ogrisel

Lime production is carbon intensive (heat + chemical C02 release from the limestone) and the hempcrete curing will take 50 years to fix back the second part. So it can be carbon negative (by fixing the carbon of the hemp itself) but only on the long term (50 years after production) and assuming that the heat used for lime production was from non-fossil source in the first place (which is not the case today: the cement industry typically produces heat by burning natural gas).

The heat could be produced from burning green hydrogen (hydrogen produced by water electrolysis using renewable power) but as of today is still too expensive compared to using fossil fuels.

> A bit like sticking a magnet on an engine block and getting loads more horsepower !

Typical motors used in electric cars are a lot more powerful and efficient than their internal combustion engine counterparts ;)

The BLAS & Lapack subset of the API of the Intel Math Kernel Library (MKL) is very well implemented in open source projects such as OpenBLAS and BLIS:

https://github.com/flame/blis

Both are well optimized for AMD CPUs.

The blog post highlights specifically usability improvements so they do pay attention to the usability.

Rather that state vague generic complaints, please be more constructive and be specific by highlighting the most important usability issues you encountered in your own experience and see if other users agree with your analysis by reporting it to the issue tracker.

The gallery of rendered images and videos is probably the quickest way of getting an idea of what this software project can achieve:

http://taichi.graphics/gallery/

Li-ion batteries are mostly useful to shift the load by a couple of hours, that is to be able to sell cheap to produce solar power in the evening when the demand is highest and therefore the gross price for electricity the highest.

So just a few hours worth of storage are actually really useful to integrate a significantly higher amount of renewables in the power mix.

Plastic micro-particles are found in tap water in most places around the world:

https://www.theguardian.com/environment/2017/sep/06/plastic-...

If I remember correctly, I believe the JetBrains / IntelliJ / PyCharm brands of the sponsor were very visible when taking the survey.

That might have influenced significantly in the selection of the Python users taking the survey: why answer a survey setup and sponsored by a private company you are not really familiar with and that you don't necessarily trust as a result?

There is also a second selection bias: the people who follow the PSF and the official python.org communication channels are probably nerdier than the average Python programmer. This is reflected in the OS statistics. The survey results suggests than 53% of the Python users do not use Windows at all.

On the scikit-learn online documentation, after removing the mobile traffic we have: Windows at 61%, macOS at 23% and Linux at 15%.

On stackoverflow, the primary OS statistics are: Windows at 47%, macOS at 27% and Linux at 23%.

Panel cleaning can be done by robots, no?

Putting a blog post would already be a positive contribution. Then reference it a bit, for instance by answering relevant questions on stackoverflow and linking to it to give more comprehensive details.

Over time your page will naturally get referenced by Google and other search engines.

Too bad it's not accepted in Wikipedia, but as long as the information is easily findable and organized as an easy to read and comprehensive enough reference, your work will be useful to many.

It's published in Nature Machine Intelligence:

https://www.nature.com/articles/s42256-018-0002-3

But at least this article seems to be available as Open Access. I am not sure if this is temporary or not.

There are other alternative technologies to store energy as heat, in particular molten silicon:

- https://reneweconomy.com.au/mit-team-gains-ground-on-molten-...

- https://1414degrees.com.au/ (which looks like a competitor for malta)

I believe the molten salt technology used for thermal energy storage is not directly related to molten salt nuclear reactors:

https://en.wikipedia.org/wiki/Thermal_energy_storage#Molten_... (sodium nitrate, potassium nitrate and calcium nitrate)

https://en.wikipedia.org/wiki/Molten_salt_reactor#Fused_salt...

Are you sure sunamp is molten salt? To me, molten salt is to store energy at a very high temperature in insulated containers.

I thought sunamp is some kind of room temperature phase change material that does not require heavy insulation.

The public is also afraid of nuclear technology because of nuclear weapons used to destroy the cities Hiroshima and Nagasaki with a single bomb each time.

Maybe, but why would EV owners would ever shop at a gas station if they can charge their car at home, work or while shopping at other regular shopping venues.

We will have to find a way to stop using fossil fuels because of climate change.

And Ionic Materials (http://ionicmaterials.com) which also received investment from big industrial companies over the last year.

You can do k-NN in the feature space of the network (activation of one of the last layers) to generate partial "explanations" of the neural network decisions.

You can write very efficient Cython code but it's true that in this case, you tend to adopt a lower level code style that is very close to C/C++. Basically, you need to think about the C/C++ code that will be generated by Cython.

C/C++ compilers might be able to generate more optimized native code than what rutsc does though. Actually, this is a question: how good is rustc with numerical / math intensive code? For instance, does it implements loop unrolling and SIMD vectorization?

I was confused. This is a hosted colaboraty service.

But the new google drive / jupyter integration is called jupyter-drive:

https://github.com/jupyter/jupyter-drive

It's basically a demo of their native code sandbox technology: (P)NaCL

https://developer.chrome.com/native-client

This new online notebook system is probably the successor of:

http://colaboratory.jupyter.org/welcome/

About manicuring a public image and fake niceness, for some reason, I find the following funny take on his public image quite authentic :)

https://twitter.com/BasedDrWorm/status/913282180280934400

Breaking with regen makes EV use the physical breaks less often as compared to combustion engine vehicles. Using breaks also emits particulates although I am not sure how it compares to the volumes of particulate emissions by combustion engines and tires abrasion.

Furthermore, adding more EVs on the roads means more EVs on home charging stations at night and when parked at work during the day. In turn this means more opportunities to develop large scale demand side management for the power grid and therefore make it easier / cheaper to integrate a larger fraction of cheap and low carbon but intermittent renewable energies such as wind and solar.

Thanks for your feedback. I am glad I could help you.

The pickling implementation of joblib has support for memory mapping numpy arrays nested in arbitrary data structures such as pandas dataframes.

Save the dataframe in a folder that can be accessed by the gunicorn worker:

    import joblib
    joblib.dump(df, '/folder/shared_data.pkl')

Then in the code run by the flask / gunicorn workers themselves:

    import joblib
    shared_df = joblib.load('/folder/shared_data.pkl', mmap_mode='r')
    # use the shared_df as usual (inplace modifications are not
    # authorized)

Some pandas function can have issues with read-only buffer though: https://github.com/pandas-dev/pandas/issues/17192 (caused by a currently unsolved bug / limitation of Cython) but it can work for your use case.

In the benchmarks reported in the article, both through-put and latency were significantly improved by the new thread pool system.

To be more factual: a majority of papers published by researchers at DeepMind do not cite any result from the neuroscience literature in their bibliography. Instead they cite other papers from the Machine Learning community.

While Demis Hassabis and probably some DeepMind researchers use some understandings from the neuroscience literature as an inspiration for their work, I am pretty sure that a majority of DeepMind researchers would rather use a combination of mathematics and trial and error experiments to build an intuition to guide the design of their next iteration of intelligent learning systems.