Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
blt
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
65 ms
·
751.
▲
by
blt
8y ago
Transit App for public transport route choosing and timetables. In fact, it is much better than Google maps for this purpose. https://transitapp.com/
752.
▲
by
blt
8y ago
They do it because C++ tempts you to strive for perfect expressive syntax and perfect efficiency at the same time. Most languages have limited expressive power (C, Fortran, Pascal) or limited efficiency (Haskell, ML, any JVM or scripting l
753.
▲
by
blt
8y ago
yeah, string interpolation is such a killer feature. I love using it in Python 3.6.
754.
▲
by
blt
8y ago
I have my display set to "scaled". I get the worst slowdown when streaming videos. 2013 15" w/ Nvidia gpu.
755.
▲
by
blt
8y ago
Psychotic disorders can turn sweet people into highly aggressive people. It's heartbreaking to see.
756.
▲
by
blt
8y ago
City planners and politicians in the USA would be wise to consider the root cause of the scooter surge rather than worrying too much about the side effects. People want to get around without buying an expensive heavy machine. Scooters solve
757.
▲
by
blt
8y ago
Lots of people would criticize USA culture by saying something like "Apparently it is culturally accepted thing for cops to shoot black people". Apparently it is, because we do not see any significant changes being made to the tra
758.
▲
by
blt
8y ago
I had the same reaction. It looks like they fit a spline to the data, and then took the analytical derivative of the spline. But there are other ways to do it like locally weighted regression. In general estimating derivatives from real wor
759.
▲
by
blt
8y ago
Did you use language interop for matplotlib, or saving/loading data files?
760.
▲
by
blt
8y ago
I'm struggling to apply this model to code for machine learning experiments. The functional core takes an extremely long time to compute, e.g. "train the model". Thus, you want to save lots of intermediate values to disk so y
761.
▲
by
blt
8y ago
You are probably right. I realize my previous post sounded pessimistic but I was only trying to state my ignorance :)
762.
▲
by
blt
8y ago
The gyro effect is pretty predictable. Anything where you're pretty confident in your model, it's better to compensate in the feedforward part of the controller than to push it into the feedback part. Think about gravity compensat
763.
▲
by
blt
8y ago
It might not be that difficult to implement a high level DSL around one of the geometric modeling kernels like Parasolid, ACIS, etc. that form the computational core of GUI CAD programs. But I don't know if any equally powerful CAD ker
764.
▲
by
blt
8y ago
a lot of mathematicians prefer chalkboards over whiteboards too, because 85% of all known whiteboard markers are nearly empty. love of math and love of technology aren't as closely linked as you believe.
765.
▲
by
blt
8y ago
I don't know anything about the problem domain but I'm guessing that you need to make some topological as well as geometric choices... In which case the evolutionary algorithms could be interesting
766.
▲
by
blt
8y ago
This is really cool but it looks like the constraint solver [1] is not much more complex than the one shown in this web page. I really want a program-driven functional CAD tool like libfive or OpenSCAD, but with powerful 3d constraint solvi
767.
▲
by
blt
8y ago
Multi-start optimization is the simplest way to deal with the initial guess problem. Is there any heuristic you can use to generate a reasonably-sized set of guesses that is likely to contain at least one guess contained in the region of co
768.
▲
by
blt
8y ago
Even if the data is not retained, it's possible that the trained model can leak information about the training examples. The field of differential privacy explores ways to make sure this doesn't happen.
769.
▲
by
blt
8y ago
They are different skills. Not saying that it's hard to learn both, but there are standardized career paths that will lead you to be good at the modeling / techniques side of "data science" without learning much about so
770.
▲
by
blt
8y ago
False. "The percentage of CS-degree holders who were women peaked in the 1980s at 34% and has been on a downward trend ever since" [1]. Women do not have an increasing presence in terms of population. Maybe the real phenomenon is
771.
▲
by
blt
8y ago
They won't sell it to embedded developers... They'll sell it to javascript developers who want to try out embedded development.
772.
▲
by
blt
8y ago
Every web designer should be required to study McMaster-Carr's interface. It's so much better than any other e-commerce site, it feels like alien technology from the future. Of course, the irony is that it could be replicated with
773.
▲
by
blt
8y ago
doubtful, nobody will ever look back wistfully on the aesthetic of early Medium pages.
774.
▲
by
blt
8y ago
what does the "-gazing" suffix mean?
775.
▲
by
blt
8y ago
rent-seeking is too seductive to the capitalist
776.
▲
by
blt
8y ago
Agreed. I think it's hard for a non programmer to estimate the difficulty of implementing a certain feature. Both in time and in technical debt. For example if some feature really doesn't fit in the program architecture and will r
777.
▲
by
blt
8y ago
Yes, and apply the same philosophy to your product manager, etc. as a developer. Developers tend to assume that the manager has carefully considered the cost of what they ask for, and is certain that the result is worth it. But this usually
778.
▲
by
blt
8y ago
... more about giving user more control than any new algorithmic techniques. Was expecting e.g. GAN discriminators
779.
▲
by
blt
8y ago
I think the general philosophy of Rust is that programmers should think about the data ownership semantics of their programs. Rather than boxing everything to make it easier to reason about closures, one should strive to write programs that
780.
▲
by
blt
8y ago
I went back for a PhD after a few years in industry. As a PhD student, my code and habits do not meet the standard of industry. This is because I'm constantly changing the whole architecture to try new ideas, so I I optimize for small
More ›