On manifolds of tensors of fixed TT-rank (Q2428543)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On manifolds of tensors of fixed TT-rank
scientific article

    Statements

    On manifolds of tensors of fixed TT-rank (English)
    0 references
    0 references
    0 references
    0 references
    26 April 2012
    0 references
    Recently, the format of TT tensors [of \textit{W. Hackbusch} and \textit{S. Kühn}, J. Fourier Anal. Appl. 15, No. 5, 706--722 (2009; Zbl 1188.15022); \textit{I. Oseledets}, ``Compact matrix form of the d-dimensional tensor decomposition'', SIAM J. Sci. Comput., submitted; \textit{I. V. Oseledets} and \textit{E. E. Tyrtyshnikov} in SIAM J. Sci. Comput. 31, No. 5, 3744--3759 (2009; Zbl 1200.65028); \textit{I. V. Oseledets} and \textit{E. E. Tyrtyshnikov}, ``Tensor tree decomposition does not need a tree'', Linear Algebra Appl., submitted] has turned out to be a promising new format for the approximation of solutions of high dimensional problems. In this paper, the authors prove some new results for the TT representation of a tensor \(U \in {\mathbb{R}}^{n_1 \times \dots \times n_d}\) and for the manifold of tensors of TT-rank \(\underline{r}\). At first, they prove that the TT (or compression) ranks \(r_{i}\) of a tensor \(U\) are unique and equal to the respective separation ranks of \(U\) if the components of the TT decomposition are required to fulfil a certain maximal rank condition. They then show that the set \(\mathbb{T}\) of TT tensors of fixed rank \(\underline{r}\) locally forms an embedded manifold in \({\mathbb{R}}^{n_1 \times \dots \times n_d}\), therefore preserving the essential theoretical properties of the Tucker format, but often showing an improved scaling behaviour. Extending a similar approach for matrices [of \textit{D. Conte} and \textit{C. Lubich}, ESAIM, Math. Model. Numer. Anal. 44, No. 4, 759--780 (2010; Zbl 1192.81125)], they introduce certain gauge conditions to obtain a unique representation of the tangent space \({\mathcal{T}_U\mathbb{T}}\) of \(\mathbb{T}\) and deduce a local parametrization of the TT manifold. The parametrisation of \({\mathcal{T}}_{U}\mathbb{T}\) is often crucial for an algorithmic treatment of high-dimensional time-dependent partial differential equations and minimisation problems [\textit{C. Lubich}, From quantum to classical molecular dynamics: Reduced models and numerical analysis. Zürich: European Mathematical Society (EMS) (2008; Zbl 1160.81001)]. They conclude with remarks on those applications and present some numerical examples.
    0 references
    0 references
    TT tensor
    0 references
    rank
    0 references
    linear manifold
    0 references
    Tucker format
    0 references
    numerical examples
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references