congoe

See this pubpeer discussion about critiques of plausibility of the papers conclusions: https://pubpeer.com/publications/155C1B85C0680A558D4431D059A...

Very excited about this! I used CedarDB a while back for my bioinformatics research when Clickhouse started running into OOMs, and the difference was astonishing (Some details: The main query was a join of 100 million rows against 30.000 rows, then aggregating statistics for the entire table. Clickhouse used up well over 64 GiB of RAM + 100 GiB of Swap and then crashed, while Cedar used only 15 GiBs and finished quickly).

Ever since attending the advanced database lecture of Thomas Neumann at TU Munich, who wrote Umbra, the academic predecessor of CedarDB, I've been looking forward to everyone getting their hands on it!

I believe Cedar will enable tons of projects and companies to simplify their data stack because it can basically handle everything at once (until reaching very significant scale).

Also discussed at https://www.youtube.com/watch?v=iMwepyyaj8I

The article is simply wrong:

> An MIT biotech researcher has been able to run the iconic computer game Doom using actual gut bacteria. Lauren Ramlan didn’t get the game going on a digital simulation of bacteria, but turned actual bacteria into pixels to display the 30-year-old FPS, as reported by Rock Paper Shotgun.

The cited Rock Paper Shotgun article doesn't support this though:

> It’s worth noting that Ramlan herself points out that “running” Doom using the cells would be an enormous undertaking due to their extremely limited ability. What she did manage to successfully do, however, is simulate using the wall of cells as a display for Doom by rendering gameplay using the illuminated E. coli.

The authors writeup can be found here (https://docs.google.com/document/d/1SFm1dS6myqq7psBKttP7CVYN...) and her explanatory video here (https://www.youtube.com/watch?v=8DnoOOgYxck). However, it appears to be only a simulation, not physically running Doom on E.coli

Parquet works well as it natively support Apache Arrow, the underlying data structure.

How could AlphaZero possibly play better chess than humans when it doesn’t even understand the history of chess theory?

RL doesn’t stop at human levels

The fundamental principle would be that enzymes have an comparative advantage when utilising and producing complex organic molecules as fuel for hydrogen production (such as glucose in wastewater) while inorganic catalysts are cheaper, more stable but limited to simple reactions

Yes, that drives cost down but even on the biggest scale, like insulin, enzymes are still orders of magnitude more expensive than inorganic catalysts (e.g. platinum for hydrogen fuel cells)

Enzymes are much too unstable and expensive to ever make sense in electricity production, especially when compared to inorganic solutions like PV.

The other way around is more sensible though, using electricity to produce hydrogen (Enzymatic electrolysis cells, EFCs), for example from distillery wastewater.

I got one by submitting an eval to there github repo. Took me about an hour.

You could have at least clicked on the LinkedIn link before assuming the author was a guy... https://www.linkedin.com/in/naomi-w-liu/

Moin Marcus,

as someone who had to make the same decision, I can strongly recommend you to take CS at TUM, the program is awesome (they just hosted the largest hackathon in Europe) and quite challenging. Let me know if you have any questions :)

So will Zeiss

That just sounds like it's for people about to die

I didn’t know that was possible, I’ll have a look at it :)

Lifespan by David Sinclair would be my recommendation if you want to energise someone for molecular biology/ biotech