tcpekin

Does anyone have a keyboard suggestion that is multilanguage and has a comma on the top layer? German without a comma is, troublesome.

Satellites are also a big use case.

This is not a superalloy (i.e. turbine blade material), just a strong normal aluminum alloy that can be additively manufactured.

This is not a medium entropy alloy, it's a standard alloy in terms of the ratio of components, which forms medium entropy intermetallic precipitates which gives the alloy it's properties. Intermetallic MEA is an odd term I'm not really familiar with and would want to look into more, but is a little suspicious. Furthermore, while MEAs (3-4 equal primary components) and HEAs (5+ equal components) do have good mechanical properties, I'd be wary of the atomic size argument, last time I've been involved in it, that argument has increasingly been questioned, as the atomic size of the elements in question are generally pretty similar.

Feel free to reach out to the email above, I think it's one of the most interesting jobs I could have gotten coming out of a postdoc due to the mix of physics and real world, technically interesting problem solving.

I don't speak for the whole company, but in our project we are heavily Python based, Julia might be used by the central research team or something though.

Here are some example positions:

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/S...

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/Q...

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/H...

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/P...

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/S...

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/S...

https://zeissgroup.wd3.myworkdayjobs.com/External/job/Jena/S...

Carl Zeiss SMT/SMS | Jena/Oberkochen/Rossdorf & remote, Germany | Multiple Roles | Full-Time | Hybrid (some onsite required) | English, some German very helpful | https://www.zeiss.de/semiconductor-manufacturing-technology/...

Zeiss SMT is the leader in extreme engineering optics for semiconductor manufacturing. We come up every so often on HN as leaders in EUV optics manufacturing for ASML scanners. SMS is a branch of SMT concerned with standalone tooling for quality assurance and defect inspection in semiconductor fabs (tools like the AIMS EUV or PROVE systems, for example). We bring complete solutions to the hypercompetitive market that is the semiconductor industry. The scale is vast – the work you can be involved in ranges from simulations at the nanometer to how to ship tools via using Boeing 747(s).

Software (and extremely diligent/precise engineering… honestly it blows my mind regularly) underpins everything we do and spans the same vastness of scale. We (Zeiss SMT, but also SMS where I work) are hiring in pretty much every IT field as we modernize systems, upgrade legacy code, and build cutting-edge new tools and platforms to remain at the front of the field. Specifically interesting to HN (I think) is my team, which works on developing extremely fast physics simulation and data processing code utilizing classical algorithms, machine learning, and _AI_, all on premise. We also build and maintain the hardware (HPC clusters) to enable these developments.

Desirable skills Python (scientific, HPC, ML, physics simulation), image analysis (big, big data processing, very fast), CUDA, C++, HPC architecture design/on premise rack configuration, HPC admin, DevOps infrastructure (physical as well as CI/CD), GPGPU programming, FPGA programming, QA engineering, Kubernetes

I can post some job links tomorrow, but if you have questions please either comment or reach out to me at thomas dot pekin at zeiss dot com. I’ve been working here for a year and it’s a lot of fun.

Grad students at Cal cost a professor ~100k a year, and then leave after 2-5 years with any optimizations they might have personally made. They also only work 6-12 hours a day, and having been said grad student, get mind numbingly bored after about 10-15 repetitive syntheses, spending lots of time on them, when the (only) interesting part, is the XRD pattern at the end... I would have absolutely advocated for such an arm if I was still there.

The others might move wafers 24/7 but they don't have three shifts of R&D scientists and engineers working to continuously analyze the results/improve yield/performance 24/7. TSMC does.

I worked as a grad student in LBL/UC Berkeley for 5 years. At the time I didn't want to make a career of it, but if I had to go back to a national lab, that might be the one. The culture was department dependent, but at the Molecular Foundry was pretty good. The campus is gorgeous, it's not isolated like other national labs at all, pay is low compared to tech for the Bay Area, but not pennies, and the conversations and people you can meet are fantastic. Always lots of new faces with students and visitors, but the core group that I interacted with were all very kind, helpful, hardworking, and just fun to talk to about science! I can definitely recommend it.

I am 99% sure I know him - shares your same first initial? Tell him hello from me (TC Pekin), I always liked talking to him during some down time, and the feeling was mutual. As a grad student I always liked talking to him and hearing about all the history and his time at NCEM!

I've made a website that does very rudimentary simulation of transmission electron microscopy diffraction patterns, as a tool for myself and other researchers. Was a big leap for me learning how to do so, but it's been fun (and frustrating) to learn about Flask, Docker, cloud instances, HTTPS etc. along the way!

https://tcpekin.com/dp_sim/

I agree with this, my alcohol consumption has really tailed off after I started training for a marathon. It is impossible for me to actually train hungover and the required amount of training just so you can finish is 4-5x a week.

I'm a researcher, so I do move/access a lot of data locally (10s of Gbs, not Tbs usually). Lots of Jupyter/IPython interactive workflows. Starting to run some more web development stuff/Docker containers.

Mac M1 Air | 16 GB RAM | 1TB SSD | Purchased Dec 2021, used daily since

    === START OF SMART DATA SECTION ===
    SMART overall-health self-assessment test result: PASSED

    SMART/Health Information (NVMe Log 0x02)
    Critical Warning:                   0x00
    Temperature:                        35 Celsius
    Available Spare:                    100%
    Available Spare Threshold:          99%
    Percentage Used:                    4%
    Data Units Read:                    268,644,677 [137 TB]
    Data Units Written:                 259,969,878 [133 TB] 
    Host Read Commands:                 1,213,441,896
    Host Write Commands:                981,264,397
    Controller Busy Time:               0
    Power Cycles:                       125
    Power On Hours:                     793
    Unsafe Shutdowns:                   7
    Media and Data Integrity Errors:    0
    Error Information Log Entries:      0

    Read 1 entries from Error Information Log failed: GetLogPage failed: system=0x38, sub=0x0, code=745

Whoops sorry for the late response - somewhat yes. You can configure a Jupyter workflow to work like this, but I don't find that I, or many other people do, as it takes more discipline to not hop around cells.

One of the main differences for me is that the .py file is run in its entirety (outside of if/else blocks for loading data). That usually corresponds to multiple cells of a Jupyter notebook that one would need to Ctrl+Enter through, where missing one would cause a problem.

The second is just how you can decouple the code and the terminal - it's a personal pet peeve that Jupyter notebooks jump around when running through cells - I don't want to be scrolling all around just to reset some variables to their original values, and it's really nice to run a whole .py script and see an output side by side, where the script is much longer than my screen. I can keep it open at the important part in VSCode, and change some intermediate process, and let all of the ad hoc plotting code remain at the bottom.

Finally, the biggest difference for me is how figures behave - the way I have it set up is that they open in their own window and remain interactive (can zoom/pan). I know you can do it in Jupyter as well, but the workflow really emphasizes inline plotting with non-interactive plots, especially when it comes to sharing them. But with the .py script and IPython command line, I can open up 5 figures, tile them however I'd like, and then refer to them by name/number in my script, so I can clear and overwrite them however I'd like, and they don't close or move around. This makes comparing things very easy, like how changing a parameter changes the rest of my analysis.

Lastly - the way it is set up is more like Matlab... whoops, but I think their workflow is much more ergonomic than a notebook. However, for sharing with other people, I usually just copy and paste the various parts of my scripts into a notebook, as that is the de facto standard.

The way I get around this is to start an IPython interpreter, and run .py files with `run -i file1.py`. This loads things into memory in the interpreter, and then I can run file2.py with the actual analysis, and iterate with file2.py until I'm happy. In the end, you can keep the files separate, or combine them into 1 file that will run top to bottom your whole analysis. As long as you keep the IPython session open everything remains in memory, just like in a notebook. The autoreload magic also works if you set it to the correct option, so if you are working on a library/package it will automatically reload them if necessary.

Mind boggling pictures - really shows just a tiny part of the scale, and just how manual the whole process still is. I can't imagine the difficulty of trying to upgrade systems that are 30-50 years old and maintain compatibility.

Is that correct? I am a postdoc at Humboldt Universität zu Berlin and I'm at the TV-L 13 level, Stufe 3, which is ~3300 post tax.

This happened to me with a 8+ year old account. By no means was I a power user, maybe having only bought a few things over those 8 years. I go to sell a pair of shoes, the leather type being `Waxed Flesh` somewhere in the description, for like 250$, and was immediately banned for life on that account. CS refused to even acknowledge my existence pretty much after that.

I guess this is expected, but still disappointing. Unlimited upload was what first got me to buy the original Pixel. I have been using the compressed photos for a while to save on space since upgrading, but didn't really know exactly how compressed they were. Here is a comparison, and for my uses that seems pretty fine:

https://www.phonearena.com/news/Google-Photos-High-quality-v...

Congrats to Josh for a 7 figure payout - I just hope it remains as barrier-free as it does now.

Does anyone know who is behind this site? I couldn't find anything poking around, and for such a cool website to (in my mind) blow up so suddenly is very surprising, with very cool content as well. I'd be interested in following anyone who is doing this.

Why is this compiler optimization beneficial with -ffinite-math-only and -fno-signed-integers?

From

    if (x > y) {
      do_something();
    } else {
      do_something_else();
    }

    if (x <= y) {
      do_something_else();
    } else {
      do_something();
    }

The number of available valence shell electrons, not the atomic nuclei is typically the controlling factor for macroscale chemical properties of elements. This typically is in balance with the nucleus' positive charge, but does not have to be.

Water molecules are molecules, and therefore have a more complex electronic structure than atoms, and therefore is harder to describe, especially at the extremes.

Postdoc in Materials Science/Physics currently at Humboldt Universität zu Berlin. I guess it's tech adjacent, but I did my PhD at Berkeley which exposed me a lot to tech and therefore this site.

I see, I wasn't thinking about kernels much smaller than the image itself. So, if you don't use blocked FFT, it only makes sense to use FFT convolution when your kernel size is > log n?

I have been trying the beta and it's pretty decent. Still alternate between signal, telegram and whatsapp for different groups/people, but the beta definitely seems better that what was before.

https://faq.whatsapp.com/general/download-and-installation/a...

I think often the FFT is used to actually speed up convolution, it isn't slower than computing convolution of your kernel shifted to every pixel position.

https://en.wikipedia.org/wiki/Convolution#Fast_convolution_a...