300nm displacements at 10Hz can be measured on their hall floor during the daytime due to human activity in the city, so they built this thing with “geometrical #antisprings”, described as “a tunable spring made by a crown of curved cantilever blades compressed each against the other” (negative #stiffness coefficient constructs?) to isolate their gravity-wave detector. They’re using electronic positive feedback to get “natural frequencies down to 0.2 Hz” and attenuate seismic noise by 80dB.
2007 #pdf #paper about #antisprings (negative-#stiffness springs) for seismic isolation, explaining how to improve their isolation by another order of magnitude with a counterweight rod thing that I don’t understand yet.
2004 #pdf #paper about #antisprings for seismic isolation. Clarifies that YES, an antispring has a negative #stiffness coefficient. In this case they’re combining antisprings with regular springs to cancel out their stiffness and get near-zero restoring force for small displacements. They managed to get a resonant frequency of 0.3 Hz with #maraging steel #springs (to reduce creep), which doesn’t sound that great to me, but whatever.
2001 #pdf #paper about Euler (buckled-column) springs for seismic isolation, also explaining #antisprings on p.3.
McIlroy #smallisbeautiful code for arithmetic on polynomials (or more precisely power series, i.e. there may be an unlimited number of coefficients) in ten lines of #Haskell.
"TensorFlow" is a high-throughput #dataflow array computing library and IDE with built-in reverse-mode #automatic-differentiation for optimization, with Python and C++ APIs and #IPython integration. At this moment in history, the growth of computer power has made a bunch of important #DSP and statistical tasks just feasible, so we are seeing things like self-driving cars, superhuman image recognition, and so on. But it’s been very difficult to take advantage of the available computational power, because it’s in the form of GPUs and clusters. So this is designed to make it easy to do exactly these things, and to scale them with your available computing power, along with libraries of the latest tricks in neural networks, machine learning (which is pretty close to "statistics").
the #tensorflow white #paper #pdf, describing its roots as a better replacement for DistBelief.
mentally ill person threatens #suicide by BART train, describes her suicide attempt last week with a drug overdose, and blames the US suicide epidemic on “radical self-reliance”, “individualism”, and “meritocracy”.
Why #Twitter is decaying due to context collapse.
TinyMOD, a 22 kilobyte #MOD #tracker. #smallisbeautiful
“open source data visualization and data analysis”, with box-and-wire visual programming.
dude, you can see which #microcontroller chips Digi-Key has the most in stock at the moment! Which is probably a pretty good proxy for which ones are the most popular. #1 is a surface-mount ATMega328(-MU) at the moment, as is #5, while the spots in between are taken up by Freescale 8-bit micros (MC9S08SG32, of the HCS08 family, the successor to the 68HC08 and the latest in the lineage of the 6800, running at 40MHz, with a kilobyte of RAM and 16K of flash). #6 and #7 are 42-cent PIC10s, and then there’s an ATTiny85 and more #AVR micros until #14, which is a five-dollar MSP430. #hardware
unsurprisingly the most popular timer/oscillator #hardware is a 555, although from TI.
this is Digi-Key’s most popular (or at least most in stock) FPGA. It’s US$6.75, thirty thousand gates, a Microsemi (??) ProASIC3 (??)
This is by a significant factor the most-stocked op-amp #hardware at Digi-Key at the moment: for US$567 you get 2500 of them. 1.4MHz, surface-mount, 2.5 to 6 volts, rail-to-rail, milspec temp range, can deal with load impedances down to 32Ω (output up to 80mA). Suggested uses: sensor signal conditioning, headphones.
This is the most-stocked memory #hardware at Digi-Key, a 32-kilobit 400kHz I²C EEPROM, costing 48 cents, with 5-millisecond page write time. Supposedly this is for smart electric meters.
This is Digi-Key’s most-stocked DC-DC switching regulator, a 97%-efficient 1.6 amp variable-voltage two-channel step-down buck converter. “Dedicated to supply core and I/O voltages of new multimedia design in portable applications powered from 1-cell Li-ion or 3 cell Alkaline/NiCd/NiMH batteries.” Wastes 60 μA on no load, which is almost a joule an hour, so this is not suitable for super-low-power sleep modes.
Digi-Key’s #2 most-stocked #FPGA #hardware is this US$14 Lattice thing, which is super interesting because Project IceStorm is for Lattice FPGAs. However, I think this is not one of the FPGAs #IceStorm supports, which are the iCE40 family; this is part of the "ECP2" family, which are “optimized for data and control path bridge and interfacing”.
This ICE40LP1K-QN84 is I think the most-stocked-by-Digi-Key #FPGA supported by Project #IceStorm. It costs US$4.29, has 64 kibibits of RAM, and has 160 CLBs (LABs?) totaling 1280 LUTs