more docs

Author: stevengj

src/dense.jl

@@ -1738,6 +1738,12 @@ algorithm instead uses pivoted QR factorization ([`qr`](@ref)).  To use an
 SVD-based algorithm, it is better to employ the SVD directly via `svd(M; rtol, atol) \\ b`,
 or to pass `rtol` and `atol` parameters via [`ldiv!`](@ref) with `svd(M)`.
 
+One can also pass `M = svd(A)` as the argument to `pinv` in order to re-use
+an existing [`SVD`](@ref) object.
+
+!!! compat "Julia 1.13"
+    Passing an `SVD` object to `pinv` requires Julia 1.13 or later.
+
 For more information, see [^pr1387], [^B96], [^S84], [^KY88].
 
 # Examples

src/svd.jl

@@ -288,6 +288,9 @@ Similar to the [`pinv`](@ref) function, the solution can be regularized by trunc
 dropping any singular values less than `max(atol, rtol*σ₁)` where `σ₁` is the largest singular value.
 The default relative tolerance is `n*ϵ`, where `n` is the size of the smallest dimension of `M`, and
 `ϵ` is the [`eps`](@ref) of the element type of `M`.
+
+!!! compat "Julia 1.13"
+    The `atol` and `rtol` arguments require Julia 1.13 or later.
 """
 function ldiv!(F::SVD{T}, B::AbstractVecOrMat; atol::Real=0, rtol::Real = (eps(real(float(oneunit(T))))*min(size(F)...))*iszero(atol)) where T
     m, n = size(F)