Succinct data structures such as wavelet trees are widely used in bioinformatics. There you often have strings that cannot be tokenized or parsed meaningfully, so you just have to deal with sequences of symbols. And because the strings are often very long and/or there can be a huge number of them, the data structures have to be space-efficient.
A wavelet tree is best seen as an intermediate data structure. It doesn't do anything particularly interesting on its own, but it can be used as a building block for higher-level data structures. For example, you can create an FM-index by storing the Burrows–Wheeler transform in a wavelet tree. (Though there are better options when the alphabet is small.) And then you can use the FM-index to find exact matches of any length between the pattern and the indexed strings.
People working with succinct data structures often talk about bitvectors rather than bitmaps. The difference is that bitmaps tend to focus on set operations, while bitvectors are more about random access with rank, select, and related queries. Then you could see wavelet trees as a generalization of bitvectors from a binary alphabet to larger alphabets. And then you have people talking about wavelet trees, when they really mean a wider class of conceptually and functionally similar data structures, regardless of the actual implementation.
Hyperlinks are often banned in various kinds of petitions and applications. Mostly to ensure that the entire application is submitted at once and does not change afterwards. Then you can process the application in multiple passes (maybe first for the formal requirements and then for the actual content), confident that the conclusions from the earlier passes are still valid.
Surface ships are still the only way to transport large quantities of troops and equipment over long distances. If you want to maintain the ability to project force beyond oceans, you need a navy to escort the vulnerable transport ships and to fight whatever threats they would be facing.
In 1945, US GDP per capita was almost $1600. Using your conversion factors, that would be almost $150k today. The actual number is something like $85k. I don't think Americans are that much poorer today than they were 80 years ago.
Poe's law strikes again. Is this supposed to be a parody of goldbugs or do you seriously think Americans were that much richer in 1945? Without a wink we don't know
"Wavelet tree" is not just a collective label but the name explicitly given by the authors of the paper where the data structure was first described in. At least Vitter had worked in image/video compression, where wavelet transforms and similar techniques are common. I believe the original idea was adapting those techniques for representing strings, and the wavelet tree data structure was the final outcome.
Industry researchers are less likely to release their findings to the public, because their bosses choose that way. They are also doing research, but the outcomes of the research are fundamentally different.
The are two fundamentally different kinds of private research funding. (Let's drop the &D part, because that's mostly unrelated to the kind of research we are talking about.)
Charitable foundations and similar organizations are not that different from government agencies funding research. They act on a smaller scale, because rich people are not actually that rich.
Then there are companies that do research as part of their business. They are typically much better funded and much narrower in scope than government-funded research. They are also biased towards topics that can be reasonably expected to work and produce economic value within the next 10-20 years. This kind of research is inherently inefficient due to redundant efforts. Instead of making their findings public, companies often keep the results secret, forcing their competitors to waste money on reinventing the wheel.
Hiring smart people who want freedom and autonomy is easy. Just give them freedom, autonomy, stability, and a good enough salary. The hard part is getting them to contribute to your business. Maybe they will contribute if they find it interesting. But if you expect them to contribute, you are clearly not giving them autonomy.
Many of the smartest people I know are good at ignoring bureaucratic requirements, or at least handling them with the minimum effort necessary. And that often includes business, which many of them see as a subcategory of bureaucracy.
The national median over what? When I was looking for similar figures in Finland, I found that the median income for households with two parents and at least one child was about twice the median over all households. And it's plausible that the median would be even higher for households with two parents and at least one child about to reach college age.
Employers, not universities. Employers were the ones who decided that most good jobs require a college degree. And then they sent administrators as their minions to take over the universities. Americans often call their universities colleges, which implies an organizational structure where the senior professionals doing the actual work are in charge. And that's how things used to work. But somehow that got turned around, with the administrators in charge and business interests treating universities as vocational schools.
One way to look at this is through different motivations to studying: internal (you want to learn things), external (you want to be successful at whatever is being measured), and pragmatic (you want to pass the classes, get the degree, and move on). Universities prefer internally motivated students – those who study to learn rather than for jobs and career success. They can try to educate people with other motivations, but what those students want is inherently in conflict with what the professors want.
A wavelet tree is best seen as an intermediate data structure. It doesn't do anything particularly interesting on its own, but it can be used as a building block for higher-level data structures. For example, you can create an FM-index by storing the Burrows–Wheeler transform in a wavelet tree. (Though there are better options when the alphabet is small.) And then you can use the FM-index to find exact matches of any length between the pattern and the indexed strings.
People working with succinct data structures often talk about bitvectors rather than bitmaps. The difference is that bitmaps tend to focus on set operations, while bitvectors are more about random access with rank, select, and related queries. Then you could see wavelet trees as a generalization of bitvectors from a binary alphabet to larger alphabets. And then you have people talking about wavelet trees, when they really mean a wider class of conceptually and functionally similar data structures, regardless of the actual implementation.
reply