Ball Lightning

After setting up a few mathematical models, I gradually got the hang of it, and the next models I constructed reflected increasing numbers of the known characteristics of ball lightning. At the same time, the models required an increasing number of calculations, and sometimes my desktop would run for days before completing a model. At New Concepts, Lin Yun networked eighteen machines, and she and I broke the models down into eighteen parts that could execute separately on those machines as close to simultaneously as possible and combine their results, greatly increasing efficiency.

When I finally completed a model that exhibited all of ball lightning’s known characteristics, the event Lin Yun had feared had already taken place. This time, she didn’t immediately start programming the model when she received it, but spent several days conducting estimates of its computational complexity. When she obtained the results, she let out a long sigh.

“We have a problem,” she said. “One round of calculations for this model will need to run for five hundred thousand hours on a single computer.”

I was shocked. “That’s… more than fifty years?”

“Yes. From past experience, every model requires several rounds of debugging before it’s operational—more in this case, since it’s such a complex model. We can only allow ten days to complete a simulation.”

I mentally estimated: “We’d need two thousand computers working simultaneously.”

Instead, we started a search for a mainframe, but it wasn’t easy. Neither Lightning Institute nor New Concepts had one; their biggest machine was an AlphaServer. The military’s mainframes were busy and had tight restrictions, and since ours wasn’t a registered military project, Lin Yun could not win us their use despite repeated attempts. So we had to place our hopes on civilian-sector mainframes, where Lin Yun and I had no connections at all. We turned to Gao Bo for ideas.

His situation wasn’t good. When he took up his position, he had converted all of the institutional departments into business units that were completely reliant on the market. He had conducted competitive rehiring and laid off a huge number of staff. His HR conduct was more impulsive than careful, and combined with his poor understanding of national conditions and human sentiment, his relationships were tense throughout the hierarchy.

His business failures were even worse. The first thing he did in office was to focus the Institute’s main strength on new surge arrestors and eliminators vastly different from conventional anti-lightning systems, including semiconductor eliminators, optimized lightning rods, laser lightning attractors, rocket lightning attractors, and water column lightning attractors. However, right around that time, new surge arrestors and eliminators were under discussion at a conference of the Chinese Society for Electrical Engineering’s High Voltage Committee, Overvoltage and Insulation Coordination Subcommittee. Minutes from the meeting showed a decision that, as there had been no theoretical or practical demonstration that these nonstandard products had any superior functionality to ordinary anti-lightning devices, and seeing as many R&D problems still remained, nonstandard anti-lightning products therefore could not be adopted in engineering projects. Due to the group’s authority and influence, the conference’s viewpoint was destined to be adopted by state-designated lightning projects, meaning that any such technology currently under development would be completely shut out of the market, wasting enormous investment. When I went to Gao Bo to talk about the mainframe, he was looking for me, to ask me to put ball lightning research on hold and concentrate my energy on developing a new lightning location system for power supply systems. He also wanted me to complete the design of an anti-lightning project for the Beijing Capital Theater. Hence the mainframe was a no-go. I had to do ball lightning research in my spare time.

Lin Yun and I tried some other leads, but it turned out that in the era of commoditized computers, mainframes were scarce.

“We’re pretty fortunate, really,” she said. “Our calculations are nothing next to the world’s supercomputing projects. I recently saw the data for a US DOE nuclear test simulation, which their current twelve teraFLOPS is far from satisfying. They’re setting up a cluster incorporating twelve thousand individual Alpha processors that can achieve speeds on the order of one hundred teraFLOPS. Our calculations rate as conventional compared to that, so we should be able to find a solution.”

She acted like a warrior at all times. No matter the difficulty, she pushed forward, minimizing my stress by understating the difficulties involved. It was something I ought to have been doing for her.