Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Google AI agents design Lean-verified quantum programs

AI summary

Google DeepMind and Google Quantum AI reported using Gemini 4-driven AI agents to design quantum programs for three scientific simulation tasks: fusion, photochemistry and drug metabolism. The programs’ correctness and resource costs were checked in Lean, and the team reported lower costs than the previous best solutions, with the largest reduction about 90%. The results were based on resource estimates rather than execution on quantum hardware. The team said it would publish the Lean library, problems and verified programs, but outside researchers could not independently inspect the proofs at the time of the report.

In-depth insight

The approach uses formal proofs to validate quantum programs without executing them on quantum hardware, potentially allowing algorithm research to proceed before sufficiently capable machines are available. However, the reported gains were preprint estimates and could not yet be independently checked.

在新药研发、电池材料和核聚变能源等领域,研究者普遍认为,规模足够大、运行足够可靠的量子计算机有望做到传统计算机做不到的事。

但现实是,目前的量子芯片规模小、错误率高,必须经过纠错才能可靠地完成长时间的计算。谷歌目前的 Willow 芯片只有 105 个量子比特,而公司的目标是在 2029 年前后造出一台“有用的、经过纠错的量子计算机”,这样一台机器大约需要 100 万个物理量子比特。

硬件之外,算法的进展同样关键,算法越省资源,硬件需要达到的规模就越小,量子计算机派上用场的那一天也就来得越早。可机器还没造出来,研究者即便写出了量子算法,也没有机器可以运行;量子比特一多,模拟所需的计算量急剧膨胀,很难再用普通计算机模拟检验。因此过去,一个量子算法的改进往往要靠专家花上多年时间手工推导。

#Google DeepMind#Google Quantum AI#Gemini 4#Lean

Original · 8 min readRead the original
Group chat

No comments yet. Start the conversation.