Berkeley Researchers Reveal New Insights into Numerical Analysis for Quantum Chemistry

May 29, 2024

May 29, 2024 — Mathematicians like Dr. Lin Lin, a faculty scientist in the Applied Mathematics and Computational Research division at Lawrence Berkeley National Laboratory (Berkeley Lab) and a Professor of Mathematics at the University of California, Berkeley, and Dr. Xin Xing, a former postdoctoral scholar from Lin’s group at UC Berkeley who is now at Apple, have years of experience practicing numerical analysis. They and many of their colleagues are increasingly recognizing the need to convey the underlying concepts and the impact of numerical analysis in ways that fellow domain scientists may find useful.

Dr. Lin Lin (left), a faculty scientist in the Applied Mathematics and Computational Research division at Berkeley Lab and a Professor of Mathematics at the University of California, Berkeley; and Dr. Xin Xing (right), a former postdoctoral scholar from Lin’s group at UC Berkeley who is now at Apple.

“Many scientific computing problems are described by continuous mathematics, but to solve them on a computer the problem must be represented using discrete quantities such as a finite number of points. This process can introduce many interesting problems,” Lin said.

For example, quantum mechanics is often used to compute microscopic physical properties of a particular material structure. These calculations are invaluable in material design, as they allow for the prediction of a material’s properties before actual chemical synthesis manufacturing. However, applying these calculations to realistic materials directly can be challenging because the number of atoms is often too big to be computationally. A common workaround is to only model a small portion of the whole system and hope that the computed results are good approximations. This difference between the results computed from a partial system and the exact values for the full system (conceptually of infinite size) is known as the finite-size error.

“Numerical analysis is the key for finding efficient algorithms for such a discretization process, or even for understanding the underlying physics phenomenon associated with the discretization, such as the finite-size error,” Lin said.

Building on previous work exploring finite-size error, in a paper published March 28 in Physical Review X (PRX), Xing and Lin introduce a new approach for understanding the finite-size error in condensed matter systems that uses coupled-cluster (CC) theory, one of the most accurate methods for treating electron correlation in quantum many-body systems. To do this, they developed novel tools in numerical analysis that can mathematically explain the various sources of the finite-size error and provide a rigorous error estimate.

Using coupled cluster doubles (CCD) theory – mathematically the simplest form of CC theory – as an example, they set out to describe and estimate the scaling of the finite-size error with respect to the size of the computational domain. Understanding the finite-size error and its scaling when the system increased in size is important in a variety of practical applications, Xing noted; in material design, for example, reducing the finite-size error means that we can more accurately predict physical properties of a target material.

Understanding the finite-size error and its finite-size scaling is important in a variety of applications. In material design, for example, reducing this error enables more accurate prediction of the physical properties of a target material. Credit: Xin Xing.

“In quantum chemistry, there are approximate models and theories with different levels of accuracy, and the coupled cluster theory serves as the ‘gold standard’ in molecular quantum chemistry,” Xing said. “One challenge is that for different theories, their finite-size error scaling can be very different. Previously, we’ve analyzed the finite-size error for models of lower accuracy, such as Hartree-Fock and low order many-body perturbation theories, which paves the way to our current work. But the coupled cluster theory is considerably more challenging to analyze.”

Numerical analysis has been fundamental to this approach, which provides the right tool to describe the source of the finite-size error and to provide a rigorous error estimate, Xing emphasized. “It connects the finite-size error in quantum chemistry calculations to a fundamental concept in numerical analysis called quadrature error analysis.”

Quadrature is “a fancy term that says you want to compute the value of an integral and replace it by a finite sum, a process that introduces a quadrature error,” Lin explained. “What was not shown before is that the finite-size error – despite its complex nature in physics – at its core boils down to a quadrature error.”

As a result, this new work implies that many tools in the numerical analysis community may be useful in reducing the finite-size error in quantum sciences. “This is a very exciting interdisciplinary direction,” Xing said.

Solving a Mystery, Filling a Gap

Even though researchers have previously proposed various analytical estimates, rigorous and comprehensive descriptions and estimates of the finite-size error for many quantum chemistry methods are mostly missing. “This new work fills that gap” and solves the mystery of an unexpected inverse volume scaling of this error in CCD calculations, Xing noted. Inverse volume scaling indicates that the finite-size error is inversely proportional to the volume of the size of the simulation cell used to model the system, resulting in much quicker error reduction compared to inverse linear scaling. In inverse linear scaling, the finite-size error decreases proportionally to the size of the simulation cell.

For example, if a 3D simulation cell has dimensions of 1 unit length along each axis, doubling the size to 2 unit lengths along each dimension would decrease the finite-size error by a factor of 1/2 in inverse linear scaling. In contrast, inverse volume scaling reduces the error by a factor of 1/8, demonstrating a more rapid decrease.

To resolve this, rigorous mathematical understanding of CCD and its finite-size error are essential. To achieve this, Xing and Lin had to overcome some new mathematical challenges. For example, the CC theory is a very complicated method even for a standalone finite-size system. This makes it technically even more challenging to analyze its finite-size error, which corresponds to analyzing its convergence when applying the CC calculation to many systems of increasing sizes (all the way up to the whole system, which conceptually is characterized as an infinite-sized one). In fact, it already takes some effort to translate the finite-size error problem into a purely numerical analysis problem (the quadrature error).

“The translated numerical quadrature error estimate problem itself is also technically challenging, which is why we developed new tools in numerical analysis to address this problem,” Lin said.

As noted by the PRX reviewer, this method of estimating finite-size error rates settles once and for all how the finite-size error should behave, Lin added. “Practitioners can now apply this inverse volume scaling law with confidence.”

It has other implications for practitioners, method developers, and theorists as well. For practitioners, reducing finite-size errors in quantum chemistry methods using techniques such as power-law extrapolation requires an in-depth understanding of the error scaling. This is particularly important when calculations are constrained to small-sized systems due to the steep increase of the computational cost with respect to the system size and limited resources.

For method developers, Xing and Lin have shown how to connect the finite-size error to the mature applied mathematical domain of numerical quadrature calculations, which points to new methods for further finite-size error reduction. Meanwhile, for theorists, some critical questions persist: How can the finite-size error analysis be integrated with the study of other equally important error sources in quantum chemistry calculations? How should the finite-size error behavior in many more complicated systems be analyzed?

“Numerical analysis is a really powerful tool that may sometimes be underappreciated in science,” Lin said. “It can be very compelling for understanding complex physics phenomena, and that is precisely what we are trying to do here. Finite-size error is a piece of complicated physics phenomena that has been bothering scientists for some time, and it is not easy to figure out empirically sometimes and not mention its origin. That is what Xin and I are trying to show in this recent work: that the big piece of the puzzle and the key to unlocking the interpretation and explanation is through numerical analysis.”

This work is supported in part by the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research and Office of Basic Energy Sciences through the Scientific Discovery through Advanced Computing program; the Applied Mathematics Program of the U.S. Department of Energy Office of Advanced Scientific Computing Research; and the Simons Investigator Award.


Source: Kathy Kincade, Berkeley Lab Computing Sciences Research

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

HPE and NVIDIA Join Forces and Plan Conquest of Enterprise AI Frontier

June 20, 2024

The HPE Discover 2024 conference is currently in full swing, and the keynote address from Hewlett-Packard Enterprise (HPE) CEO Antonio Neri on Tuesday, June 18, was an unforgettable event. Other than being the first busi Read more…

Slide Shows Samsung May be Developing a RISC-V CPU for In-memory AI Chip

June 19, 2024

Samsung may have unintentionally revealed its intent to develop a RISC-V CPU, which a presentation slide showed may be used in an AI chip. The company plans to release an AI accelerator with heavy in-memory processing, b Read more…

ASC24 Student Cluster Competition: Who Won and Why?

June 18, 2024

As is our tradition, we’re going to take a detailed look back at the recently concluded the ASC24 Student Cluster Competition (Asia Supercomputer Community) to see not only who won the various awards, but to figure out Read more…

Qubits 2024: D-Wave’s Steady March to Quantum Success

June 18, 2024

In his opening keynote at D-Wave’s annual Qubits 2024 user meeting, being held in Boston, yesterday and today, CEO Alan Baratz again made the compelling pitch that D-Wave’s brand of analog quantum computing (quantum Read more…

Apple Using Google Cloud Infrastructure to Train and Serve AI

June 18, 2024

Apple has built a new AI infrastructure to deliver AI features introduced in its devices and is utilizing resources available in Google's cloud infrastructure.  Apple's new AI backend includes: A homegrown foun Read more…

Argonne’s Rick Stevens on Energy, AI, and a New Kind of Science

June 17, 2024

The world is currently experiencing two of the largest societal upheavals since the beginning of the Industrial Revolution. One is the rapid improvement and implementation of artificial intelligence (AI) tools, while the Read more…

HPE and NVIDIA Join Forces and Plan Conquest of Enterprise AI Frontier

June 20, 2024

The HPE Discover 2024 conference is currently in full swing, and the keynote address from Hewlett-Packard Enterprise (HPE) CEO Antonio Neri on Tuesday, June 18, Read more…

Slide Shows Samsung May be Developing a RISC-V CPU for In-memory AI Chip

June 19, 2024

Samsung may have unintentionally revealed its intent to develop a RISC-V CPU, which a presentation slide showed may be used in an AI chip. The company plans to Read more…

Qubits 2024: D-Wave’s Steady March to Quantum Success

June 18, 2024

In his opening keynote at D-Wave’s annual Qubits 2024 user meeting, being held in Boston, yesterday and today, CEO Alan Baratz again made the compelling pitch Read more…

Shutterstock_666139696

Argonne’s Rick Stevens on Energy, AI, and a New Kind of Science

June 17, 2024

The world is currently experiencing two of the largest societal upheavals since the beginning of the Industrial Revolution. One is the rapid improvement and imp Read more…

Under The Wire: Nearly HPC News (June 13, 2024)

June 13, 2024

As managing editor of the major global HPC news source, the term "news fire hose" is often mentioned. The analogy is quite correct. In any given week, there are Read more…

Labs Keep Supercomputers Alive for Ten Years as Vendors Pull Support Early

June 12, 2024

Laboratories are running supercomputers for much longer, beyond the typical lifespan, as vendors prematurely deprecate the hardware and stop providing support. Read more…

MLPerf Training 4.0 – Nvidia Still King; Power and LLM Fine Tuning Added

June 12, 2024

There are really two stories packaged in the most recent MLPerf  Training 4.0 results, released today. The first, of course, is the results. Nvidia (currently Read more…

Highlights from GlobusWorld 2024: The Conference for Reimagining Research IT

June 11, 2024

The Globus user conference, now in its 22nd year, brought together over 180 researchers, system administrators, developers, and IT leaders from 55 top research Read more…

Atos Outlines Plans to Get Acquired, and a Path Forward

May 21, 2024

Atos – via its subsidiary Eviden – is the second major supercomputer maker outside of HPE, while others have largely dropped out. The lack of integrators and Atos' financial turmoil have the HPC market worried. If Atos goes under, HPE will be the only major option for building large-scale systems. Read more…

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Everyone Except Nvidia Forms Ultra Accelerator Link (UALink) Consortium

May 30, 2024

Consider the GPU. An island of SIMD greatness that makes light work of matrix math. Originally designed to rapidly paint dots on a computer monitor, it was then Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Some Reasons Why Aurora Didn’t Take First Place in the Top500 List

May 15, 2024

The makers of the Aurora supercomputer, which is housed at the Argonne National Laboratory, gave some reasons why the system didn't make the top spot on the Top Read more…

Leading Solution Providers

Contributors

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Intel’s Next-gen Falcon Shores Coming Out in Late 2025 

April 30, 2024

It's a long wait for customers hanging on for Intel's next-generation GPU, Falcon Shores, which will be released in late 2025.  "Then we have a rich, a very Read more…

Google Announces Sixth-generation AI Chip, a TPU Called Trillium

May 17, 2024

On Tuesday May 14th, Google announced its sixth-generation TPU (tensor processing unit) called Trillium.  The chip, essentially a TPU v6, is the company's l Read more…

The NASA Black Hole Plunge

May 7, 2024

We have all thought about it. No one has done it, but now, thanks to HPC, we see what it looks like. Hold on to your feet because NASA has released videos of wh Read more…

Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study

June 10, 2024

Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst fir Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

GenAI Having Major Impact on Data Culture, Survey Says

February 21, 2024

While 2023 was the year of GenAI, the adoption rates for GenAI did not match expectations. Most organizations are continuing to invest in GenAI but are yet to Read more…

AMD Clears Up Messy GPU Roadmap, Upgrades Chips Annually

June 3, 2024

In the world of AI, there's a desperate search for an alternative to Nvidia's GPUs, and AMD is stepping up to the plate. AMD detailed its updated GPU roadmap, w Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire