--- Log opened Mon Mar 21 00:00:32 2022 00:37 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap 03:29 -!- Llamamoe [~Llamamoe@178235178105.dynamic-4-waw-k-1-2-0.vectranet.pl] has joined #hplusroadmap 03:39 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has joined #hplusroadmap 05:56 < lsneff> There’s gotta be a way to so distributed learning like that but with distributed computing. 05:57 < lsneff> I was looking at the wikipedia page for simulated annealing the other day and it’s a global process that simulated a physical process that only uses local information 05:57 < lsneff> There must be a way to formalizing these such that they only use local information 06:05 < Llamamoe> There exist papers doing more or less that 06:05 < Llamamoe> https://slideslive.com/38923276/putting-an-end-to-endtoend-gradientisolated-learning-of-representations 06:06 < Llamamoe> And there exist distributed architectures, too 06:33 < lsneff> super interesting 06:33 < lsneff> That vaguely reminds me of prediction coding 06:33 < lsneff> Which still doesn’t have an efficient method 06:38 < lsneff> Llamamoe: is there a paper that goes along with that talk? 06:43 < Llamamoe> lsneff: There is, same title, I thought it was linked there, sorry 06:49 < lsneff> .t https://arxiv.org/abs/1905.11786 06:49 < saxo> [1905.11786] Putting An End to End-to-End: Gradient-Isolated Learning of Representations 08:27 < nmz787_> lsneff: I would have been ready to pitch nGauge's AFM to a company I work part-time for.... but their product drivers are Windows only with no scripting interface, so I told them it wasn't worth my time and to let me know if their software side improves 08:31 -!- mirage335 [~mirage335@64.79.52.86] has quit [Quit: Client closed] 08:31 -!- mirage335 [~mirage335@64.79.52.86] has joined #hplusroadmap 09:04 < lsneff> nmz787_: wow, seems like they have some seriously bad internal management happening 09:05 < lsneff> If only they’d just sell me the tips 09:18 < lsneff> nmz787_: do you happen to have the email of the ceo or someone at that level? If I don’t get a reply back from the “Vice President Business Development & Global Sales”, I’d like to go a bit higher 09:28 < lsneff> Nvm, I’ll just find them on LinkedIn if it comes down to it 09:55 -!- mirage335 [~mirage335@64.79.52.86] has quit [Quit: Client closed] 09:55 -!- mirage335 [~mirage335@64.79.52.86] has joined #hplusroadmap 11:27 < nmz787_> lsneff: sounds like you were also talking to "Jim" 11:29 < lsneff> Yeah, it was Jim 11:30 < nmz787_> lsneff: here's his bullet point list of responses to some questions I asked: 11:30 < nmz787_> No scripting or API; Can’t auto stitch; Precision manual X-Y stage (1um sensitivity); Shipping to Internationally no problem (India, Singapore); No ITAR or other crazy import restrictions or tariffs; Yes, Gwyddion for analysis software…is open-source; Price ~$28K for the research tool including the instrument, laptop pre-loaded with operating and analysis software, accessories, etc…everything 11:30 < nmz787_> you need to be up and scanning in a few minutes; Tips are diamond-like carbon…very hard and long life (1,000’s of scans) 11:30 < nmz787_> For our current instruments, we do have an Education Version (n3) that would be about $8K cheaper. I’ve attached the spec differences. It can be upgraded at a later date. 11:32 < lsneff> They ship a laptop with the AFM?... gross 11:32 < lsneff> What a mess 11:35 < nmz787_> lol 11:35 < nmz787_> yeeeeaaaaahhhhh 11:35 < nmz787_> I just got a load of surplus lab equipment and it came with 5 or 6 desktop PCs 11:36 < nmz787_> 2 for the 2 EDS detectors, presumably one for the TOF mass spec, another for the DLTS spectrometer... I can't quite tell what they're all for 11:43 < lsneff> Wow, that's ridiculous 11:43 < lsneff> All this stuff needs to be open-sourrce 11:44 < lsneff> Do you think there's any chance I'll be able to convince them to sell the tips individually? 11:45 < nmz787_> I doubt it, but I guess it depends on how you frame it 11:45 < nmz787_> i.e. some student project or something 11:46 < nmz787_> you could ask the relevant college(s) admins if they have ever purchased one of the nGauge systems 11:46 < nmz787_> and then ask that prof to order some tips for you 11:46 < nmz787_> or give you an old one 11:46 < nmz787_> (presumably your engineering college and science college have different purchasing/procurement admins) 11:49 < nmz787_> but I would think they are scared about the "wrong" (right?) people getting ahold of the tips and then engineering a $2k mechanical system around them 11:49 < nmz787_> or reversing their micro/nano fab 11:50 < lsneff> Yeah, that's probably what they're thinking ultimately. It's just too bad 11:51 < lsneff> And they're kind of right. I need an AFM so I can design and build open-source photolithography equipment 11:52 < nmz787_> you mean to QA your developed structures? 11:52 < lsneff> Yeah 11:52 < nmz787_> or you want to do litho pen stuff 11:52 < nmz787_> well, you can just use your campus SEM 11:52 < nmz787_> for QA 11:52 < lsneff> No, for QA 11:52 < lsneff> I'm not going to be on campus by the time I get this up and running 11:52 < nmz787_> I also was thinking the nGauge for litho QA, but the price isn't justified vs the "difficulty" of sputter coating and SEM imaging 11:53 < nmz787_> if you're at PSU.edu, they work with external folks no problem 11:53 < nmz787_> why not just build the dan berard AFM? 11:53 < nmz787_> err, STM 11:53 < lsneff> I guess I can always find a nearby uni with an SEM, but the time lost by having to schedule things and move the wafers around would be annoying 11:53 < lsneff> That's what I'm ultimately going to do probabky 11:54 < nmz787_> you don't /need/ wafers 11:54 < lsneff> "wafers", more like small pieces of polysilicon 11:54 < nmz787_> I image chips and slivers of plastic 11:54 < nmz787_> usually just end up taping them to a wafer for something easy to load into the machine 11:54 < lsneff> I'm at psu now, but I'm moving in about 9 months to somewhere else 11:54 < lsneff> Not sure where yet 11:55 < nmz787_> lately I've been getting free SEM time 11:55 < nmz787_> just because the shop owner doesn't think my jobs are difficult/long enough to warrant charging me 11:55 < lsneff> lmao 11:56 < lsneff> If I can't figure something out by the time I build the machine in like a year or two, would you SEM image my etchings if I mail them to you? 11:56 < nmz787_> sure 11:56 < nmz787_> I definitely plan on having my cleanroom ready by then 11:56 < lsneff> Thanks 11:57 < nmz787_> the FIB is already close to being web-enabled 11:57 < nmz787_> there's the issue of segregating users from each others samples, since by web-enabled currently I just mean remote desktop 11:57 < nmz787_> ideally I could lock the stage to a given area or something 11:58 < nmz787_> then I could just bulk load samples from people and let them time-share the system remotely 11:58 < lsneff> That would be cool 11:59 < nmz787_> the SEM needs re-assembled, but I have all the parts, maybe except a single metal adapter plate which should be a breeze for this local guy with a CNC plasma table and mill 11:59 < nmz787_> that's all analog though, so connecting it to the web will be a bit of work 11:59 < nmz787_> but could be stupid like, an arduino controlling a motor which moves a physical switch 12:01 < lsneff> Is the sofware just windows? 12:01 < nmz787_> the FIB, yes 12:01 < lsneff> There's no way to spin up an actual web-interface? 12:01 < nmz787_> the SEM... it's all just analog controls and a CRT 12:01 < lsneff> Ahh, gottcha 12:01 < lsneff> lmao 12:01 < nmz787_> I have been reversing, slowly, the FIB software 12:02 < nmz787_> it's written in Delphi, which was something I've never used... so that alone was an architectural novelty 12:03 < lsneff> Delphi? 12:03 < lsneff> Wow 12:03 < lsneff> Have you seen the mochii SEM? https://mymochii.com/ 12:04 < nmz787_> yeah, in person in fact, at M&M (microscope and microanalsis conference) 12:05 < nmz787_> their resolution is pretty low, and the stage is terribly small with no way to swap to a larger stage (at least by design) 12:05 < nmz787_> the price tag is /almost/ low enough to be an "impulse" buy 12:07 < lsneff> How much was it? 12:11 < nmz787_> $65k a year or two, or three? ago 12:12 < nmz787_> lsneff: you might try building this https://doi.org/10.3938/JKPS.63.1287 12:12 < nmz787_> .title 12:12 < saxo> Design and fabrication of a scanning electron microscope (SEM) with an electrostatic column for process embedment | SpringerLink 12:12 < lsneff> Ah, interesting 12:13 < lsneff> I will look at that 12:13 < nmz787_> found this just now too, looks interesting https://arxiv.org/pdf/1504.06962.pdf 12:13 < nmz787_> Design and Implementation of a Micron-SizedElectron ColumnFabricated by Focused Ion Beam Milling 12:16 < lsneff> Thanks for the link 12:17 < lsneff> I imagine the project will ultimately involve designing AFMs, SEMs, and many types of machines 12:17 < lsneff> Because of how awfully the industry has handled the proliferation of nanofabrication tech, we're going to have to basically start from scratch 12:20 < nmz787_> well, there's at least a fair amount of surplus equipment to be had, which at least gets us past having to design ALL the pieces 12:21 < nmz787_> lsneff: I've worked with Michael Labella before https://www.mri.psu.edu/research-capabilities-categories/nanofabrication?page=1 12:21 < nmz787_> if you have any interest in getting trained on those heidelberg machines, I need to get some jobs run (paid jobs) 12:22 < nmz787_> was going to ask Michael about contracting them out 12:22 < lsneff> I wish I had time haha 12:22 < lsneff> I'm leaving psu this summer, so I won't be here 12:24 < nmz787_> ah, ok, I think I need it done in the next month or two 12:24 < nmz787_> have you visited https://abservices.psu.edu/lion-surplus 12:27 < lsneff> I have, but it’s mostly old computers and furniture 12:27 < lsneff> I didn’t see any lab equipment there last time I was there 12:27 < nmz787_> ah 12:35 < lsneff> nmz787_: since wafers need to be exposed like 100 times for all the layers, how do they position the shuttles in exactly the same place each time? 12:36 < lsneff> Is there some sort of fiducial that can be imaged with an SEM? 12:54 -!- mirage335 [~mirage335@64.79.52.86] has quit [Quit: Client closed] 12:54 -!- mirage335 [~mirage335@64.79.52.86] has joined #hplusroadmap 13:18 < kanzure> https://www.science.org/content/article/scientists-say-they-can-read-nearly-whole-genome-ivf-created-embryo 13:19 < kanzure> "Whole-genome risk prediction of common diseases in human preimplantation embryos" https://www.nature.com/articles/s41591-022-01735-0 14:00 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has quit [Remote host closed the connection] 14:10 -!- mirage335 [~mirage335@64.79.52.86] has quit [Quit: Client closed] 14:10 -!- mirage335 [~mirage335@64.79.52.86] has joined #hplusroadmap 15:07 -!- Codaraxis [~Codaraxis@user/codaraxis] has joined #hplusroadmap 15:20 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::bb3b] has joined #hplusroadmap 15:26 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 240 seconds] 16:22 -!- Malvolio is now known as Guest8260 16:22 -!- Guest8260 [~Malvolio@user/malvolio] has quit [Killed (erbium.libera.chat (Nickname regained by services))] 16:24 -!- Malvolio [~Malvolio@user/malvolio] has joined #hplusroadmap 16:37 < lsneff> It’s going to be interesting to build a reactive ion etcher 16:44 -!- Llamamoe [~Llamamoe@178235178105.dynamic-4-waw-k-1-2-0.vectranet.pl] has quit [Quit: Leaving.] 16:53 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::bb3b] has quit [Read error: Connection reset by peer] 16:53 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::bb3b] has joined #hplusroadmap 17:01 -!- mirage335 [~mirage335@64.79.52.86] has quit [Quit: Client closed] 17:01 -!- mirage335 [~mirage335@64.79.52.86] has joined #hplusroadmap 17:12 -!- Jay_Dugger [~jwd@47-185-200-188.dlls.tx.frontiernet.net] has joined #hplusroadmap 17:19 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 252 seconds] 17:43 -!- mirage335 [~mirage335@64.79.52.86] has quit [Quit: Client closed] 17:43 -!- mirage335 [~mirage335@64.79.52.86] has joined #hplusroadmap 18:08 -!- Croran [~Croran@71.231.214.173] has quit [Quit: leaving] 18:55 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::bb3b] has quit [Quit: Leaving] 19:10 -!- Jay_Dugger [~jwd@47-185-200-188.dlls.tx.frontiernet.net] has quit [Ping timeout: 240 seconds] 19:30 -!- ptrcmd [~ptrcmd@user/ptrcmd] has quit [Ping timeout: 250 seconds] 19:30 -!- ptrcmd [~ptrcmd@user/ptrcmd] has joined #hplusroadmap 23:53 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap --- Log closed Tue Mar 22 00:00:32 2022