Most backward error bounds for numerical linear algebra algorithms are of the form , for a machine precision and a problem size . The dependence on of these bounds is known to be pessimistic: together with Nick Higham, our recent probabilistic analysis [SIAM J. Sci. Comput., 41 (2019), pp. A2815–A2835], which assumes rounding errors to be independent random variables of mean zero, proves that can be replaced by a small multiple of with high probability. However, even these smaller bounds can still be pessimistic, as the figure below illustrates.
The figure plots the backward error for summation (in single precision) of floating-point numbers randomly sampled from a uniform distribution. For numbers in the distribution, the bound is almost sharp and accurately predicts the error growth. However, for the distribution, the error is much smaller, seemingly not growing with . This strong dependence of the backward error on the data cannot be explained by the existing bounds, which do not depend on the values of the data.
In our recent preprint, we perform a new probabilistic analysis that combines a probabilistic model of the rounding errors with a second probabilistic model of the data. Our analysis reveals a strong dependence of the backward error on the mean of the data : indeed, our new backward error bounds are proportional to . Therefore, for data with small or zero mean, these new bounds are much sharper as they bound the backward error by a small multiple of the machine precision independent of the problem size .
Motivated by this observation, we also propose new algorithms that transform the data to have zero mean, so as to benefit from these more favorable bounds. We implement this idea for matrix multiplication and show that our new algorithm can produce significantly more accurate results than standard matrix multiplication.
Embedded systems are based on low-power, low-performance processors and can be found in various medical devices, smart watches, various communication devices, cars, planes, mobile phones and many other places. These systems come in a hardware and software package optimized for specific computational tasks and most commonly have real-time constraints. As these systems usually have energy usage and cost constraints too, sophisticated numerical hardware that can process floating-point data is not included, but rather only integer arithmetic, which is simpler in terms of area and power of the processors.
ISO 18037:2008 is a standard for embedded C programming language support. It lays out various rules that C compilers should support to make embedded systems easier to program using a high-level language. One of the most important definitions in this standard is fixed-point arithmetic data types and operations. Support for fixed-point arithmetic is highly desirable, since if it is not provided integers with scaling factors have to be used, which makes code hard to maintain and debug and most commonly requires assembler level changes or completely new implementations for each different platform.
The GCC compiler provides some support of the fixed-point arithmetic defined in this standard for ARM processors. However, in my recent technical report (https://arxiv.org/abs/2001.01496) I demonstrated various numerical pitfalls that programmers of embedded systems based on ARM and using GCC can get into. The issues demonstrated include
larger than half machine epsilon errors in rounding decimal constants to fixed-point data types,
errors in conversions between different data types,
incorrect pre-rounding of arguments of mixed-format arithmetic operations before the operation is performed, and
lack of rounding of the outputs of arithmetic operations.
These findings can be used to improve the accuracy of various embedded numerical libraries that might be using this compiler. To demonstrate one of the issues, here is a piece of test code:
The multiplication operation is a mixed-format operation, since it multiplies an unsigned long fract argument with an accum argument, therefore it is subject to prerounding of the unsigned long fract argument as described in the report. Since the comparison step in the if () sees that the argument a is larger than zero and b larger than 1, the code is executed with a hope that c will not be set to zero. However, in the arithmetic operation, a is incorrectly pre-rounded to 0, which causes c = 0*b, an unexpected outcome and a bug that is hard to detect and fix.
In many applications requiring the solution of a linear system , the matrix possesses a block low-rank (BLR) property: most of its off-diagonal blocks are of low numerical rank and can therefore be well approximated by low-rank matrices. This property arises for example in the solution of discretized partial differential equations, because the numerical rank of a given off-diagonal block is closely related to the distance between the two subdomains associated with this block. BLR matrices have been exploited to significantly reduce the cost of solving , in particular in the context of sparse direct solvers such as MUMPS, PaStiX, and STRUMPACK.
However, the impact of these low-rank approximations on the numerical stability of the solution in floating-point arithmetic has not been previously analyzed. The difficulty of such an analysis lies in the fact that, unlike for classical algorithms without low-rank approximations, there are two kinds of errors to analyze: floating-point errors (which depend on the unit roundoff ), and low-rank truncation errors (which depend on the the low-rank threshold , the parameter controlling the accuracy of the blockwise low-rank approximations). Moreover, the two kinds of error cannot easily be isolated in the analysis, because the BLR compression and factorization stages are often interlaced.
In our recent preprint, with Nick Higham, we present rounding error analysis for the solution of a linear system by LU factorization of BLR matrices. Assuming that a stable pivoting scheme is used, we prove backward stability: the relative backward error is bounded by a modest constant times , and does not depend on the unit roundoff as long as is safely smaller than . This is a very desirable theoretical guarantee that can now be given to the users, who can therefore control the numerical behavior of BLR solvers simply by setting to the target accuracy. We illustrate this key result in the figure below, which shows a strong correlation between the backward error and the threshold for 26 matrices from the SuiteSparse collection coming from a wide range of real-life applications.
The adjective “Bohemian” was used for the first time in a linear algebra context by Robert Corless and Steven Thornton to describe the eigenvalues of matrices whose entries are taken from a finite discrete set, usually of integers. The term is a partial acronym for “BOunded Height Matrix of Integers”, but the origin of the term was soon forgotten, and the expression “Bohemian matrix” is now widely accepted.
As Olga Taussky observed already in 1960, the study of matrices with integer elements is “very vast and very old”, with early work of Sylvester and Hadamard that dates back to the second half of the nineteenth century. These names are the first two in a long list of mathematicians that worked on what is now known as the “Hadamard conjecture”: for any positive integer multiple of 4, there exists an by matrix , with entries and , such that .
If this is the best-known open problem surrounding Bohemian matrices, it is far from being the only one. During the 3-day workshop “Bohemian Matrices and Applications” that our group hosted in June last year, Steven Thornton released the Characteristic Polynomial Database, which collects the determinants and characteristic polynomials of billions of samples from certain families of structured as well as unstructured Bohemian matrices. All the available data led Steven to formulate a number of conjectures regarding the determinants of several families of Bohemian upper Hessenberg matrices.
Gian Maria Negri Porzio and I attended the workshop, and set ourselves the task of solving at least one of these open problems. In our recent preprint, we enumerate all the possible determinants of Bohemian upper Hessenberg matrices with ones on the subdiagonal. We consider also the special case of families with main diagonal fixed to zero, whose determinants turn out to be related to some generalizations of Fibonacci numbers. Many of the conjectures stated in the Characteristic Polynomial Database follow from our results.
Summing numbers is a key computational task at the heart of many numerical algorithms. When performed in floating-point arithmetic, summation is subject to rounding errors: for a machine precision , the error bound for the most basic summation algorithms, such as recursive summation, is proportional to .
Nowadays, with the growing interest in low floating-point precisions and ever increasing in applications, such error bounds have become unacceptably large. While summation algorithms leading to smaller error bounds are known (compensated summation is an example), they are computationally expensive.
In our recent preprint, Pierre Blanchard, Nick Higham and I propose a class of fast and accurate summation algorithms called FABsum. We show that FABsum has an error bound of the form , where is a block size, which is independent of to first order. As illustrated by the figure below, which plots the measured error using single precision as a function of , FABsum can deliver substantially more accurate results than recursive summation. Moreover, FABsum can be easily incorporated in high-performance numerical linear algebra kernels in order to boost accuracy with only a modest drop in performance, as we demonstrate in the paper with the PLASMA library.
In earlier blog posts, I wrote about the benefits of using half precision arithmetic (fp16) and about the problems of overflow and underflow in fp16 and how to avoid them. But how can one experiment with fp16, or other low precision formats such as bfloat16, in order to study how algorithms behave in these arithmetics? (For an accessible introduction to fp16 and bfloat16 see the blog post by Nick Higham.)
As of now, fp16 is supported by several GPUs, but these are specialist devices and they can be very expensive. Moreover, architectures that support bfloat16 have not yet not been released. Therefore software that simulates these floating-point formats is needed.
In our latest EPrint, Nick Higham and I investigate algorithms for simulating fp16, bfloat16 and other low precision formats. We have also written a MATLAB function chop that can be incorporated into other MATLAB codes to simulate low precision arithmetic. It can easily be used to study the effect of low precision formats on various algorithms.
Imagine a hypothetical situation where the computer can just represent integers. Then the question is how do we represent numbers like 4/3? An obvious answer would be to represent it via the integer closest to it, 1 in this case. However, one will have to come with a convention to handle the case where the number is in the centre. Now replace the integer in the example with floating-point numbers, and a similar question arises. This process of converting any given number to a floating-point number is called rounding. If we adopt a rule where we choose the closest floating-point number (as above), then we formally call it as ‘round to nearest’. There are other ways to round as well, and different rounding modes can yield different results for the same code. However meddling with the parameters of a floating-point format without a proper understanding of their consequences can be a recipe for disaster. Cleve Moler in his blog on sub-normal numbers makes this point by warning ‘don’t try this at home’. The MATLAB software we have written provides a safe environment to experiment with the effects of changing any parameter of a floating-point format (such as rounding modes and support of subnormal numbers) on the output of a code. All the technical details can be found in the Eprint and our MATLAB codes.
The wave-kernel functions and arise in the solution of second order differential equations such as with initial conditions at . Here, is an arbitrary square matrix and . The square root in these formulas is illusory, as both functions can be expressed as power series in , so there are no questions about existence of the functions.
How can these functions be computed efficiently? In Computing the Wave-Kernel Matrix Functions (SIAM J. Sci. Comput., 2018) Prashanth Nadukandi and I develop an algorithm based on Padé approximation and the use of double angle formulas. The amount of scaling and the degree of the Padé approximant are chosen to minimize the computational cost subject to achieving backward stability for in exact arithmetic.
In the derivation we show that the backward error of any approximation to can be explicitly expressed in terms of a hypergeometric function. To bound the backward error we derive and exploit a new bound for in terms of the norms of lower powers of ; this bound is sharper than one previously obtained by Al-Mohy and Higham.
Numerical experiments show that the algorithm behaves in a forward stable manner in floating-point arithmetic and is superior in this respect to the general purpose Schur–Parlett algorithm applied to these functions.
The fundamental regions of the
function cosh(sqrt(z)), needed for the backward error analysis