HN user

cocoablazing

244 karma
Posts0
Comments126
View on HN
No posts found.

This had more to do with safety system failres and procedures regarding operation of critical valves than poison kinetics.

LM designs were abandoned prior to this accident due to availability and maintainability issues.

To what end?

The USN has, to a material degree, perfected the compact PWR.

Their experience in fuel design, materials, cooling, support systems, control systems, and processes pose an enormous barrier to new systems other than those that could truly radically simplify the reactor system and yield a huge gain in reduced manpower and increased reliability.

I think the best marginal gains are coming and will come from developments in tactical and ISR systems.

This business requires many to be confronted with complexity at scale. I think these problems are universal but few encounter them due to the distribution of employees in organizational hierarchies.

It’s possible that this code has bugs that have yet to be detected. It’s a fact that quality is important if it can generate losses. Style may not be particularly important, but reliability is.

Apply. I would like nothing more than a flow of applications from smart software engineers with BSs over scientists exiting of postdocing and MCFs.

The skill I look for over all others is an understanding of the standard principles of software design, and principles of testing is a bonus. I want to see that the candidate is honestly interested in our language to the extent that they are familiar with the obvious pitfalls of the standard language features. Candidates who prepared by writing option pricers and passing leetcode challenges won't necessarily understand any of these, because they are simply writing functions.

I assume that anyone who got through a rigorous engineering undergrad has the mathematical preparation to perform as a quantitative developer. This won't be true for all roles, but it's true for all of the QD roles that I have contact with at BB.

If you are interested in a dev role in finance, and see requirements for basic financial knowledge, risk, etc. -- as long as the requirement isn't quite specific, such as experience pricing specific option types, pricing complex products, or with a specific product, I would just ignore it and apply. Many of these concepts are quite simple compared to what SEs are trained to model and implement, and the people on the team who need help know that.

I've seen GPU and C++ skills requested for a Pandas shop. Just apply.

I think hiring managers and HR think they will attract better candidates by adding these things, but it is counterproductive.

From what I can tell from salary surveys, we pay an entry-level dev better than many others. >150 guaranteed all-in first year, and better than that afterwards if you just do your job. Promotion path is good because modern skills are in dire need.

There are rooms in some of these facilities that have been known as infinity rooms because no readily available radiation meter would not saturate in them.

The radiation field of a freshly removed power reactor fuel element is so strong that you couldn’t run past it fast enough to survive the dose.

The issue is intra-predictor correlation. In the extreme case that a predictor is duplicated, the correct beta might be {betaa, beta(1-a)} for a in [0, 1], which an algorithm may not estimate in a stable manner. A significant degree of correlation introduces this general problem.

What’s the advantage of the file system/repo/bespoke diag database over storing the numpy arrays in the existing database infrastructure?

Doesn’t implementing this system with HDF5 cause headaches for concurrency in either direction?

I think deploying a world-killing nuclear arsenal hasn’t been bad for economic and political relationships with critical partners.

I think this has been the case with NATO, Israel, Saudi Arabia, Korea, Taiwan, and Pakistan.

On the balance, this obviously doesn’t make the overall strategy sound in the long-run, but the nuclear enterprise has been myopic from the beginning.

None of those small banks are subject to the regulation that large BHC are. The potential barrier is off by orders of magnitude. The development cycle on those models is multi year. It would take a massive effort to recruit and build the quant teams alone, much less the finance, compliance, and legal personnel with the expertise.

The bulk of research funding in the departments I studied in went to development of technology to support identification and attribution of rogue nuclear weapons attacks, over any other peaceful or constructive use of nuclear technology.

Nuclear weapons states and their force commanders are well aware of this capability. It’s been publicly advertised for decades for deterrence.

Both RF and US operate FBM forces as deterrent second-strike forces. This is not only the operational reality but also the treaty-regulated deployed configuration.

Optimizing deployment in the manner you describe is not aligned to the current mission of these forces / our deterrence relationship.

It's a rule-of-thumb. The fission of Uranium and Plutonium results in an incredible diversity of both isotopes and nuclear isomers.

Most of these isotopes exist on distinct decay chains. An unstable nuclear state may decay in more than one mode, each with a distinct time constant. Many of these decays occur as groups of decay chains that occur on extremely short time-scales, and may temporarily remain in metastable states (isomers).

So this general rule results from the superposition of many thousands of reactions, that proceed stochastically with distinct time constants and energy released.

There is a mathematical rule relating the time constants and energies of these reactions, but it is not simple: https://en.wikipedia.org/wiki/Fermi%27s_golden_rule

Your whole thread of comments simply sounds paranoid and detached from the reality that the US maintains and funds the continued development of the most advanced strategic warfare system in the world.

The Russians and Chinese can't even deploy a reliable survivable force. Their boomers are either rusting or poorly tested and their forces underfunded.

They can't even regularly deploy attack submarines to the Atlantic Ocean -- how are they going to simultaneously fix, track, and neutralize the US boomer force?

Your entire train of thought on US/RF/PRC deterrence relationships is fantastical.