dimensions, not all low-rank covariance approximations yield approximate posterior distributions which are equivalent to the posterior and prior distribution, we characterise the low-rank covariance approximations which do yield this equivalence, and their respective inverses, or `precisions'. For such approximations, a family of measure approximation problems is solved by identifying the low-rank approximations which are optimal for various losses simultaneously. These loss functions include the family of R\'enyi divergences, the Amari $\alpha$-divergences for $\alpha\in(0,1)$, the Hellinger metric and the Kullback-Leibler divergence. Our results extend those of Spantini et al. (SIAM J. Sci. Comput. 2015) to Hilbertian parameter spaces, and provide theoretical underpinning for the construction of low-rank approximations of discretised versions of the infinite-dimensional inverse problem, by formulating discretisation independent results.