Author: Eiko

Tags: matrix, matrix analysis, eigenvalues, singular values, p-adic

Time: 2024-11-29 17:41:45 - 2024-12-06 22:20:59 (UTC)

Reference: p-adic Differential Equations by Kiran S. Kedlaya

Matrix With Complex Numbers

For complex numbers we use the L2 norm for vectors and the operator norm |A|=supv0|Av||v| for matrices.

We always arrange the eigenvalues λ1,,λn of A in absolute value decreasing order.

Singular Values

It is familiar by inner product and orthogonalization that real symmetric matrices have real eigenvalues. Semi-positive definite matrices have non-negative eigenvalues, which also holds for complex matrices.

Definition. The singular values of A are the square roots of the eigenvalues of AA.

For a real symmetric matrix these are basically the absolute values of the eigenvalues.

Singular Value Decomposition

There are unitary U,VU(n) such that

UAV=Σ=Σ(σ1,,σn)

is a diagonal matrix, where σ1σ2σn0 are the singular values of A.

  • Singular value decomposition preserves metric on both sides. So it is a good measurement of the size of the matrix.

  • Changing notation ,we can write it as AV=UΣ, which basically means you can find orthogonal basis such that Avi=σiui.

  • For any vector v we have |Av|σ1|v|.

  • For any two dimensional subspace W there is a non-zero vector w such that |Aw|σ2|w|. This generalizes to higher dimensions.

  • σd is the smallest number such that for every d dimensional subspace you can find a non-zero vector v such that |Av|σd|v|. i.e.

    σd(A)=supdimW=dinfvW{0}|Av||v|.

  • Using this formula it is obvious that all singular value does not exceed operator norm, σi|A|. Actually σ1=|A|. So you can consider singular values as generalizations of operator norm! Instead of one number, you have a sequence of numbers describing the size of the matrix in different dimensions.

Remarks. Singular values are the invariants of a linear map between two inner product spaces (V,,V,f,W,,W), actually for matrices, here V,W are given basis and equipped with standard inner product. The structures of two inner products on source and target spaces are preserved.

Interaction With Exterior Algebra

For an inner product space V with orthonormal basis e1,,en, the exterior space kV has inner product structure and an orthonormal basis ei1eik for i1<<ik.

Since we can map

kA(vi1vik)=Avi1Avik=σi1σikui1uik,

  • we can see the singular values of kA are {σi1σik:i1<<ik}.

  • Similarly we can see the eigenvalues of kA are {λi1λik:i1<<ik}.

Singular Values Bounds Minors

  • In fact this observation gives another interpretation of singular values in terms of exterior product and orthonormal basis,

    (σ1σk)(A)=σ1(kA)=sup|o|=|o|=1|o,(kA)o|.

  • In particular this means all minors of A are bounded by the partial products of singular values, if |I|=|J|=k then

    |detAIJ|σ1σk.

  • From Ae1=λ1e1 and |λ1||e1|=|Ae1|σ1|e1| we see that σ1|λ1|. We have

    |λ1|σ1=|A|.

  • Applying above to kA we obtain Weyl’s inequality

    σ1σk|λ1λk|,

    which is an equality when k=0,n.

  • Weyl’s inequality has a converse, if you have {σi}R0 and {λi}C such that σ1σk|λ1λk| for all k then there is a matrix A such that σi are the singular values and λi are the eigenvalues.

Remarks. Unlike eigenvalues, singular values do not behave well under polynomial compositions, since the singular value decomposition essentially assumes the source and target spaces be different spaces. But they do work under inverses, from the singular value decomposition you can see the singular values of A1 are 1/σi.

Perturbations

  • The singular values behave better with additive perturbations,

    |σi(A+B)σi(A)||B|.

    This can be easily proved by the curious formula of σi in sup inf, and the two facts that |x+y||x|+|y|,|x+y||x||y|.

  • The eigenvalues cannot easily be controlled by additive perturbations, but we can say about the characteristic polynomials.

    |χA+B[tnm]χA[tnm]|(2m1)(nm)|B|j<mmax(σj,|B|).

    This can be seen as the χM[tni]=IdetMII term is the sum of all i-principal minors, we have

    χA+B[tnm]=|I|=mdet(A+B)II=|I|=mdetAII+|I|=mγs{αs,βs},sIdet(γi1γim)II=χA[tnm]+O1((2m1)(nm)supk<mσ1(A)σk(A)|B|mk)

  • For multiplicative perturbation, if B is invertible we have

    σi(BA)|B|σi(A).

    This can be seen using the formula

    σi(BA)=supinf|BAv||v||B|supinf|Av||v|=|B|σi(A).

  • A similar formula σi(BA)σi(A)|B| also holds, you can see this by applying transpose (also called dual or adjoint) and observe that singular values are invariant under transpose (dual) operation.

Matrix With p-adic Numbers

For p-adic fields (or actually, any other non-archimedean valued fields) the infinity norm on vectors is more appropriate

|v|:=maxi|vi|.

The operator norm is defined as usual using this norm.

  • As a corollary we have

    |Av|=maxi|aijvj|maxi(maxj|aij||vj|)maxi,j|aij|(maxj|vj|)maxi,j|aij||v|.

    This means |A|maxi,j|aij|.

  • Note also that |A|maxj|Aej||ej|=maxi,j|aij|, therefore

    |A|=max|aij|.

  • Matrix in PGLn(Op) preserves the vector norm, |Pv|=|v|. This can be seen by

    • |P|=max|pij|1,

    • |Pv||P||v||v|,

    • |v|=|P1Pv||P1||Pv||Pv|.

    • Equality holds also tells us that |P|=|P1|=1.

Hodge Polygon

For a sequence (si)i=1nRn its associated polygon is the path joined by {(n+k,s1++sk):k=0,,n}. The si are understood as successive slopes, and if si are non-decreasing, the polygon is convex.

Given a matrix A, let si be a sequence whose partial sum Si=s1++si satisfy

Si=min|I|=|J|=iv(detAIJ).

  • si are called elementary divisors or invariant factors of A.

  • The Hodge polygon of A is the associated polygon of this sequence, and the singular values of A are given by σi=psi, here p should be the base used in the absolute value ||=pv.

Singular Values Computes Maximal Minors

  • As a corollary of the definition,

    σ1σk=max|I|=|J|=k|detAIJ|.

  • These si are are clearly invariant under the action of GLn(Op)×GLn(Op)op, as a result the singular values and Hodges polygons are also invariant.

  • Since we are in a valuation ring, through the same process of putting matrix with PID coefficients into normal form, we have the smith normal form of A as PAQ=Σ(c1,,cn) the diagonal matrix, where P,QGLn(Op) and |ci|=σi=psi. This definition of singular value make sense because the vector norms on source and target spaces are preserved by GLn(Op).

  • The process guarantees that c1||cn which means s1sn, i.e. Hodge polygon is convex.

  • Similarly we have the subspace characterization of singular values

    σd(A)=supdimW=dinfvW{0}|Av||v|.

Remarks. Here GLn preserves vector norm, so it is natural choice for action. Unlike the complex case we do not try to preserve the inner product structure but we choose to preserve the vector norm. This is because the p-adic norm is not obtained from an inner product. The singular values are the invariants of a linear map between two normed spaces (V,||V,f,W,||W).

Newton Polygon Associated To Eigenvalues

For a matrix A with eigenvalues λ1,,λn the Newton polygon is the associated polygon of the sequence v(λ1)v(λn), so that |λ1||λn|.

  • As with eigenvalues, Newton polygon is invariant under conjugation by GLn(k).

  • Weyl’s inequality holds as well

    σ1σk|λ1λk|,

    with equality when k=0,n. This means the Newton polygon lies above the Hodge polygon, with the same starting and ending points.

Hodge-Newton Decomposition

If the Hodge and Newton polygons of a matrix ‘breaks into two parts’, i.e. for some i we have

|λi|>|λi+1|,σ1σi=|λ1λi|

the matrix has a Hodge-Newton decomposition at i, there is an integral matrix UGLn(Op) such that

U1AU=(BC0D)

If moreover σi>σi+1 then C can be chosen to be zero.

Perturbations Of p-adic Matrices

A lot of the perturbation results for complex matrices have a p-adic analog.

Non-Archimedean Property For Vectors

  • Recall that for numbers we have |a+b|max(|a|,|b|), with equality when |a||b|.

    There is a vector non-archimedean property (for the -norm) for vectors v,w, since |v+w|=maxi|vi+wi|maximax(|vi|,|wi|)=max(|v|,|w|), we have

    |v+w|max(|v|,|w|)

  • If |w|<|v|, there exists at least on i such that |wi|<|vi|, at this time the equality holds.

    |w|<|v||v+w|=|v|.

Perturbations

  • If |B|<σi, then σj(A+B)=σj(A) for all ji.

    This is the direct consequence of the vector non-archimedean property and the sup-inf property of singular values, as for the vector v that attains the sup-inf of σj(A), we have |Bv|<σi|v|=|Av| therefore |Av+Bv|=|Av|.

  • Let χA and χA+B be the characteristic polynomials. Then

    |χA+B[tnm]χA[tnm]||B|j<mmax(σj,|B|).

    Here the coefficients appearing in the complex version of the statement are replaced by 1, due to the archimedean property of p-adic numbers.

  • For multiplicative perturbation, the result is the same

    σi(BA)=supinf|BAv||v||B|supinf|Av||v|=|B|σi(A).