The Incomparable QSimulate
The Arrow Not of Time, But of Quantum
The Quantum Dragon is a role-player. The goal of every featured image is to illustrate, in “a picture is worth a thousand words” fashion, the key point being made by the article. But for the first time ever, I’ve got nothin’ for ya. According to QSimulate, no one is doing anything like this, and I don’t have a challenge for that.
Normally, The Quantum Dragon focuses on quantum technologies. QSimulate will eventually use quantum computers to improve the accuracy of its pharma and materials simulations, maybe starting in 2-3 years, but they’re only testing it now. In the meantime, the company is claiming a 1,000X+ speedup over conventional GPU-based quantum mechanics simulations with its QUELO simulation platform.
Balls and Sticks vs Arrows
Human biology is too complex to simulate molecular-level drug and protein interactions with conventional AI methods. QSimulate uses a quantum physics-first approach that directly models molecular behavior at a subatomic level. Consuming only milliseconds per snapshot is the basis for the 1,000X+ speedup claim over traditional methods.
At a super high level, imagine you’ve got a molecule drawn with the typical balls and sticks representing atoms and their bonds. QSimulate removes the crude classical calculations and redraws the molecule using flowing arrows. The calculations are quantum, and that’s how they illustrate that. Simply put: the algorithm is better. Furthermore, they use faithful representations, not small chunks.
Real, Paying Customers
It’s important to note that QUELO has real, paying customers today. Unlike the way the quantum industry develops something and then tries to make it useful, QSimulate starts with a customer’s drug discovery and materials science R&D challenges. They take a vertical approach, as opposed to industry-standard horizontal approaches, and they measure success in terms of the actual problems they’re solving. As the first of its kind solution for peptide drugs and other larger molecules, QUELO is reportedly turning what used to take customers months into a matter of hours.
Repercussions for Quantum Computing
According to QSimulate, the pharmaceutical industry doesn’t care how the simulations are happening. It doesn’t matter if the calculations are classical or quantum. What matters is customer constraints. Customers don’t want to wait for hours, and they have limits to how much they can spend on compute, and QSimulate, or any provider for that matter, needs to work within customer constraints. The good news is that quantum computers are viewed as having the potential to alleviate some of these constraints.
Furthermore, there will be opportunities to use machine learning on the results of future quantum computation. Whether this will be classical machine learning or quantum remains an open question, but there is expected to be value in quantum computation beyond what will execute on real hardware.
Keeping Costs Low
Simulating one compound with QUELO is designed to require only 2 GPUs, running for several hours at an approximate cost of $1 per hour. Some customers would otherwise run 100 GPUs concurrently and continuously to simulate large numbers of drug molecules, so we don’t need quantum calculations to realize that QUELO must help mitigate energy costs.
Conclusion
Speaking of keeping costs low, The Quantum Dragon sure saved a lot of energy this week. He walked into the empty studio, I snapped the featured image above, and… “lunch!”
Analogies require comparisons, after all, and comparisons require having something to compare something to. QSimulate has nothing to compare QUELO to and, quite frankly, neither do I. As much as I would prefer to have some action-packed featured image this week, the illustration needs to fit the article, and we consequently have a blank image. Like this week’s The Quantum Dragon, QUELO stands alone.
