Abstract:Quasi-Monte Carlo (QMC) methods have developed over several decades. With the explosion in computational science, there is a need for great software that implements QMC algorithms. We summarize the QMC software that has been developed to date, propose some criteria for developing great QMC software, and suggest some steps toward achieving great software. We illustrate these criteria and steps with the Quasi-Monte Carlo Python library (QMCPy), an open-source community software framework, extensible by design with common programming interfaces to an increasing number of existing or emerging QMC libraries developed by the greater community of QMC researchers.

What problem does this paper attempt to address?

The paper primarily explores the challenges faced in developing high-quality Quasi-Monte Carlo (QMC) software and proposes standards and steps for creating excellent QMC software. Specifically, the paper attempts to address the following key issues: 1. **Importance of QMC Software**: The paper emphasizes the need for excellent software to implement QMC algorithms as computational science advances. QMC methods have been widely applied in various fields, such as quantitative finance, uncertainty quantification, and image rendering. 2. **Characteristics of QMC Software**: The paper outlines several key characteristics for evaluating excellent QMC software: - **Integrability**: The ability to integrate with related libraries, allowing users to flexibly choose and combine different components. - **Correctness**: Ensuring the accuracy of the software's computational results to avoid issues caused by incorrect implementations. - **Efficiency**: Optimizing code to reduce computation time and memory usage, thereby improving software performance. - **Accessibility**: Making the software easy to obtain, install, and use, including providing documentation, tutorials, and support. - **Sustainability**: Ensuring the long-term development and updates of the software through community support and maintenance. 3. **Methods to Overcome Challenges**: To achieve the above goals, the paper discusses in detail how to overcome the challenges behind each characteristic, using the Quasi-Monte Carlo Python library (QMCPy) as an example. QMCPy is an open-source community software framework designed with compatibility and extensibility with other QMC libraries in mind. 4. **Future Work**: Finally, the paper points out that QMC methods can be extended to other mathematical problem areas, such as function approximation, Markov chain simulation, and density estimation. It also emphasizes the importance of establishing connections with existing mature software to apply QMC techniques to a broader range of computational mathematics and statistics problems. In summary, this paper aims to provide guidance for developing high-quality QMC software to promote the practical application and development of QMC theory.

Challenges in Developing Great Quasi-Monte Carlo Software

PyQMC: An all-Python real-space quantum Monte Carlo module in PySCF

DeepQMC: an open-source software suite for variational optimization of deep-learning molecular wave functions

Software engineering to sustain a high-performance computing scientific application: QMCPACK

On Bounding and Approximating Functions of Multiple Expectations using Quasi-Monte Carlo

Wasserstein Quantum Monte Carlo: A Novel Approach for Solving the Quantum Many-Body Schrödinger Equation

Quasi-Monte Carlo Beyond Hardy-Krause

Sequential quasi-Monte Carlo: Introduction for Non-Experts, Dimension Reduction, Application to Partly Observed Diffusion Processes

Discussion of "Sequential Quasi-Monte-Carlo Sampling" by M. Gerber and N. Chopin

Quasi-Newton Hamiltonian Monte Carlo.

Quantum computing challenges and solutions in software industry—A multivocal literature review

Dimension Reduction and Smoothing in Quasi-Monte Carlo Method for Financial Engineering

Quantum Dynamical Hamiltonian Monte Carlo

hIPPYlib-MUQ: A Bayesian Inference Software Framework for Integration of Data with Complex Predictive Models under Uncertainty

Nonasymptotic Convergence Rate of Quasi-Monte Carlo: Applications to Linear Elliptic PDEs with Lognormal Coefficients and Importance Samplings

The importance of being scrambled: supercharged Quasi Monte Carlo

Enhancing Quasi-Monte Carlo Methods by Exploiting Additive Approximation for Problems in Finance

Convergence of sequential quasi-Monte Carlo smoothing algorithms

Quasi-Monte Carlo and discontinuous Galerkin

MQT Qudits: A Software Framework for Mixed-Dimensional Quantum Computing

Improving approximate Bayesian computation via quasi-Monte Carlo