Home | History | Annotate | Download | only in docs
      1 %!TEX root = ceres-solver.tex
      2 \chapter{Solving}
      3 Effective use of Ceres requires some familiarity with the basic components of a nonlinear least squares solver, so before we describe how to configure the solver, we will begin by taking a brief look at how some of the core optimization algorithms in Ceres work and the various linear solvers and preconditioners that power it.
      4 
      5 \section{Trust Region Methods}
      6 \label{sec:trust-region}
      7 Let $x \in \mathbb{R}^{n}$ be an $n$-dimensional vector of variables, and
      8 $ F(x) = \left[f_1(x),   \hdots,  f_{m}(x) \right]^{\top}$ be a $m$-dimensional function of $x$.  We are interested in solving the following optimization problem~\footnote{At the level of the non-linear solver, the block and residual structure is not relevant, therefore our discussion here is in terms of an optimization problem defined over a state vector of size $n$.},
      9 \begin{equation}
     10         \arg \min_x \frac{1}{2}\|F(x)\|^2\ .
     11         \label{eq:nonlinsq}
     12 \end{equation}
     13 Here, the Jacobian $J(x)$ of $F(x)$ is an $m\times n$ matrix, where $J_{ij}(x) = \partial_j f_i(x)$  and the gradient vector $g(x) = \nabla  \frac{1}{2}\|F(x)\|^2 = J(x)^\top F(x)$. Since the efficient global optimization of~\eqref{eq:nonlinsq} for general $F(x)$ is an intractable problem, we will have to settle for finding a local minimum.
     14 
     15 The general strategy when solving non-linear optimization problems is to solve a sequence of approximations to the original problem~\cite{nocedal2000numerical}. At each iteration, the approximation is solved to determine a correction $\Delta x$ to the vector $x$. For non-linear least squares, an approximation can be constructed by using the linearization $F(x+\Delta x) \approx F(x) + J(x)\Delta x$, which leads to the following linear least squares  problem:
     16 \begin{equation}
     17          \min_{\Delta x} \frac{1}{2}\|J(x)\Delta x + F(x)\|^2
     18         \label{eq:linearapprox}
     19 \end{equation}
     20 Unfortunately, na\"ively solving a sequence of these problems and
     21 updating $x \leftarrow x+ \Delta x$ leads to an algorithm that may not
     22 converge.  To get a convergent algorithm, we need to control the size
     23 of the step $\Delta x$. And this is where the idea of a trust-region
     24 comes in. Algorithm~\ref{alg:trust-region} describes the basic  trust-region loop for non-linear least squares problems.
     25 
     26 \begin{algorithm}
     27 \caption{The basic trust-region algorithm.\label{alg:trust-region}}
     28 \begin{algorithmic}
     29 \REQUIRE Initial point $x$ and a trust region radius $\mu$.
     30 \LOOP
     31 \STATE{Solve $\arg \min_{\Delta x} \frac{1}{2}\|J(x)\Delta x + F(x)\|^2$ s.t. $\|D(x)\Delta x\|^2 \le \mu$}
     32 \STATE{$\rho = \frac{\displaystyle \|F(x + \Delta x)\|^2 - \|F(x)\|^2}{\displaystyle \|J(x)\Delta x + F(x)\|^2 - \|F(x)\|^2}$}
     33 \IF {$\rho > \epsilon$}
     34 \STATE{$x = x + \Delta x$}
     35 \ENDIF
     36 \IF {$\rho > \eta_1$}
     37 \STATE{$\rho = 2 * \rho$}
     38 \ELSE
     39 \IF {$\rho < \eta_2$}
     40 \STATE {$\rho = 0.5 * \rho$}
     41 \ENDIF
     42 \ENDIF
     43 \ENDLOOP
     44 \end{algorithmic}
     45 \end{algorithm}
     46 
     47 Here, $\mu$ is the trust region radius, $D(x)$ is some matrix used to define a metric on the domain of $F(x)$ and $\rho$ measures the quality of the step $\Delta x$, i.e., how well did the linear model predict the decrease in the value of the non-linear objective. The idea is to increase or decrease the radius of the trust region depending on how well the linearization predicts the behavior of the non-linear objective, which in turn is reflected in the value of $\rho$.
     48 
     49 The key computational step in a trust-region algorithm is the solution of the constrained optimization problem
     50 \begin{align}
     51         \arg\min_{\Delta x}& \frac{1}{2}\|J(x)\Delta x + F(x)\|^2 \\
     52         \text{such that}&\quad  \|D(x)\Delta x\|^2 \le \mu
     53 \label{eq:trp}
     54 \end{align}
     55 
     56 There are a number of different ways of solving this problem, each giving rise to a different concrete trust-region algorithm. Currently Ceres, implements two trust-region algorithms - Levenberg-Marquardt and  Dogleg.
     57 
     58 \subsection{Levenberg-Marquardt}
     59 The Levenberg-Marquardt algorithm~\cite{levenberg1944method, marquardt1963algorithm} is the most popular algorithm for solving non-linear least squares problems.  It was also the first trust region algorithm to be developed~\cite{levenberg1944method,marquardt1963algorithm}. Ceres implements an exact step~\cite{madsen2004methods} and an inexact step variant of the Levenberg-Marquardt algorithm~\cite{wright1985inexact,nash1990assessing}.
     60 
     61 It can be shown, that the solution to~\eqref{eq:trp} can be obtained by solving an unconstrained optimization of the form
     62 \begin{align}
     63         \arg\min_{\Delta x}& \frac{1}{2}\|J(x)\Delta x + F(x)\|^2 +\lambda  \|D(x)\Delta x\|^2
     64 \end{align}
     65 Where, $\lambda$ is a Lagrange multiplier that is inverse related to $\mu$. In Ceres, we solve for
     66 \begin{align}
     67         \arg\min_{\Delta x}& \frac{1}{2}\|J(x)\Delta x + F(x)\|^2 + \frac{1}{\mu} \|D(x)\Delta x\|^2
     68 \label{eq:lsqr}
     69 \end{align}
     70 The matrix $D(x)$ is a non-negative diagonal matrix, typically the square root of the diagonal of the matrix $J(x)^\top J(x)$.
     71 
     72 Before going further, let us make some notational simplifications. We will assume that the matrix $\sqrt{\mu} D$ has been concatenated at the bottom of the matrix $J$ and similarly a vector of zeros has been added to the bottom of the vector $f$ and the rest of our discussion will be in terms of $J$ and $f$, \ie the linear least squares problem.
     73 \begin{align}
     74  \min_{\Delta x} \frac{1}{2} \|J(x)\Delta x + f(x)\|^2 .
     75  \label{eq:simple}
     76 \end{align}
     77 For all but the smallest problems the solution of~\eqref{eq:simple} in each iteration of the Levenberg-Marquardt algorithm is the dominant computational cost in Ceres. Ceres provides a number of different options for solving~\eqref{eq:simple}. There are two major classes of methods - factorization and iterative.
     78 
     79 The factorization methods are based on computing an exact solution of~\eqref{eq:lsqr} using a Cholesky or a QR factorization and lead to an exact step Levenberg-Marquardt algorithm. But it is not clear if an exact solution of~\eqref{eq:lsqr} is necessary at each step of the LM algorithm to solve~\eqref{eq:nonlinsq}. In fact, we have already seen evidence that this may not be the case, as~\eqref{eq:lsqr} is itself a regularized version of~\eqref{eq:linearapprox}. Indeed, it is possible to construct non-linear optimization algorithms in which the linearized problem is solved approximately. These algorithms are known as inexact Newton or truncated Newton methods~\cite{nocedal2000numerical}.
     80 
     81 An inexact Newton method requires two ingredients. First, a cheap method for approximately solving systems of linear equations. Typically an iterative linear solver like the Conjugate Gradients method is used for this purpose~\cite{nocedal2000numerical}. Second, a termination rule for the iterative solver. A typical termination rule is of the form
     82 \begin{equation}
     83         \|H(x) \Delta x + g(x)\| \leq \eta_k \|g(x)\|. \label{eq:inexact}
     84 \end{equation}
     85 Here, $k$ indicates the Levenberg-Marquardt iteration number and $0 < \eta_k <1$ is known as the forcing sequence.  Wright \& Holt \cite{wright1985inexact} prove that a truncated Levenberg-Marquardt algorithm that uses an inexact Newton step based on~\eqref{eq:inexact} converges for any sequence $\eta_k \leq \eta_0 < 1$ and the rate of convergence depends on the choice of the forcing sequence $\eta_k$.
     86 
     87 Ceres supports both exact and inexact step solution strategies. When the user chooses a factorization based linear solver, the exact step Levenberg-Marquardt algorithm is used. When the user chooses an iterative linear solver, the inexact step Levenberg-Marquardt algorithm is used.
     88 
     89 \subsection{Dogleg}
     90 \label{sec:dogleg}
     91 Another strategy for solving the trust region problem~\eqref{eq:trp} was introduced by M. J. D. Powell. The key idea there is to compute two vectors
     92 \begin{align}
     93         \Delta x^{\text{Gauss-Newton}} &= \arg \min_{\Delta x}\frac{1}{2} \|J(x)\Delta x + f(x)\|^2.\\
     94         \Delta x^{\text{Cauchy}} &= -\frac{\|g(x)\|^2}{\|J(x)g(x)\|^2}g(x).
     95 \end{align}
     96 Note that the vector $\Delta x^{\text{Gauss-Newton}}$ is the solution
     97 to~\eqref{eq:linearapprox} and $\Delta x^{\text{Cauchy}}$ is the
     98 vector that minimizes the linear approximation if we restrict
     99 ourselves to moving along the direction of the gradient. Dogleg methods finds a vector $\Delta x$ defined by $\Delta
    100 x^{\text{Gauss-Newton}}$ and $\Delta x^{\text{Cauchy}}$ that solves
    101 the trust region problem. Ceres supports two
    102 variants.
    103 
    104 \texttt{TRADITIONAL\_DOGLEG} as described by Powell,
    105 constructs two line segments using the Gauss-Newton and Cauchy vectors
    106 and finds the point farthest along this line shaped like a dogleg
    107 (hence the name) that is contained in the
    108 trust-region. For more details on the exact reasoning and computations, please see Madsen et al~\cite{madsen2004methods}.
    109 
    110  \texttt{SUBSPACE\_DOGLEG} is a more sophisticated method
    111 that considers the entire two dimensional subspace spanned by these
    112 two vectors and finds the point that minimizes the trust region
    113 problem in this subspace\cite{byrd1988approximate}.
    114 
    115 The key advantage of the Dogleg over Levenberg Marquardt is that if the step computation for a particular choice of $\mu$ does not result in sufficient decrease in the value of the objective function, Levenberg-Marquardt solves the linear approximation from scratch with a smaller value of $\mu$. Dogleg on the other hand, only needs to compute the interpolation between the Gauss-Newton and the Cauchy vectors, as neither of them depend on the value of $\mu$.
    116 
    117 The Dogleg method can only be used with the exact factorization based linear solvers.
    118 
    119 \subsection{Inner Iterations}
    120 \label{sec:inner}
    121 Some non-linear least squares problems have additional structure in
    122 the way the parameter blocks interact that it is beneficial to modify
    123 the way the trust region step is computed. e.g., consider the
    124 following regression problem
    125 
    126 \begin{equation}
    127   y = a_1 e^{b_1 x} + a_2 e^{b_3 x^2 + c_1}
    128 \end{equation}
    129 
    130 Given a set of pairs $\{(x_i, y_i)\}$, the user wishes to estimate
    131 $a_1, a_2, b_1, b_2$, and $c_1$.
    132 
    133 Notice that the expression on the left is linear in $a_1$ and $a_2$,
    134 and given any value for $b_1$, $b_2$ and $c_1$, it is possible to use
    135 linear regression to estimate the optimal values of $a_1$ and
    136 $a_2$. It's possible to analytically eliminate the variables
    137 $a_1$ and $a_2$ from the problem entirely. Problems like these are
    138 known as separable least squares problem and the most famous algorithm
    139 for solving them is the Variable Projection algorithm invented by
    140 Golub \& Pereyra~\cite{golub-pereyra-73}.
    141 
    142 Similar structure can be found in the matrix factorization with
    143 missing data problem. There the corresponding algorithm is
    144 known as Wiberg's algorithm~\cite{wiberg}.
    145 
    146 Ruhe \& Wedin  present an analysis of
    147 various algorithms for solving separable non-linear least
    148 squares problems and refer to {\em Variable Projection} as
    149 Algorithm I in their paper~\cite{ruhe-wedin}.
    150 
    151 Implementing Variable Projection is tedious and expensive. Ruhe \&
    152 Wedin present a simpler algorithm with comparable convergence
    153 properties, which they call Algorithm II.  Algorithm II performs an
    154 additional optimization step to estimate $a_1$ and $a_2$ exactly after
    155 computing a successful Newton step.
    156 
    157 
    158 This idea can be generalized to cases where the residual is not
    159 linear in $a_1$ and $a_2$, i.e.,
    160 
    161 \begin{equation}
    162   y = f_1(a_1, e^{b_1 x}) + f_2(a_2, e^{b_3 x^2 + c_1})
    163 \end{equation}
    164 
    165 In this case, we solve for the trust region step for the full problem,
    166 and then use it as the starting point to further optimize just $a_1$
    167 and $a_2$. For the linear case, this amounts to doing a single linear
    168 least squares solve. For non-linear problems, any method for solving
    169 the $a_1$ and $a_2$ optimization problems will do. The only constraint
    170 on $a_1$ and $a_2$ (if they are two different parameter block) is that
    171 they do not co-occur in a residual block.
    172 
    173 This idea can be further generalized, by not just optimizing $(a_1,
    174 a_2)$, but decomposing the graph corresponding to the Hessian matrix's
    175 sparsity structure into a collection of non-overlapping independent sets
    176 and optimizing each of them.
    177 
    178 Setting \texttt{Solver::Options::use\_inner\_iterations} to true
    179 enables
    180 the use of this non-linear generalization of Ruhe \& Wedin's Algorithm
    181 II.  This version of Ceres has a higher iteration complexity, but also
    182 displays better convergence behavior per iteration.
    183 
    184 Setting \texttt{Solver::Options::num\_threads} to the maximum number
    185 possible is highly recommended.
    186 
    187 \subsection{Non-monotonic Steps}
    188 \label{sec:non-monotonic}
    189 Note that the basic trust-region algorithm described in
    190 Algorithm~\ref{alg:trust-region} is a descent algorithm  in that they
    191 only accepts a point if it strictly reduces the value of the objective
    192 function.
    193 
    194 Relaxing this requirement allows the algorithm to be more
    195 efficient in the long term at the cost of some local increase
    196 in the value of the objective function.
    197 
    198 This is because allowing for non-decreasing objective function
    199 values in a princpled manner allows the algorithm to ``jump over
    200 boulders'' as the method is not restricted to move into narrow
    201 valleys while preserving its convergence properties.
    202 
    203 Setting \texttt{Solver::Options::use\_nonmonotonic\_steps} to \texttt{true}
    204 enables the non-monotonic trust region algorithm as described by
    205 Conn,  Gould \& Toint in~\cite{conn2000trust}.
    206 
    207 Even though the value of the objective function may be larger
    208 than the minimum value encountered over the course of the
    209 optimization, the final parameters returned to the user are the
    210 ones corresponding to the minimum cost over all iterations.
    211 
    212 The option to take non-monotonic is available for all trust region
    213 strategies.
    214 
    215 \section{\texttt{LinearSolver}}
    216 Recall that in both of the trust-region methods described above, the key computational cost is the solution of a linear least squares problem of the form
    217 \begin{align}
    218  \min_{\Delta x} \frac{1}{2} \|J(x)\Delta x + f(x)\|^2 .
    219  \label{eq:simple2}
    220 \end{align}
    221 
    222 
    223 Let $H(x)= J(x)^\top J(x)$ and $g(x) = -J(x)^\top  f(x)$. For notational convenience let us also drop the dependence on $x$. Then it is easy to see that solving~\eqref{eq:simple2} is equivalent to solving the {\em normal equations}
    224 \begin{align}
    225 H \Delta x  &= g \label{eq:normal}
    226 \end{align}
    227 
    228 Ceres provides a number of different options for solving~\eqref{eq:normal}.
    229 
    230 \subsection{\texttt{DENSE\_QR}}
    231 For small problems (a couple of hundred parameters and a few thousand residuals) with relatively dense Jacobians, \texttt{DENSE\_QR} is the method of choice~\cite{bjorck1996numerical}. Let $J = QR$ be the QR-decomposition of $J$, where $Q$ is an orthonormal matrix and $R$ is an upper triangular matrix~\cite{trefethen1997numerical}. Then it can be shown that the solution to~\eqref{eq:normal} is given by
    232 \begin{align}
    233     \Delta x^* = -R^{-1}Q^\top f
    234 \end{align}
    235 Ceres uses \texttt{Eigen}'s dense QR factorization routines.
    236 
    237 \subsection{\texttt{DENSE\_NORMAL\_CHOLESKY} \& \texttt{SPARSE\_NORMAL\_CHOLESKY}}
    238 Large non-linear least square problems are usually sparse. In such cases, using a dense QR factorization is inefficient. Let $H = R^\top R$ be the Cholesky factorization of the normal equations, where $R$ is an upper triangular matrix, then the  solution to ~\eqref{eq:normal} is given by
    239 \begin{equation}
    240     \Delta x^* = R^{-1} R^{-\top} g.
    241 \end{equation}
    242 The observant reader will note that the $R$ in the Cholesky
    243 factorization of $H$ is the same upper triangular matrix $R$ in the QR
    244 factorization of $J$. Since $Q$ is an orthonormal matrix, $J=QR$
    245 implies that $J^\top J = R^\top Q^\top Q R = R^\top R$. There are two variants of Cholesky factorization -- sparse and
    246 dense.
    247 
    248 \texttt{DENSE\_NORMAL\_CHOLESKY}  as the name implies performs a dense
    249 Cholesky factorization of the normal equations. Ceres uses
    250 \texttt{Eigen}'s dense LDLT factorization routines.
    251 
    252 \texttt{SPARSE\_NORMAL\_CHOLESKY}, as the name implies performs a
    253 sparse Cholesky factorization of the normal equations. This leads to
    254 substantial savings in time and memory for large sparse
    255 problems. Ceres uses the sparse Cholesky factorization routines in  Professor Tim Davis'  \texttt{SuiteSparse} or
    256 \texttt{CXSparse} packages~\cite{chen2006acs}.
    257 
    258 \subsection{\texttt{DENSE\_SCHUR} \& \texttt{SPARSE\_SCHUR}}
    259 While it is possible to use \texttt{SPARSE\_NORMAL\_CHOLESKY} to solve bundle adjustment problems, bundle adjustment problem have a special structure, and a more efficient scheme for solving~\eqref{eq:normal} can be constructed.
    260 
    261 Suppose that the SfM problem consists of $p$ cameras and $q$ points and the variable vector $x$ has the  block structure $x = [y_{1},\hdots,y_{p},z_{1},\hdots,z_{q}]$. Where, $y$ and $z$ correspond to camera and point parameters, respectively.  Further, let the camera blocks be of size $c$ and the point blocks be of size $s$ (for most problems $c$ =  $6$--$9$ and $s = 3$). Ceres does not impose any constancy requirement on these block sizes, but choosing them to be constant simplifies the exposition.
    262 
    263 A key characteristic of the bundle adjustment problem is that there is no term $f_{i}$ that includes two or more point blocks.  This in turn implies that the matrix $H$ is of the form
    264 \begin{equation}
    265         H =  \left[
    266                 \begin{matrix} B & E\\ E^\top & C
    267                 \end{matrix}
    268                 \right]\ ,
    269 \label{eq:hblock}
    270 \end{equation}
    271 where, $B \in \reals^{pc\times pc}$ is a block sparse matrix with $p$ blocks of size $c\times c$ and  $C \in \reals^{qs\times qs}$ is a block diagonal matrix with $q$ blocks of size $s\times s$. $E \in \reals^{pc\times qs}$ is a general block sparse matrix, with a block of size $c\times s$ for each observation. Let us now block partition $\Delta x = [\Delta y,\Delta z]$ and $g=[v,w]$ to restate~\eqref{eq:normal} as the block structured linear system
    272 \begin{equation}
    273         \left[
    274                 \begin{matrix} B & E\\ E^\top & C
    275                 \end{matrix}
    276                 \right]\left[
    277                         \begin{matrix} \Delta y \\ \Delta z
    278                         \end{matrix}
    279                         \right]
    280                         =
    281                         \left[
    282                                 \begin{matrix} v\\ w
    283                                 \end{matrix}
    284                                 \right]\ ,
    285 \label{eq:linear2}
    286 \end{equation}
    287 and apply Gaussian elimination to it. As we noted above, $C$ is a block diagonal matrix, with small diagonal blocks of size $s\times s$.
    288 Thus, calculating the inverse of $C$ by inverting each of these blocks is  cheap. This allows us to  eliminate $\Delta z$ by observing that $\Delta z = C^{-1}(w - E^\top \Delta y)$, giving us
    289 \begin{equation}
    290         \left[B - EC^{-1}E^\top\right] \Delta y = v - EC^{-1}w\ .  \label{eq:schur}
    291 \end{equation}
    292 The matrix
    293 \begin{equation}
    294 S = B - EC^{-1}E^\top\ ,
    295 \end{equation}
    296 is the Schur complement of $C$ in $H$. It is also known as the {\em reduced camera matrix}, because the only variables participating in~\eqref{eq:schur} are the ones corresponding to the cameras. $S \in \reals^{pc\times pc}$ is a block structured symmetric positive definite matrix, with blocks of size $c\times c$. The block $S_{ij}$ corresponding to the pair of images $i$ and $j$ is non-zero if and only if the two images observe at least one common point.
    297 
    298 Now, \eqref{eq:linear2}~can  be solved by first forming $S$, solving for $\Delta y$, and then back-substituting $\Delta y$ to obtain the value of $\Delta z$.
    299 Thus, the solution of what was an $n\times n$, $n=pc+qs$ linear system is reduced to the inversion of the block diagonal matrix $C$, a few matrix-matrix and matrix-vector multiplies, and the solution of block sparse $pc\times pc$ linear system~\eqref{eq:schur}.  For almost all  problems, the number of cameras is much smaller than the number of points, $p \ll q$, thus solving~\eqref{eq:schur} is significantly cheaper than solving~\eqref{eq:linear2}. This is the {\em Schur complement trick}~\cite{brown-58}.
    300 
    301 This still leaves open the question of solving~\eqref{eq:schur}. The
    302 method of choice for solving symmetric positive definite systems
    303 exactly is via the Cholesky
    304 factorization~\cite{trefethen1997numerical} and depending upon the
    305 structure of the matrix, there are, in general, two options. The first
    306 is direct factorization, where we store and factor $S$ as a dense
    307 matrix~\cite{trefethen1997numerical}. This method has $O(p^2)$ space complexity and $O(p^3)$ time
    308 complexity and is only practical for problems with up to a few hundred
    309 cameras. Ceres implements this strategy as the \texttt{DENSE\_SCHUR} solver.
    310 
    311 
    312  But, $S$ is typically a fairly sparse matrix, as most images
    313 only see a small fraction of the scene. This leads us to the second
    314 option: sparse direct methods. These methods store $S$ as a sparse
    315 matrix, use row and column re-ordering algorithms to maximize the
    316 sparsity of the Cholesky decomposition, and focus their compute effort
    317 on the non-zero part of the factorization~\cite{chen2006acs}.
    318 Sparse direct methods, depending on the exact sparsity structure of the Schur complement,
    319 allow bundle adjustment algorithms to significantly scale up over those based on dense
    320 factorization. Ceres implements this strategy as the \texttt{SPARSE\_SCHUR} solver.
    321 
    322 \subsection{\texttt{CGNR}}
    323 For general sparse problems, if the problem is too large for \texttt{CHOLMOD} or a sparse linear algebra library is not linked into Ceres, another option is the \texttt{CGNR} solver. This solver uses the Conjugate Gradients solver on the {\em normal equations}, but without forming the normal equations explicitly. It exploits the relation
    324 \begin{align}
    325     H x = J^\top J x = J^\top(J x)
    326 \end{align}
    327 When the user chooses \texttt{ITERATIVE\_SCHUR} as the linear solver, Ceres automatically switches from the exact step algorithm to an inexact step algorithm.
    328 
    329 \subsection{\texttt{ITERATIVE\_SCHUR}}
    330 Another option for bundle adjustment problems is to apply PCG to the reduced camera matrix $S$ instead of $H$. One reason to do this is that $S$ is a much smaller matrix than $H$, but more importantly, it can be shown that $\kappa(S)\leq \kappa(H)$.  Ceres implements PCG on $S$ as the \texttt{ITERATIVE\_SCHUR} solver. When the user chooses \texttt{ITERATIVE\_SCHUR} as the linear solver, Ceres automatically switches from the exact step algorithm to an inexact step algorithm.
    331 
    332 The cost of forming and storing the Schur complement $S$ can be prohibitive for large problems. Indeed, for an inexact Newton solver that computes $S$ and runs PCG on it, almost all of its time is spent in constructing $S$; the time spent inside the PCG algorithm is negligible in comparison. Because  PCG only needs access to $S$ via its product with a vector, one way to evaluate $Sx$ is to observe that
    333 \begin{align}
    334   x_1 &= E^\top x \notag \\
    335   x_2 &= C^{-1} x_1 \notag\\
    336   x_3 &= Ex_2 \notag\\
    337   x_4 &= Bx \notag\\
    338   Sx &= x_4 - x_3\ .\label{eq:schurtrick1}
    339 \end{align}
    340 Thus, we can run PCG on $S$ with the same computational effort per iteration as PCG on $H$, while reaping the benefits of a more powerful preconditioner. In fact, we do not even need to compute $H$, \eqref{eq:schurtrick1} can be implemented using just the columns of $J$.
    341 
    342 Equation~\eqref{eq:schurtrick1} is closely related to {\em Domain Decomposition methods} for solving large linear systems that arise in structural engineering and partial differential equations. In the language of Domain Decomposition, each point in a bundle adjustment problem is a domain, and the cameras form the interface between these domains. The iterative solution of the Schur complement then falls within the sub-category of techniques known as Iterative Sub-structuring~\cite{saad2003iterative,mathew2008domain}.
    343 
    344 \section{Preconditioner}
    345 The convergence rate of Conjugate Gradients  for solving~\eqref{eq:normal} depends on the distribution of eigenvalues of $H$~\cite{saad2003iterative}. A useful upper bound is $\sqrt{\kappa(H)}$, where, $\kappa(H)$f is the condition number of the matrix $H$. For most bundle adjustment problems, $\kappa(H)$ is high and a direct application of Conjugate Gradients to~\eqref{eq:normal} results in extremely poor performance.
    346 
    347 The solution to this problem is to replace~\eqref{eq:normal} with a {\em preconditioned} system.  Given a linear system, $Ax =b$ and a preconditioner $M$ the preconditioned system is given by $M^{-1}Ax = M^{-1}b$. The resulting algorithm is known as Preconditioned Conjugate Gradients algorithm (PCG) and its  worst case complexity now depends on the condition number of the {\em preconditioned} matrix $\kappa(M^{-1}A)$.
    348 
    349 The computational cost of using a preconditioner $M$ is the cost of computing $M$ and evaluating the product $M^{-1}y$ for arbitrary vectors $y$. Thus, there are two competing factors to consider: How much of $H$'s structure is captured by $M$ so that the condition number $\kappa(HM^{-1})$ is low, and the computational cost of constructing and using $M$.  The ideal preconditioner would be one for which $\kappa(M^{-1}A) =1$. $M=A$ achieves this, but it is not a practical choice, as applying this preconditioner would require solving a linear system equivalent to the unpreconditioned problem.  It is usually the case that the more information $M$ has about $H$, the more expensive it is use. For example, Incomplete Cholesky factorization based preconditioners  have much better convergence behavior than the Jacobi preconditioner, but are also much more expensive.
    350 
    351 
    352 The simplest of all preconditioners is the diagonal or Jacobi preconditioner, \ie,  $M=\operatorname{diag}(A)$, which for block structured matrices like $H$ can be generalized to the block Jacobi preconditioner.
    353 
    354 For \texttt{ITERATIVE\_SCHUR} there are two obvious choices for block diagonal preconditioners for $S$. The block diagonal of the matrix $B$~\cite{mandel1990block} and the block diagonal $S$, \ie the block Jacobi preconditioner for $S$. Ceres's implements both of these preconditioners and refers to them as  \texttt{JACOBI} and \texttt{SCHUR\_JACOBI} respectively.
    355 
    356 For bundle adjustment problems arising in reconstruction from community photo collections, more effective preconditioners can be constructed by analyzing and exploiting the camera-point visibility structure of the scene~\cite{kushal2012}. Ceres implements the two visibility based preconditioners described by Kushal \& Agarwal as \texttt{CLUSTER\_JACOBI} and \texttt{CLUSTER\_TRIDIAGONAL}. These are fairly new preconditioners and Ceres' implementation of them is in its early stages and is not as mature as the other preconditioners described above.
    357 
    358 \section{Ordering}
    359 \label{sec:ordering}
    360 The order in which variables are eliminated in a linear solver can
    361 have a significant of impact on the efficiency and accuracy of the
    362 method. For example when doing sparse Cholesky factorization, there are
    363 matrices for which a good ordering will give a Cholesky factor with
    364 O(n) storage, where as a bad ordering will result in an completely
    365 dense factor.
    366 
    367 Ceres allows the user to provide varying amounts of hints to the
    368 solver about the variable elimination ordering to use. This can range
    369 from no hints, where the solver is free to decide the best ordering
    370 based on the user's choices like the linear solver being used, to an
    371 exact order in which the variables should be eliminated, and a variety
    372 of possibilities in between.
    373 
    374 Instances of the \texttt{ParameterBlockOrdering} class are used to communicate this
    375 information to Ceres.
    376 
    377 Formally an ordering is an ordered partitioning of the parameter
    378 blocks. Each parameter block belongs to exactly one group, and
    379 each group has a unique integer associated with it, that determines
    380 its order in the set of groups. We call these groups {\em elimination
    381   groups}.
    382 
    383 Given such an ordering, Ceres ensures that the parameter blocks in the
    384 lowest numbered elimination group are eliminated first, and then the
    385 parameter blocks in the next lowest numbered elimination group and so
    386 on. Within each elimination group, Ceres is free to order the
    387 parameter blocks as it chooses. e.g. Consider the linear system
    388 
    389 \begin{align}
    390 x + y &= 3\\
    391    2x + 3y &= 7
    392 \end{align}
    393 
    394 There are two ways in which it can be solved. First eliminating $x$
    395 from the two equations, solving for y and then back substituting
    396 for $x$, or first eliminating $y$, solving for $x$ and back substituting
    397 for $y$. The user can construct three orderings here.
    398 
    399 \begin{enumerate}
    400 \item   $\{0: x\}, \{1: y\}$: Eliminate $x$ first.
    401 \item  $\{0: y\}, \{1: x\}$: Eliminate $y$ first.
    402 \item   $\{0: x, y\}$: Solver gets to decide the elimination order.
    403 \end{enumerate}
    404 
    405 Thus, to have Ceres determine the ordering automatically using
    406 heuristics, put all the variables in the same elimination group. The
    407 identity of the group does not matter. This is the same as not
    408 specifying an ordering at all. To control the ordering for every
    409 variable, create an elimination group per variable, ordering them in
    410 the desired order.
    411 
    412 If the user is using one of the Schur solvers (\texttt{DENSE\_SCHUR},
    413 \texttt{SPARSE\_SCHUR},\ \texttt{ITERATIVE\_SCHUR}) and chooses to
    414 specify an ordering, it must have one important property. The lowest
    415 numbered elimination group must form an independent set in the graph
    416 corresponding to the Hessian, or in other words, no two parameter
    417 blocks in in the first elimination group should co-occur in the same
    418 residual block. For the best performance, this elimination group
    419 should be as large as possible. For standard bundle adjustment
    420 problems, this corresponds to the first elimination group containing
    421 all the 3d points, and the second containing the all the cameras
    422 parameter blocks.
    423 
    424 If the user leaves the choice to Ceres, then the solver uses an
    425 approximate maximum independent set algorithm to identify the first
    426 elimination group~\cite{li2007miqr} .
    427 \section{\texttt{Solver::Options}}
    428 
    429 \texttt{Solver::Options} controls the overall behavior of the
    430 solver. We list the various settings and their default values below.
    431 
    432 \begin{enumerate}
    433 
    434 \item{\texttt{trust\_region\_strategy\_type }}
    435   (\texttt{LEVENBERG\_MARQUARDT}) The  trust region step computation
    436   algorithm used by Ceres. Currently \texttt{LEVENBERG\_MARQUARDT }
    437   and \texttt{DOGLEG} are the two valid choices.
    438 
    439 \item{\texttt{dogleg\_type}} (\texttt{TRADITIONAL\_DOGLEG})  Ceres
    440   supports two different dogleg strategies.
    441   \texttt{TRADITIONAL\_DOGLEG} method by Powell and the
    442   \texttt{SUBSPACE\_DOGLEG} method described by Byrd et al.
    443 ~\cite{byrd1988approximate}. See Section~\ref{sec:dogleg} for more details.
    444 
    445 \item{\texttt{use\_nonmonotoic\_steps}} (\texttt{false})
    446 Relax the requirement that the trust-region algorithm take strictly
    447 decreasing steps. See Section~\ref{sec:non-monotonic} for more details.
    448 
    449 \item{\texttt{max\_consecutive\_nonmonotonic\_steps}} (5)
    450 The window size used by the step selection algorithm to accept
    451 non-monotonic steps.
    452 
    453 \item{\texttt{max\_num\_iterations }}(\texttt{50}) Maximum number of
    454   iterations for Levenberg-Marquardt.
    455 
    456 \item{\texttt{max\_solver\_time\_in\_seconds }} ($10^9$) Maximum
    457   amount of time for which the solver should run.
    458 
    459 \item{\texttt{num\_threads }}(\texttt{1}) Number of threads used by
    460   Ceres to evaluate the Jacobian.
    461 
    462 \item{\texttt{initial\_trust\_region\_radius } ($10^4$)} The size of
    463   the initial trust region. When the \texttt{LEVENBERG\_MARQUARDT}
    464   strategy is used, the reciprocal of this number is the initial
    465   regularization parameter.
    466 
    467 \item{\texttt{max\_trust\_region\_radius } ($10^{16}$)} The trust
    468   region radius is not allowed to grow beyond this value.
    469 
    470 \item{\texttt{min\_trust\_region\_radius } ($10^{-32}$)} The solver
    471   terminates, when the trust region becomes smaller than this value.
    472 
    473 \item{\texttt{min\_relative\_decrease }}($10^{-3}$) Lower threshold
    474   for relative decrease before a Levenberg-Marquardt step is acceped.
    475 
    476 \item{\texttt{lm\_min\_diagonal } ($10^6$)} The
    477   \texttt{LEVENBERG\_MARQUARDT} strategy, uses a diagonal matrix to
    478   regularize the the trust region step. This is the lower bound on the
    479   values of this diagonal matrix.
    480 
    481 \item{\texttt{lm\_max\_diagonal } ($10^{32}$)}  The
    482   \texttt{LEVENBERG\_MARQUARDT} strategy, uses a diagonal matrix to
    483   regularize the the trust region step. This is the upper bound on the
    484   values of this diagonal matrix.
    485 
    486 \item{\texttt{max\_num\_consecutive\_invalid\_steps } (5)} The step
    487   returned by a trust region strategy can sometimes be numerically
    488   invalid, usually because of conditioning issues. Instead of crashing
    489   or stopping the optimization, the optimizer can go ahead and try
    490   solving with a smaller trust region/better conditioned problem. This
    491   parameter sets the number of consecutive retries before the
    492   minimizer gives up.
    493 
    494 \item{\texttt{function\_tolerance }}($10^{-6}$) Solver terminates if
    495 \begin{align}
    496 \frac{|\Delta \text{cost}|}{\text{cost}} < \texttt{function\_tolerance}
    497 \end{align}
    498 where, $\Delta \text{cost}$ is the change in objective function value
    499 (up or down) in the current iteration of Levenberg-Marquardt.
    500 
    501 \item \texttt{Solver::Options::gradient\_tolerance } Solver terminates if
    502 \begin{equation}
    503     \frac{\|g(x)\|_\infty}{\|g(x_0)\|_\infty} < \texttt{gradient\_tolerance}
    504 \end{equation}
    505 where $\|\cdot\|_\infty$ refers to the max norm, and $x_0$ is the vector of initial parameter values.
    506 
    507 \item{\texttt{parameter\_tolerance }}($10^{-8}$) Solver terminates if
    508 \begin{equation}
    509     \frac{\|\Delta x\|}{\|x\| + \texttt{parameter\_tolerance}} < \texttt{parameter\_tolerance}
    510 \end{equation}
    511 where $\Delta x$ is the step computed by the linear solver in the current iteration of Levenberg-Marquardt.
    512 
    513 \item{\texttt{linear\_solver\_type }(\texttt{SPARSE\_NORMAL\_CHOLESKY})}
    514 
    515 \item{\texttt{linear\_solver\_type
    516     }}(\texttt{SPARSE\_NORMAL\_CHOLESKY}/\texttt{DENSE\_QR}) Type of
    517   linear solver used to compute the solution to the linear least
    518   squares problem in each iteration of the Levenberg-Marquardt
    519   algorithm. If Ceres is build with \suitesparse linked in  then the
    520   default is \texttt{SPARSE\_NORMAL\_CHOLESKY}, it is
    521   \texttt{DENSE\_QR} otherwise.
    522 
    523 \item{\texttt{preconditioner\_type }}(\texttt{JACOBI}) The
    524   preconditioner used by the iterative linear solver. The default is
    525   the block Jacobi preconditioner. Valid values are (in increasing
    526   order of complexity) \texttt{IDENTITY},\texttt{JACOBI},
    527   \texttt{SCHUR\_JACOBI}, \texttt{CLUSTER\_JACOBI} and
    528   \texttt{CLUSTER\_TRIDIAGONAL}.
    529 
    530 \item{\texttt{sparse\_linear\_algebra\_library }
    531     (\texttt{SUITE\_SPARSE})} Ceres supports the use of two sparse
    532   linear algebra libraries, \texttt{SuiteSparse}, which is enabled by
    533   setting this parameter to \texttt{SUITE\_SPARSE} and
    534   \texttt{CXSparse}, which can be selected by setting this parameter
    535   to $\texttt{CX\_SPARSE}$. \texttt{SuiteSparse} is a sophisticated
    536   and complex sparse linear algebra library and should be used in
    537   general. If your needs/platforms prevent you from using
    538   \texttt{SuiteSparse}, consider using \texttt{CXSparse}, which is a
    539   much smaller, easier to build library. As can be expected, its
    540   performance on large problems is not comparable to that of
    541   \texttt{SuiteSparse}.
    542 
    543 
    544 \item{\texttt{num\_linear\_solver\_threads }}(\texttt{1}) Number of
    545   threads used by the linear solver.
    546 
    547 \item{\texttt{use\_inner\_iterations} (\texttt{false}) } Use a
    548   non-linear version of a simplified variable projection
    549   algorithm. Essentially this amounts to doing a further optimization
    550   on each Newton/Trust region step using a coordinate descent
    551   algorithm.  For more details, see the discussion in ~\ref{sec:inner}
    552 
    553 \item{\texttt{inner\_iteration\_ordering} (\texttt{NULL})} If
    554   \texttt{Solver::Options::inner\_iterations} is true, then the user
    555   has two choices.
    556 
    557 \begin{enumerate}
    558 \item Let the solver heuristically decide which parameter blocks to
    559   optimize in each inner iteration. To do this, set
    560   \texttt{inner\_iteration\_ordering} to {\texttt{NULL}}.
    561 
    562 \item Specify a collection of of ordered independent sets. The lower
    563   numbered groups are optimized before the higher number groups during
    564   the inner optimization phase. Each group must be an independent set.
    565 \end{enumerate}
    566 
    567 \item{\texttt{linear\_solver\_ordering} (\texttt{NULL})} An instance
    568   of the ordering object informs the solver about the desired order in
    569   which parameter blocks should be eliminated by the linear
    570   solvers. See section~\ref{sec:ordering} for more details.
    571 
    572   If \texttt{NULL}, the solver is free to choose an ordering that it
    573   thinks is best. Note: currently, this option only has an effect on
    574   the Schur type solvers, support for the
    575   \texttt{SPARSE\_NORMAL\_CHOLESKY} solver is forth coming.
    576 
    577 \item{\texttt{use\_block\_amd } (\texttt{true})} By virtue of the
    578   modeling layer in Ceres being block oriented, all the matrices used
    579   by Ceres are also block oriented.  When doing sparse direct
    580   factorization of these matrices, the fill-reducing ordering
    581   algorithms can either be run on the block or the scalar form of
    582   these matrices. Running it on the block form exposes more of the
    583   super-nodal structure of the matrix to the Cholesky factorization
    584   routines. This leads to substantial gains in factorization
    585   performance. Setting this parameter to true, enables the use of a
    586   block oriented Approximate Minimum Degree ordering
    587   algorithm. Settings it to \texttt{false}, uses a scalar AMD
    588   algorithm. This option only makes sense when using
    589   \texttt{sparse\_linear\_algebra\_library = SUITE\_SPARSE} as it uses
    590   the \texttt{AMD} package that is part of \texttt{SuiteSparse}.
    591 
    592 \item{\texttt{linear\_solver\_min\_num\_iterations }}(\texttt{1})
    593   Minimum number of iterations used by the linear solver. This only
    594   makes sense when the linear solver is an iterative solver, e.g.,
    595   \texttt{ITERATIVE\_SCHUR}.
    596 
    597 \item{\texttt{linear\_solver\_max\_num\_iterations }}(\texttt{500})
    598   Minimum number of iterations used by the linear solver. This only
    599   makes sense when the linear solver is an iterative solver, e.g.,
    600   \texttt{ITERATIVE\_SCHUR}.
    601 
    602 \item{\texttt{eta }} ($10^{-1}$)
    603  Forcing sequence parameter. The truncated Newton solver uses this
    604  number to control the relative accuracy with which the Newton step is
    605  computed. This constant is passed to ConjugateGradientsSolver which
    606  uses it to terminate the iterations when
    607 \begin{equation}
    608 \frac{Q_i - Q_{i-1}}{Q_i} < \frac{\eta}{i}
    609 \end{equation}
    610 
    611 \item{\texttt{jacobi\_scaling }}(\texttt{true}) \texttt{true} means
    612   that the Jacobian is scaled by the norm of its columns before being
    613   passed to the linear solver. This improves the numerical
    614   conditioning of the normal equations.
    615 
    616 \item{\texttt{logging\_type }}(\texttt{PER\_MINIMIZER\_ITERATION})
    617 
    618 
    619 \item{\texttt{minimizer\_progress\_to\_stdout }}(\texttt{false})
    620 By default the Minimizer progress is logged to \texttt{STDERR}
    621 depending on the \texttt{vlog} level. If this flag is
    622 set to true, and \texttt{logging\_type } is not \texttt{SILENT}, the
    623 logging output
    624 is sent to \texttt{STDOUT}.
    625 
    626 \item{\texttt{return\_initial\_residuals }}(\texttt{false})
    627 \item{\texttt{return\_final\_residuals }}(\texttt{false})
    628 If true, the vectors \texttt{Solver::Summary::initial\_residuals } and
    629 \texttt{Solver::Summary::final\_residuals } are filled with the
    630 residuals before and after the optimization. The entries of these
    631 vectors are in the order in which ResidualBlocks were added to the
    632 Problem object.
    633 
    634 \item{\texttt{return\_initial\_gradient }}(\texttt{false})
    635 \item{\texttt{return\_final\_gradient }}(\texttt{false})
    636 If true, the vectors \texttt{Solver::Summary::initial\_gradient } and
    637 \texttt{Solver::Summary::final\_gradient } are filled with the
    638 gradient before and after the optimization. The entries of these
    639 vectors are in the order in which ParameterBlocks were added to the
    640 Problem object.
    641 
    642 Since \texttt{AddResidualBlock } adds ParameterBlocks to the
    643 \texttt{Problem } automatically if they do not already exist, if you
    644 wish to have explicit control over the ordering of the vectors, then
    645 use \texttt{Problem::AddParameterBlock } to explicitly add the
    646 ParameterBlocks in the order desired.
    647 
    648 \item{\texttt{return\_initial\_jacobian }}(\texttt{false})
    649 \item{\texttt{return\_initial\_jacobian }}(\texttt{false})
    650 If true, the Jacobian matrices before and after the optimization are
    651 returned in \texttt{Solver::Summary::initial\_jacobian } and
    652 \texttt{Solver::Summary::final\_jacobian } respectively.
    653 
    654 The rows of these matrices are in the same order in which the
    655 ResidualBlocks were added to the Problem object. The columns are in
    656 the same order in which the ParameterBlocks were added to the Problem
    657 object.
    658 
    659 Since \texttt{AddResidualBlock } adds ParameterBlocks to the
    660 \texttt{Problem } automatically if they do not already exist, if you
    661 wish to have explicit control over the column ordering of the matrix,
    662 then use \texttt{Problem::AddParameterBlock } to explicitly add the
    663 ParameterBlocks in the order desired.
    664 
    665 The Jacobian matrices are stored as compressed row sparse
    666 matrices. Please see \texttt{ceres/crs\_matrix.h } for more details of
    667 the format.
    668 
    669 \item{\texttt{lsqp\_iterations\_to\_dump }} List of iterations at
    670   which the optimizer should dump the linear least squares problem to
    671   disk. Useful for testing and benchmarking. If empty (default), no
    672   problems are dumped.
    673 
    674 \item{\texttt{lsqp\_dump\_directory }} (\texttt{/tmp})
    675  If \texttt{lsqp\_iterations\_to\_dump} is non-empty, then this
    676  setting determines the directory to which the files containing the
    677  linear least squares problems are written to.
    678 
    679 
    680 \item{\texttt{lsqp\_dump\_format }}(\texttt{TEXTFILE}) The format in
    681   which linear least squares problems should be logged
    682 when \texttt{lsqp\_iterations\_to\_dump} is non-empty.  There are three options
    683 \begin{itemize}
    684 \item{\texttt{CONSOLE }} prints the linear least squares problem in a human readable format
    685   to \texttt{stderr}. The Jacobian is printed as a dense matrix. The vectors
    686    $D$, $x$ and $f$ are printed as dense vectors. This should only be used
    687    for small problems.
    688 \item{\texttt{PROTOBUF }}
    689    Write out the linear least squares problem to the directory
    690    pointed to by \texttt{lsqp\_dump\_directory} as a protocol
    691    buffer. \texttt{linear\_least\_squares\_problems.h/cc} contains routines for
    692    loading these problems. For details on the on disk format used,
    693    see \texttt{matrix.proto}. The files are named
    694    \texttt{lm\_iteration\_???.lsqp}. This requires that
    695    \texttt{protobuf} be linked into Ceres Solver.
    696 \item{\texttt{TEXTFILE }}
    697    Write out the linear least squares problem to the directory
    698    pointed to by \texttt{lsqp\_dump\_directory} as text files
    699    which can be read into \texttt{MATLAB/Octave}. The Jacobian is dumped as a
    700    text file containing $(i,j,s)$ triplets, the vectors $D$, $x$ and $f$ are
    701    dumped as text files containing a list of their values.
    702 
    703    A \texttt{MATLAB/Octave} script called \texttt{lm\_iteration\_???.m} is also output,
    704    which can be used to parse and load the problem into memory.
    705 \end{itemize}
    706 
    707 
    708 
    709 \item{\texttt{check\_gradients }}(\texttt{false})
    710  Check all Jacobians computed by each residual block with finite
    711      differences. This is expensive since it involves computing the
    712      derivative by normal means (e.g. user specified, autodiff,
    713      etc), then also computing it using finite differences. The
    714      results are compared, and if they differ substantially, details
    715      are printed to the log.
    716 
    717 \item{\texttt{gradient\_check\_relative\_precision }} ($10^{-8}$)
    718   Relative precision to check for in the gradient checker. If the
    719   relative difference between an element in a Jacobian exceeds
    720   this number, then the Jacobian for that cost term is dumped.
    721 
    722 \item{\texttt{numeric\_derivative\_relative\_step\_size }} ($10^{-6}$)
    723  Relative shift used for taking numeric derivatives. For finite
    724      differencing, each dimension is evaluated at slightly shifted
    725      values, \eg for forward differences, the numerical derivative is
    726 
    727 \begin{align}
    728  \delta &= \texttt{numeric\_derivative\_relative\_step\_size}\\
    729  \Delta f &= \frac{f((1 + \delta)  x) - f(x)}{\delta x}
    730 \end{align}
    731 
    732 The finite differencing is done along each dimension. The
    733 reason to use a relative (rather than absolute) step size is
    734 that this way, numeric differentiation works for functions where
    735 the arguments are typically large (e.g. $10^9$) and when the
    736 values are small (e.g. $10^{-5}$). It is possible to construct
    737 "torture cases" which break this finite difference heuristic,
    738 but they do not come up often in practice.
    739 
    740 \item{\texttt{callbacks }}
    741   Callbacks that are executed at the end of each iteration of the
    742 \texttt{Minimizer}. They are executed in the order that they are
    743 specified in this vector. By default, parameter blocks are
    744 updated only at the end of the optimization, i.e when the
    745 \texttt{Minimizer} terminates. This behavior is controlled by
    746 \texttt{update\_state\_every\_variable}. If the user wishes to have access
    747 to the update parameter blocks when his/her callbacks are
    748 executed, then set \texttt{update\_state\_every\_iteration} to true.
    749 
    750 The solver does NOT take ownership of these pointers.
    751 
    752 \item{\texttt{update\_state\_every\_iteration }}(\texttt{false})
    753 Normally the parameter blocks are only updated when the solver
    754 terminates. Setting this to true update them in every iteration. This
    755 setting is useful when building an interactive application using Ceres
    756 and using an \texttt{IterationCallback}.
    757 
    758 \item{\texttt{solver\_log}}  If non-empty, a summary of the execution of the solver is
    759  recorded to this file.  This file is used for recording and Ceres'
    760  performance. Currently, only the iteration number, total
    761  time and the objective function value are logged. The format of this
    762  file is expected to change over time as the performance evaluation
    763  framework is fleshed out.
    764 \end{enumerate}
    765 
    766 \section{\texttt{Solver::Summary}}
    767 TBD
    768