Vectors are write as boldface lowercase letters (e.g. $\mat a$, $\mat b$), matrices use boldface uppercase or Greek letters (e.g. $\mat A$, $\mat B$, $\mat\alpha$, $\mat\Delta$). The identity matrix of dimensions $p\times p$ is denoted by $\mat{I}_p$ and the commutation matrix as $\mat{K}_{p, q}$ or $\mat{K}_p$ is case of $p = q$. Tensors, meaning multi-dimensional arrays of order at least 3, use uppercase calligraphic letters (e.g. $\ten{A}$, $\ten{B}$, $\ten{X}$, $\ten{Y}$, $\ten{F}$). Boldface indices (e.g. $\mat{i}, \mat{j}, \mat{k}$) denote multi-indices $\mat{i}=(i_1, ..., i_r)\in[\mat{d}]$ where the bracket notation is a shorthand for $[r]=\{1, ..., r\}$ which in conjunction with a multi-index as argument means $[\mat{d}]=[d_1]\times ... \times[d_K]$.
Let $\ten{A}=(a_{i_1,...,i_r})\in\mathbb{R}^{d_1\times ...\times d_r}$ be an order\footnote{Also called rank, therefore the variable name $r$, but this term is \emph{not} used as it leads to confusion with the rank as in ``the rank of a matrix''.}$r$ tensor where $r\in\mathbb{N}$ is the number of modes or axis of $\ten{A}$. For matrices $\mat{B}_k\in\mathbb{R}^{p_k\times d_k}$ with $k\in[r]=\{1, 2, ..., r\}$ the \emph{multi-linear multiplication} is defined element wise as
which results in an order $r$ tensor of dimensions $p_1\times ...\times p_k)$. With this the \emph{$k$-mode product} between the tensor $\ten{A}$ with the matrix $\mat{B}_k$ is given by
Furthermore, the notation $\ten{A}\times_{k\in S}$ is a short hand for writing the iterative application if the mode product for all indices in $S\subset[r]$. For example $\ten{A}\times_{k\in\{2, 5\}}\mat{B}_k =\ten{A}\times_2\mat{B}_2\times_5\mat{B}_5$. By only allowing $S$ to be a set this notation is unambiguous because the mode products commutes for different modes $j\neq k\Rightarrow\ten{A}\times_j\mat{B}_j\times_k\mat{B}_k =\ten{A}\times_k\mat{B}_k\times_j\mat{B}_j$.
The \emph{inner product} between two tensors of the same order and dimensions is
with which the \emph{Frobenius Norm}$\|\ten{A}\|_F =\sqrt{\langle\ten{A}, \ten{A}\rangle}$. Of interest is also the \emph{maximum norm}$\|\ten{A}\|_{\infty}=\max_{i_1, ..., i_K} a_{i_1, ..., i_K}$. Furthermore, the Frobenius and maximum norm are also used for matrices while for a vector $\mat{a}$ the \emph{2 norm} is $\|\mat{a}\|_2=\sqrt{\langle\mat{a}, \mat{a}\rangle}$.
Matrices and tensor can be \emph{vectorized} by the \emph{vectorization} operator $\vec$. For tensors of order at least $2$ the \emph{flattening} (or \emph{unfolding} or \emph{matricization}) is a reshaping of the tensor into a matrix along an particular mode. For a tensor $\ten{A}$ of order $r$ and dimensions $d_1, ..., d_r$ the $k$-mode unfolding $\ten{A}_{(k)}$ is a $d_k\times\prod_{l=1, l\neq k}d_l$ matrix. For the tensor $\ten{A}=(a_{i_1,...,i_r})\in\mathbb{R}^{d_1, ..., d_r}$ the elements of the $k$ unfolded tensor $\ten{A}_{(k)}$ are
The rank of a tensor $\ten{A}$ of dimensions $d_1\times ...\times d_r$ is given by a vector $\rank{\ten{A}}=(a_1, ..., a_r)\in[d_1]\times...\times[d_r]$ where $a_k =\rank(\ten{A}_{(k)})$ is the usual matrix rank of the $k$ unfolded tensor.
{\color{red}$\mathcal{S}^p$, $\mathcal{S}_{+}^p$, $\mathcal{S}_{++}^p$ symmetric matrices of dimensions $p\times p$, or call it $\operatorname{Sym}(p)$}
{\color{red}The group of orthogonas matrices $O(p)$ of dim $p\times p$, where $O(p, q)$ are the $p\times q$ matrices (a.k.a. the Stiefel manifold)}
\begin{theorem}[Tensor Normal to Multi-Variate Normal equivalence]
For a multi-dimensional random variable $\ten{X}$ of order $r$ with dimensions $p_1\times ..., p_r$. Let $\ten{\mu}$ be the mean of the same order and dimensions as $\ten{X}$ and the mode covariance matrices $\mat{\Delta}_i$ of dimensions $p_i\times p_i$ for $i =1, ..., n$. Then the tensor normal distribution is equivalent to the multi-variate normal distribution by the relation
A straight forward way is to rewrite the Tensor Normal density as the density of a Multi-Variate Normal distribution depending on the vectorization of $\ten{X}$. First consider
Next, using a property of the determinant of a Kronecker product $|\mat{\Delta}_1\otimes\mat{\Delta}_2| = |\mat{\Delta}_1|^{p_2}|\mat{\Delta}_2|^{p_1}$ yields
When sampling from the Multi-Array Normal one way is to sample from the Multi-Variate Normal and then reshaping the result, but this is usually very inefficient because it requires to store the multi-variate covariance matrix which is very big. Instead, it is more efficient to sample $\ten{Z}$ as a tensor of the same shape as $\ten{X}$ with standard normal entries and then transform the $\ten{Z}$ to follow the Multi-Array Normal as follows
\caption{\label{tab:dimensions}\small Summary listing of dimensions with the corresponding sample versions $\mat{X}_i, \mat{R}_i, \mat{r}_i, \mat{f}_{y_i}$ for $i =1, ..., n$ as well as estimates $\widehat{\mat{\alpha}}, \widehat{\mat{\beta}}, \widehat{\mat\Delta}, \widehat{\mat\Delta}_1$ and $\widehat{\mat\Delta}_2$.}
\end{minipage}
\end{table}
The log-likelihood $l$ given $n$ i.i.d. observations assuming that $\mat{X}_i\mid(Y = y_i)$ is normal distributed as
Note that the log-likelihood estimate $\hat{l}$ only depends on $\mat\alpha, \mat\beta$. Next, we compute the gradient for $\mat\alpha$ and $\mat\beta$ of $\hat{l}$ used to formulate a Gradient Descent base estimation algorithm for $\mat\alpha, \mat\beta$ as the previous algorithmic. The main reason is to enable an estimation for bigger dimensions of the $\mat\alpha, \mat\beta$ coefficients since the previous algorithm does \emph{not} solve the high run time problem for bigger dimensions.
Start with the general case of $\mat X_i|(Y_i = y_i)$ is multivariate normal distributed with the covariance $\mat\Delta$ being a $p q\times p q$ positive definite symmetric matrix \emph{without} an further assumptions. We have $i =1, ..., n$ observations following
Substitution of the MLE estimates into the log-likelihood $l(\mat\mu, \mat\Delta, \mat\alpha, \mat\beta)$ gives the estimated log-likelihood $\hat{l}(\mat\alpha, \mat\beta)$ as
We are interested in the gradients $\nabla_{\mat\alpha}\hat{l}(\mat\alpha, \mat\beta)$, $\nabla_{\mat\beta}\hat{l}(\mat\alpha, \mat\beta)$ of the estimated log-likelihood. Therefore, we consider the differential of $\hat{l}$.
&= \underbrace{-\frac{n}{2}\log|\widehat{\mat{\Delta}}| - \frac{1}{2}\sum_{i = 1}^n \t{\mat{r}_i}(\d\widehat{\mat{\Delta}}^{-1}) \mat{r}_i}_{=\,0\text{ due to }\widehat{\mat{\Delta}}\text{ beeing the MLE}}\label{eq:deriv1}
Now, substitution of $\d\mat{r}_i$ into \eqref{eq:deriv1} gives the gradients (not dimension standardized versions of $\D\hat{l}(\mat\alpha)$, $\D\hat{l}(\mat\beta)$) by identification of the derivatives from the differentials (see: \todo{appendix})
These quantities are very verbose as well as completely unusable for an implementation. By detailed analysis of the gradients we see that the main parts are only element permutations with a high sparsity. By defining the following compact matrix
\mat G = \vec^{-1}_{q r}\bigg(\Big( \sum_{j = 1}^n \vec\mat{f}_{y_j}\otimes\widehat{\mat\Delta}^{-1}\mat{r}_j \Big)_{\pi(i)}\bigg)_{i = 1}^{p q k r}{\color{gray}\qquad(q r \times p k)}
with $\pi$ being a permutation of $p q k r$ elements corresponding to permuting the axis of a 4D tensor of dimensions $p\times q\times k\times r$ by $(2, 4, 1, 3)$. As a generalization of transposition this leads to a rearrangement of the elements corresponding to the permuted 4D tensor with dimensions $q\times r\times p\times k$ which is then vectorized and reshaped into a matrix of dimensions $q r \times p k$. With $\mat G$ the gradients simplify to \todo{validate this mathematically}
but the error tensor $\ten{\epsilon}\sim\mathcal{TN}(0, \mat{\Delta}_1, ..., \mat{\Delta}_r)$ is Tensor Normal distributed with mean zero and covariance matrices $\mat{\Delta}_1, ..., \mat{\Delta}_r$.
The sample model for $n$ observations has the same form with an additional sample axis in the last mode of $\ten{X}$ and $\ten{Y}$ with dimensions $p_1\times ...\times p_r\times n$ and $q_1\times ...\times q_r\times n$, respectively. Let the residual tensor be
By the definition of the Tensor Normal, using the notation $p_{\lnot j}=\prod_{k\neq j}p_j$, we get for observations $\ten{X}, \ten{F}$ the log-likelihood in terms of the residuals as
Note that the log-likelihood depends on the covariance matrices $\mat{\Delta}_j$, $j =1, ..., r$ as well as the mean $\mu$ and the parameter matrices $\mat{\alpha}_j$, $j =1, ..., r$ through the residuals $\ten{R}$.
\subsection{MLE estimates}
For deriving the MLE estimates we compute the differential of the log-likelihood given the data as
Using $\d\log|\mat{A}| =\tr(\mat{A}^{-1}\d\mat{A})$ and $\d\mat{A}^{-1}=-\mat{A}^{-1}(\d\mat{A})\mat{A}^{-1}$ as well as $\langle\ten{A}, \ten{B}\rangle=\tr(\ten{A}_{(j)}\t{\ten{B}_{(j)}})$ for any $j =1, ..., r$ we get the differential of the estimated log-likelihood as
as well as gradients (even though they are not realy used, except in the case of a pure gradient based estimation procedure which might ease the estimation burden as all the MLE estimates are cross dependent)
\qquad{\color{gray}p_j\times q_j}&& j = 1, ..., r.
\end{align*}
% Now we assume the residuals covariance has the form $\mat\Delta = \mat\Delta_1\otimes\mat\Delta_2$ where $\mat\Delta_1$, $\mat\Delta_2$ are $q\times q$, $p\times p$ covariance matrices, respectively. This is analog to the case that $\mat{R}_i$'s are i.i.d. Matrix Normal distribution
% The density of the Matrix Normal (with mean zero) is equivalent to the vectorized quantities being multivariate normal distributed with Kronecker structured covariance
% Out first order of business is to derive the MLE estimated of the covariance matrices $\mat\Delta_1$, $\mat\Delta_2$ (the mean estimate $\widehat{\mat\mu}$ is trivial). Therefore, we look at the differentials with respect to changes in the covariance matrices as
% We first take a closer look at the sum. After a bit of algebra using $\d\mat A^{-1} = -\mat A^{-1}(\d\mat A)\mat A^{-1}$ and the definitions of $\widehat{\mat\Delta}_1$, $\widehat{\mat\Delta}_2$ the sum can be rewritten
\paragraph{Comparison to the general case:} There are two main differences, first the general case has a closed form solution for the gradient due to the explicit nature of the MLE estimate of $\widehat{\mat\Delta}$ compared to the mutually dependent MLE estimates $\widehat{\mat\Delta}_1$, $\widehat{\mat\Delta}_2$. On the other hand the general case has dramatically bigger dimensions of the covariance matrix ($p q \times p q$) compared to the two Kronecker components with dimensions $q \times q$ and $p \times p$. This means that in the general case there is a huge performance penalty in the dimensions of $\widehat{\mat\Delta}$ while in the other case an extra estimation is required to determine $\widehat{\mat\Delta}_1$, $\widehat{\mat\Delta}_2$.
which means that $\E\widetilde{\mat\Delta}_1=\mat\Delta_1\tr(\mat\Delta_2)$ and in analogy $\E\widetilde{\mat\Delta}_2=\mat\Delta_2\tr(\mat\Delta_1)$. Now, we need to handle the scaling which can be estimated unbiasedly by
because with $\|\mat{R}_i\|_F^2=\tr\mat{R}_i\t{\mat{R}_i}=\tr\t{\mat{R}_i}\mat{R}_i$ the scale estimate $\tilde{s}=\tr(\widetilde{\mat\Delta}_1)=\tr(\widetilde{\mat\Delta}_2)$. Then $\E\tilde{s}=\tr(\E\widetilde{\mat\Delta}_1)=\tr{\mat\Delta}_1\tr{\mat\Delta}_2=\tr({\mat\Delta}_1\otimes{\mat\Delta}_2)$. Leading to the estimate of the covariance as
\todo{ prove they are consistent, especially $\widetilde{\mat\Delta}=\tilde{s}^{-1}(\widetilde{\mat\Delta}_1\otimes\widetilde{\mat\Delta}_2)$!}
The hoped for a benefit is that these covariance estimates are in a closed form which means there is no need for an additional iterative estimations step. Before we start with the derivation of the gradients define the following two quantities
Now, the matrix normal with the covariance matrix of the vectorized quantities of the form $\mat{\Delta}= s^{-1}(\mat{\Delta}_1\otimes\mat{\Delta}_2)$ has the form
The second form is due to the property of the determinant for scaling and the Kronecker product giving that $|\widetilde{\mat\Delta}| =(\tilde{s}^{-1})^{p q}|\widetilde{\mat{\Delta}}_1|^p |\widetilde{\mat{\Delta}}_2|^q$ as well as an analog Kronecker decomposition as in the MLE case.
The last one is tedious but straight forward. Its computation extensively uses the symmetry of $\widetilde{\mat{\Delta}}_1$, $\widetilde{\mat{\Delta}}_2$, the cyclic property of the trace and the relation $\d\mat{A}^{-1}=-\mat{A}^{-1}(\d\mat{A})\mat{A}^{-1}$.
Putting it all together
\begin{align*}
\d\tilde{l}(\mat{\alpha}, \mat{\beta})
&= \frac{n p q}{2}\Big(-\frac{2}{n\tilde{s}}\Big)\sum_{i = 1}^n \tr(\t{\mat{f}_{y_i}}\t{\mat{\beta}}\mat{R}_i\d\mat{\alpha} + \mat{f}_{y_i}\t{\mat{\alpha}}\t{\mat{R}_i}\d\mat{\beta}) \\
Observe that the bracketed expressions before $\d\mat{\alpha}$ and $\d\mat{\beta}$ are transposes. Lets denote the expression for $\d\mat{\alpha}$ as $\mat{G}_i$ which has the form
\todo{This section uses an alternative notation as it already tries to generalize to general multi-dimensional arrays. Furthermore, one of the main differences is that the observation are indexed in the \emph{last} mode. The benefit of this is that the mode product and parameter matrix indices match not only in the population model but also in sample versions.}
Let $\ten{X}, \ten{F}$ be order $r$ tensors of dimensions $p_1\times ... \times p_r$ and $q_1\times ... \times q_r$, respectively. Also denote the error tensor $\epsilon$ of the same order and dimensions as $\ten{X}$. The considered model for the $i$'th observation is
where we assume $\ten{\epsilon}_i$ to be i.i.d. mean zero tensor normal distributed $\ten{\epsilon}\sim\mathcal{TN}(0, \mat{\Delta}_1, ..., \mat{\Delta}_r)$ for $\mat{\Delta}_j\in\mathcal{S}^{p_j}_{++}$, $j =1, ..., r$. Given $i =1, ..., n$ observations the collected model containing all observations
which is almost identical as the observations $\ten{X}_i, \ten{F}_i$ are stacked on an addition $r +1$ mode leading to response, predictor and error tensors $\ten{X}, \ten{F}$ of order $r +1$ and dimensions $p_1\times...\times p_r\times n$ for $\ten{X}, \ten{\epsilon}$ and $q_1\times...\times q_r\times n$ for $\ten{F}$.
In the following we assume w.l.o.g that $\ten{\mu}=0$, as if this is not true we simply replace $\ten{X}_i$ with $\ten{X}_i -\ten{\mu}$ for $i =1, ..., n$ before collecting all the observations in the response tensor $\ten{X}$.
for the $j =1, ..., r$ modes. Using this relation we construct an iterative estimation process by setting the initial estimates of $\hat{\mat{\alpha}}_j^{(0)}=\mat{U}_j[, 1:q_j]$ which are the first $q_j$ columns of $\mat{U}_j$.
where the normal equations for $\mat{\alpha}_j$ depend on all the other $\mat{\alpha}_k$. With the initial estimates from above this allows an alternating approach. Index with $t =1, ...$ the current iteration, then a new estimate $\widehat{\mat{\alpha}}_j^{(t)}$ given the previous estimates $\widehat{\mat{\alpha}}_k^{(t-1)}$, $k =1, ..., r$ is computed as
for $j =1, ..., r$ until convergence or a maximum number of iterations is exceeded. The final estimates are the least squares estimates by this procedure.
% The first example (which by it self is \emph{not} exemplary) is the estimation with parameters $n = 200$, $p = 11$, $q = 5$, $k = 14$ and $r = 9$. The ``true'' matrices $\mat\alpha$, $\mat\beta$ generated by sampling there elements i.i.d. standard normal like the responses $y$. Then, for each observation, $\mat{f}_y$ is computed as $\sin(s_{i, j} y + o_{i j})$ \todo{ properly describe} to fill the elements of $\mat{f}_y$. Then the $\mat{X}$'s are samples as
\caption{Mean (standard deviation) for simulated runs of $20$ repititions for the model $\mat{X}=\mat{\beta}\mat{f}_y\t{\mat{\alpha}}$ of dimensinos $(p, q)=(11, 7)$, $(k, r)=(3, 5)$ with a sample size of $n =200$. The covariance structure is $\mat{\Delta}=\mat{\Delta}_2\otimes\mat{\Delta}_1$ for $\Delta_i =\text{AR}(\sqrt{0.5})$, $i =1, 2$. The functions applied to the standard normal response $y$ are $\sin, \cos$ with increasing frequency.}
In this section we summarize a few useful matrix identities, for more details see for example \cite{MatrixAlgebra-AbadirMagnus2005}.
For two matrices $\mat A$ of dimensions $q\times r$ and $\mat B$ of dimensions $p\times k$ holds
\begin{equation}\label{eq:vecKron}
\vec(\mat A\kron\mat B) = (\mat{I}_r\kron\mat{K}_{k,q}\kron\mat{I}_p)(\vec\mat A\kron\vec\mat B).
\end{equation}
Let $\mat A$ be a $p\times p$ dimensional non-singular matrix. Furthermore, let $\mat a, \mat b$ be $p$ vectors such that $\t{\mat b}A^{-1}\mat a\neq-1$, then
\det(\mat A + \mat a\t{\mat b}) = \det(\mat A)(1 + \t{\mat b}{\mat A}^{-1}\mat a)
\end{displaymath}
which even holds in the case $\t{\mat b}A^{-1}\mat a =-1$. This is known as Sylvester's determinant theorem.
\section{Commutation Matrix and Permutation Identities}
\begin{center}
Note: In this section we use 0-indexing for the sake of simplicity!
\end{center}
In this section we summarize relations between the commutation matrix and corresponding permutation. We also list some extensions to ``simplify'' or represent some term. This is mostly intended for implementation purposes and understanding of terms occurring in the computations.
Let $\mat A$ be an arbitrary $p\times q$ matrix. The permutation matrix $\mat K_{p, q}$ satisfies
\begin{displaymath}
\mat{K}_{p, q}\vec{\mat{A}} = \vec{\t{\mat{A}}}\quad\Leftrightarrow\quad (\vec{\mat{A}})_{\pi_{p, q}(i)} = (\vec{\t{\mat{A}}})_{i}, \quad\text{for } i = 0, ..., p q - 1
\end{displaymath}
where $\pi_{p, q}$ is a permutation of the indices $i =0, ..., p q -1$ such that
\begin{displaymath}
\pi_{p, q}(i + j p) = j + i q, \quad\text{for }i = 0, ..., p - 1; j = 0, ..., q - 1.
\end{displaymath}
\begin{table}[!htp]
\centering
\begin{minipage}{0.8\textwidth}
\centering
\begin{tabular}{l c l}
$\mat{K}_{p, q}$&$\hat{=}$&$\pi_{p, q}(i + j p)= j + i q$\\
$\mat{I}_r\kron\mat{K}_{p, q}$&$\hat{=}$&$\tilde{\pi}_{p, q, r}(i + j p + k p q)= j + i q + k p q$\\
$\mat{K}_{p, q}\kron\mat{I}_r$&$\hat{=}$&$\hat{\pi}_{p, q, r}(i + j p + k p q)= r(j + i q)+ k$
\end{tabular}
\caption{\label{tab:commutation-permutation}Commutation matrix terms and corresponding permutations. Indices are all 0-indexed with the ranges; $i =0, ..., p -1$, $j =0, ..., q -1$ and $k =0, ..., r -1$.}
\end{minipage}
\end{table}
\section{Matrix and Tensor Operations}
The \emph{Kronecker product}\index{Operations!Kronecker@$\kron$ Kronecker product} is denoted as $\kron$ and the \emph{Hadamard product} uses the symbol $\circ$. We also need the \emph{Khatri-Rao product}\index{Operations!KhatriRao@$\hada$ Khatri-Rao product}
$\hada$ as well as the \emph{Transposed Khatri-Rao product}$\odot_t$ (or \emph{Face-Splitting product}). There is also the \emph{$n$-mode Tensor Matrix Product}\index{Operations!ttm@$\ttm[n]$$n$-mode tensor product} denoted by $\ttm[n]$ in conjunction with the \emph{$n$-mode Matricization} of a Tensor $\mat{T}$ written as $\mat{T}_{(n)}$, which is a matrix. See below for definitions and examples of these operations.\todo{ Definitions and Examples}
\todo{ resolve confusion between Khatri-Rao, Column-wise Kronecker / Khatri-Rao, Row-wise Kronecker / Khatri-Rao, Face-Splitting Product, .... Yes, its a mess.}
\paragraph{Kronecker Product $\kron$:}
\paragraph{Khatri-Rao Product $\hada$:}
\paragraph{Transposed Khatri-Rao Product $\odot_t$:} This is also known as the Face-Splitting Product and is the row-wise Kronecker product of two matrices. If relates to the Column-wise Kronecker Product through
\paragraph{$n$-mode unfolding:}\emph{Unfolding}, also known as \emph{flattening} or \emph{matricization}, is an reshaping of a tensor into a matrix with rearrangement of the elements such that mode $n$ corresponds to columns of the result matrix and all other modes are vectorized in the rows. Let $\ten{T}$ be a tensor of order $m$ with dimensions $t_1\times ... \times t_n\times ... \times t_m$ and elements indexed by $(i_1, ..., i_n, ..., i_m)$. The $n$-mode flattening, denoted $\ten{T}_{(n)}$, is defined as a $(t_n, \prod_{k\neq n}t_k)$ matrix with element indices $(i_n, j)$ such that $j =\sum_{k =1, k\neq n}^m i_k\prod_{l =1, l\neq n}^{k -1}t_l$.
\todo{ give an example!}
\paragraph{$n$-mode Tensor Product $\ttm[n]$:}
The \emph{$n$-mode tensor product}$\ttm[n]$ between a tensor $\mat{T}$ of order $m$ with dimensions $t_1\times t_2\times ... \times t_n\times ... \times t_m$ and a $p\times t_n$ matrix $\mat{M}$ is defined element-wise as
where $i_1, ..., i_{n-1}, i_{n+1}, ..., i_m$ run from $1$ to $t_1, ..., t_{n-1}, t_{n+1}, ..., t_m$, respectively. Furthermore, the $n$-th fiber index $j$ of the product ranges from $1$ to $p$. This gives a new tensor $\mat{T}\ttm[n]\mat{M}$ of order $m$ with dimensions $t_1\times t_2\times ... \times p\times ... \times t_m$.
\begin{example}[Matrix Multiplication Analogs]
Let $\mat{A}$, $\mat{B}$ be two matrices with dimensions $t_1\times t_2$ and $p\times q$, respectively. Then $\mat{A}$ is also a tensor of order $2$, now the $1$-mode and $2$-mode products are element wise given by
In other words, the $1$-mode product equals $\mat{A}\ttm[1]\mat{B}=\mat{B}\mat{A}$ and the $2$-mode is $\mat{A}\ttm[2]\mat{B}=\t{(\mat{B}\t{\mat{A}})}$ in the case of the tensor $\mat{A}$ being a matrix.
\end{example}
\begin{example}[Order Three Analogs]
Let $\mat{A}$ be a tensor of the form $t_1\times t_2\times t_3$ and $\mat{B}$ a matrix of dimensions $p\times q$, then the $n$-mode products have the following look
\begin{align*}
(\mat{A}\ttm[1]\mat{B})_{i,j,k}&= \sum_{l = 1}^{t_1}\mat{A}_{l,j,k}\mat{B}_{i,l}&\text{for }t_1 = q, \\
(\mat{A}\ttm[2]\mat{B})_{i,j,k}&= \sum_{l = 1}^{t_2}\mat{A}_{i,l,k}\mat{B}_{j,l}\equiv (\mat{B}\mat{A}_{i,:,:})_{j,k}&\text{for }t_2 = q, \\
or in other words, the $i$-th slice of the tensor product $\ten{F}\ttm[2]\mat{\beta}\ttm[3]\mat{\alpha}$ contains $\mat{\beta}\mat{f}_{y_i}\t{\mat{\alpha}}$ for $i =1, ..., n$.
In this section we give a short summary of alternative but equivalent operations.
Using the notation $\widehat{=}$ to indicate that two expressions are identical in the sense that they contain the same element in the same order but may have different dimensions. Meaning, when vectorizing ether side of $\widehat{=}$, they are equal ($\mat{A}\widehat{=}\mat{B}\ :\Leftrightarrow\ \vec{\mat{A}}=\vec{\mat{B}}$).
Therefore, we use $\mat{A}, \mat{B}, \mat{X}, \mat{F}, \mat{R}, ...$ for matrices. 3-Tensors are written as $\ten{A}, \ten{B}, \ten{T}, \ten{X}, \ten{F}, \ten{R}, ...$.
% if and only if $\vec\mat{X}\sim\mathcal{N}_{p q}(\vec\mat\mu, \mat\Delta_1\otimes\mat\Delta_2)$. Note the order of the covariance matrices $\mat\Delta_1, \mat\Delta_2$. Its density is given by
% \begin{displaymath}
% f(\mat{X}) = \frac{1}{(2\pi)^{p q / 2}|\mat\Delta_1|^{p / 2}|\mat\Delta_2|^{q / 2}}\exp\left(-\frac{1}{2}\tr(\mat\Delta_1^{-1}\t{(\mat X - \mat \mu)}\mat\Delta_2^{-1}(\mat X - \mat \mu))\right).
% \end{displaymath}
% \section{Sampling form a Multi-Array Normal Distribution}
% Let $\ten{X}$ be an order (rank) $r$ Multi-Array random variable of dimensions $p_1\times...\times p_r$ following a Multi-Array (or Tensor) Normal distributed
This section contains short summaries of the main references with each sub-section concerning one paper.
\subsection{}
\subsection{Generalized Tensor Decomposition With Features on Multiple Modes}
The \cite{TensorDecomp-HuLeeWang2022} paper proposes a multi-linear conditional mean model for a constraint rank tensor decomposition. Let the responses $\ten{Y}\in\mathbb{R}^{d_1\times ... \times\d_K}$ be an order $K$ tensor. Associated with each mode $k\in[K]$ they assume feature matrices $\mat{X}_k\in\mathbb{R}^{d_k\times p_k}$. Now, they assume that conditional on the feature matrices $\mat{X}_k$ the entries of the tensor $\ten{Y}$ are independent realizations. The rank constraint is specified through $\mat{r}=(r_1, ..., r_K)$, then the model is given by
The order $K$ tensor $\ten{C}\in\mathbb{R}^{r_1\times...\times r_K}$ is an unknown full-rank core tensor and the matrices $\mat{M}_k\in\mathbb{R}^{p_k\times r_k}$ are unknown factor matrices. The function $f$ is applied element wise and serves as the link function based on the assumed distribution family of the tensor entries. Finally, the operation $\times$ denotes the tensor-by-matrix product using a short hand
with $\ttm[k]$ denoting the $k$-mode tensor matrix product.
The algorithm for estimation of $\ten{C}$ and $\mat{M}_1, ..., \mat{M}_K$ assumes the individual conditional entries of $\ten{Y}$ to be independent and to follow a generalized linear model with link function $f$. The proposed algorithm is an iterative algorithm for minimizing the negative log-likelihood
where $b = f'$ it the derivative of the canonical link function $f$ in the generalized linear model the conditioned entries of $\ten{Y}$ follow. The algorithm utilizes the higher-order SVD (HOSVD) to enforce the rank-constraint.
The main benefit is that this approach generalizes well to a multitude of different structured data sets.
\todo{ how does this relate to the $\mat{X}=\mat{\mu}+\mat{\beta}\mat{f}_y\t{\mat{\alpha}}+\mat{\epsilon}$ model.}
