--- Log opened Fri Aug 18 00:00:52 2017
00:45 -!- emeraldgreen [~user@92.222.68.248] has joined ##hplusroadmap
00:46 -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-qgxrgpacnsnemker] has joined ##hplusroadmap
00:51 -!- emeraldgreen [~user@92.222.68.248] has quit [Ping timeout: 240 seconds]
01:38 -!- Cory [~Cory@unaffiliated/cory] has quit [Read error: Connection reset by peer]
01:39 -!- CRM114 [~urchin@unaffiliated/urchin] has joined ##hplusroadmap
01:50 -!- jtimon [~quassel@143.29.134.37.dynamic.jazztel.es] has quit [Ping timeout: 240 seconds]
01:52 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has quit [Remote host closed the connection]
02:26 -!- Cory [~Cory@unaffiliated/cory] has joined ##hplusroadmap
02:30 -!- augur [~augur@198-27-215-123.static.sonic.net] has joined ##hplusroadmap
02:34 -!- augur [~augur@198-27-215-123.static.sonic.net] has quit [Remote host closed the connection]
02:35 -!- augur [~augur@198-27-215-123.static.sonic.net] has joined ##hplusroadmap
02:45 -!- augur [~augur@198-27-215-123.static.sonic.net] has quit [Remote host closed the connection]
03:54 < kanzure> "Comprehensive single-cell transcriptional profiling of a multicellular organism" http://science.sciencemag.org/content/357/6352/661
03:54 < fltrz> the TSV index was crap, even opencsv chokes on it, I got the sql file now, no more trying to guess the custom tsv format quirks
03:55 < kanzure> .tw https://twitter.com/neurowitz/status/897617040445505536
03:55 < yoleaux> @minikerri 4/ Clearly there is a deeper truth to nervous system function, given the near infinite functional reconfigurability of even 3 cell networks (@neurowitz, in reply to tw:897616501385154562)
03:57 < fltrz> hm the abstract does not explain how it covers multicellular organism
03:57 < kanzure> "Distinct neural circuits for the formation and retrieval of episodic memories" http://www.cell.com/cell/abstract/S0092-8674(17)30820-6
03:57 < kanzure>                                     sleeptyping disabled (~~ enjoy the rest of your sleep)
04:00 < fltrz> what is he referring to with "infinite functional reconfigurability" for 3 cells?
04:07 -!- wrldpcmbl [sid145438@gateway/web/irccloud.com/x-ounmfvskzfaxllke] has quit [Quit: Connection closed for inactivity]
04:42 < archels> yeah well if we're talking about biology then obviously those 3 cells are going to contain a forest of receptors and signaling pathways
04:43 < archels> if we're talking about 3 Hodgkin-Huxley neurons then infinite functional reconfigurability is a bit of an overstatement
04:45 -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-qgxrgpacnsnemker] has quit [Quit: Connection closed for inactivity]
04:45 -!- augur [~augur@198-27-215-123.static.sonic.net] has joined ##hplusroadmap
04:49 -!- augur [~augur@198-27-215-123.static.sonic.net] has quit [Ping timeout: 240 seconds]
05:04 -!- jtimon [~quassel@143.29.134.37.dynamic.jazztel.es] has joined ##hplusroadmap
05:36 -!- y0ur1 [~y0ur1@gateway/tor-sasl/y0ur1] has quit [Quit: y0ur1]
05:45 -!- CRM114 is now known as Urchin
06:09 -!- cluckj [~cluckj@static-173-59-27-112.phlapa.ftas.verizon.net] has joined ##hplusroadmap
06:45 -!- jtimon [~quassel@143.29.134.37.dynamic.jazztel.es] has quit [Remote host closed the connection]
06:49 < kanzure> biological neurons, yeah.
06:52 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
07:03 -!- y0no [~y0no@2001:bc8:212d:201:ff01::a] has quit [Quit: Bye]
07:04 -!- cluckj [~cluckj@static-173-59-27-112.phlapa.ftas.verizon.net] has quit [Read error: Connection reset by peer]
07:04 -!- cluckj [~cluckj@nat.philau.edu] has joined ##hplusroadmap
07:05 < kanzure> .tw https://twitter.com/pfau/status/849421400683229186
07:05 < yoleaux> @poolio I remain baffled that the machine learning field has simply decided it doesn't care about overfitting if something makes pretty pictures. (@pfau, in reply to tw:849124995896758276)
07:07 < kanzure> "Cell diversity and network dynamics in photosensitive human brain organoids" http://www.nature.com/nature/journal/v545/n7652/full/nature22047.html https://twitter.com/eboyden3/status/857296997233152016
07:12 < kanzure> .tw https://twitter.com/JennyBryan/status/898372123281858560
07:12 < yoleaux> Low-tech common sense about filenames. My slides may be homely, but this practical advice holds up well. http://www2.stat.duke.edu/~rcs46/lectures_2015/01-markdown-git/slides/naming-slides/naming-slides.pdf (@JennyBryan)
07:14 -!- jcluck [~cluckj@static-173-59-27-112.phlapa.ftas.verizon.net] has joined ##hplusroadmap
07:18 -!- cluckj [~cluckj@nat.philau.edu] has quit [Ping timeout: 248 seconds]
07:27 -!- sachy [~sachy@nat.brmlab.cz] has joined ##hplusroadmap
07:46 -!- augur [~augur@198-27-215-123.static.sonic.net] has joined ##hplusroadmap
07:50 -!- augur [~augur@198-27-215-123.static.sonic.net] has quit [Ping timeout: 240 seconds]
08:35 -!- Gurkenglas_ [~Gurkengla@dslb-178-000-218-087.178.000.pools.vodafone-ip.de] has quit [Ping timeout: 255 seconds]
09:03 -!- y0no [y0no@2001:bc8:212d:201:ff01::a] has joined ##hplusroadmap
09:06 < kanzure> robot abuse at boston dynamics https://www.youtube.com/watch?v=zkv-_LqTeQA
09:31 < mrdata> marching toward the robot wars
09:38 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap
09:41 -!- Storyteller [~Storytell@unaffiliated/storyteller] has left ##hplusroadmap []
09:57 -!- y0ur1 [~y0ur1@gateway/tor-sasl/y0ur1] has joined ##hplusroadmap
10:00 -!- y0ur1 [~y0ur1@gateway/tor-sasl/y0ur1] has quit [Remote host closed the connection]
10:03 -!- y0ur1 [~y0ur1@gateway/tor-sasl/y0ur1] has joined ##hplusroadmap
10:03 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Remote host closed the connection]
10:06 < kanzure> statistical renaming during javascript deobfuscation http://jsnice.org/
10:06 < kanzure> .title https://news.ycombinator.com/item?id=15043106
10:06 < yoleaux> Neural Nets Can Learn Function Type Signatures from Binaries [pdf] | Hacker News
10:39 -!- cluckj [~cluckj@static-173-59-27-112.phlapa.ftas.verizon.net] has joined ##hplusroadmap
10:42 -!- jcluck [~cluckj@static-173-59-27-112.phlapa.ftas.verizon.net] has quit [Ping timeout: 248 seconds]
10:43 < fenn> i'd have to disagree with "BRAFWTNEGASSAY" being a good filename
10:46 < fenn> it's like one of those "find the word boundaries" challenges
10:47 < kanzure> filename-examples-bad.txt
11:07 -!- Douhet [~Douhet@unaffiliated/douhet] has quit [Ping timeout: 240 seconds]
11:07 -!- Douhet [~Douhet@unaffiliated/douhet] has joined ##hplusroadmap
11:22 -!- Douhet [~Douhet@unaffiliated/douhet] has quit [Read error: Connection reset by peer]
11:23 -!- Douhet [~Douhet@unaffiliated/douhet] has joined ##hplusroadmap
11:46 -!- augur [~augur@198-27-215-123.static.sonic.net] has joined ##hplusroadmap
11:51 -!- augur [~augur@198-27-215-123.static.sonic.net] has quit [Ping timeout: 255 seconds]
11:57 -!- augur [~augur@198-27-215-123.static.sonic.net] has joined ##hplusroadmap
12:11 -!- augur [~augur@198-27-215-123.static.sonic.net] has quit [Remote host closed the connection]
12:54 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has joined ##hplusroadmap
12:56 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has quit [Remote host closed the connection]
12:57 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has joined ##hplusroadmap
13:21 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 240 seconds]
14:23 < nmz787> kanzure: re semisynbio, I posted about that 14 months ago too 2016-06-01.log:18:13
14:24 < nmz787> and seems we spoke a bit on it the next month
14:56 -!- y0no [y0no@2001:bc8:212d:201:ff01::a] has quit [Ping timeout: 255 seconds]
15:02 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap
15:07 -!- esmerelda [~mabel@174-24-228-29.tukw.qwest.net] has quit [Ping timeout: 240 seconds]
15:09 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 240 seconds]
15:11 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap
15:13 < kanzure> yeah
15:13 < kanzure> i was invited to join them
15:13 < kanzure> so that's why i was looking at their documents again
15:18 -!- jaboja [~jaboja@jaboja.pl] has quit [Remote host closed the connection]
15:32 -!- y0ur1 [~y0ur1@gateway/tor-sasl/y0ur1] has quit [Remote host closed the connection]
15:33 < kanzure> .tw https://twitter.com/hardmaru/status/898446369605664769
15:33 < yoleaux> Neural Architecture Search discover efficient ConvNets that require much fewer parameters and use less computation. https://arxiv.org/abs/1707.07012 https://pbs.twimg.com/media/DHfsLMJUIAA-TDe.jpg (@hardmaru)
15:35 -!- y0ur1 [~y0ur1@gateway/tor-sasl/y0ur1] has joined ##hplusroadmap
15:53 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
15:53 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Client Quit]
15:54 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
16:38 < nmz787> kanzure: neat
17:16 -!- c0rw1n [~c0rw1n@cpc109847-bagu17-2-0-cust223.1-3.cable.virginm.net] has quit [Remote host closed the connection]
17:17 -!- c0rw1n [~c0rw1n@cpc109847-bagu17-2-0-cust223.1-3.cable.virginm.net] has joined ##hplusroadmap
17:26 < nmz787> I have infiltrated semisynbio :)
17:27 < kanzure> intel?
17:30 < kanzure> "workshop results" https://www.src.org/program/grc/semisynbio/semisynbio-consortium-roadmap/6043-full-report-dna-based-storage-final-twg1.pdf
17:37 < nmz787> yeah
17:38 < nmz787> they are already a full member, so I have full access... their intel rep is local to me too
17:39 < nmz787> who is on the exec tech advisory board
17:51 < kanzure> uhh no idea. i've only seen one list of people names but i don't know what the list was.
17:51 < kanzure> anyway, going through the intel person was a good idea.
18:01 < nmz787> oh, sorry, that was a statment
18:01 < nmz787> that the intel person was on that board
18:07 < kanzure> hah okay.
18:17 < kanzure> max gigaprinter bandwidth is number of spots in a row that require the same chemical as the next step in their reaction. so worst case, it's none of them. best case it's all of them. and average case it's ... well it depends on how you lay out the spots on the surface.
18:17 < kanzure> and their sequences.
18:21 < fenn> if 1 billion nozzles is too many could you settle for 1 billion droplets per second?
18:21 < kanzure> "Development of high-throughput glass inkjet devices for pharmaceutical applications" http://onlinelibrary.wiley.com/doi/10.1002/jps.24192/full
18:21 < kanzure> "... glass capillary tubes were manufactured with orifice diameters of 5, 10 and 20 μm and were actuated with diaphragm piezoelectric disks. Also, a pressure capsule was formed by creating a manifold at a distance from the orifice tip. Placing the piezoelectric disk at 0.5 mm distance from the tip allowed the formation of a jet at 3.2 MHz in certain designs, but for a short period of time ...
18:21 < kanzure> ...because of overheating. The length of the pressure capsule, its inlet diameter, and the nozzle tip geometry were crucial to lower the required power. Actuating an inkjet device with 10 μm orifice diameter comfortably at 900 kHz and drying the droplets from 1% salbutamol sulphate solution allowed the formation of particles with diameters of 1.76 ± 0.15 μm and the geometric standard ...
18:22 < kanzure> ...deviation of 1.08"
18:22 < kanzure> overheating is very very solvable with cooling
18:22 < kanzure> i think 3 MHz is the highest rate i've seen so far for a single nozzle
18:23 < kanzure> yeah 1 billion drops per second is still useful, although i'd be surprised if >3k nozzles is impractical
19:36 < fltrz> "<kanzure> max gigaprinter bandwidth is number of spots in a row that require the same chemical as the next step in their reaction. so worst case, it's none of them. best case it's all of them. and average case it's ... well it depends on how you lay out the spots on the surface."
19:37 < fltrz> this reminds me of Gray codes
19:37 < fltrz> they are often used in experiments, for example suppose you have 8 binary parameters and you need to do an experiment with all settings
19:38 < fltrz> thats 256 runs, suppose that changing a binary parameter (say hot -> cold; or physically toggle a switch; etc) takes some time
19:38 < fltrz> then the naive approach of going through 0-255 and map each bit to a control would cause the lowest bit to be constantly toggled
19:39 < fltrz> with gray codes you can minimize that to only 1 bit needing a change per run, so in total only 255 changes
19:40 < fltrz> otherwise its 255+127+63+31+15+7+3+1=512 changes
19:43 < kanzure> huh how do you know cause->effect separation though?
19:43 < fltrz> you don't, just like you dont in the linear 0-255 run
19:44 < fltrz> they just give the same results (same experiments performed) with less changes between experiment
19:44 < fltrz> the 255 experiments are independent experiments, its a scheduling optimization
19:45 < kanzure> in the linear 0-255 run you know that you only toggled one bit and got one result.
19:47 < fltrz> no in the linear run you often toggle multiple bits, i.e. from setting "00111111" (63) to "01000000" (64) you had to change 7 of the 8 settings
19:47 < fltrz> some parameters for settings require time to change, i.e. temperature, or using a crane to move a block of lead in some physics experiment
19:48 < fltrz> so you want to minimize the number of changes between experiments
19:48 < kanzure> ok wait you mean you keep the previous state from the last experiment, and then change the bitmask
19:48 < kanzure> by a single bit.
19:49 < fltrz> yes, with gray codes you can minimize in this case to 1 bit or setting that was changed
19:49 < fltrz> so in total 255 changes
19:49 < kanzure> ok, and then the goal is to minimize the total number of bit flips, while still visiting all states or all bit flips or something.
19:49 < fltrz> while for naive binary progression, the least significant bit toggles between every experiment, the second las bit between every 2 experiments etc
19:50 < fltrz> yes, while still visiting all states
19:50 < kanzure> this is assume no correlation between different bits and different outcomes
19:50 < kanzure> who taught you debugging :P
19:51 < fltrz> imagine your printhead only dispersed 2 kinds of fluid, black and white, then ordering the sequences such in gray code only one nozzle would have to change state
19:51 < kanzure> in biology you end up in weird situations where merely your presence happened to alter the entire experiment (like because of nucleases flying off of your skin and hitting the experiment materials....)
19:52 < fltrz> yes, it assume no memory effects, if there are you need to clean up between experiments (both in the linear progression case, and in the gray code progression case)
19:53 < fltrz> in a biology experiment a fresh subject from the original non-experimented-on-population would be introduced between every run
19:53 < fltrz> so you can study the effect of 8 binary parameters on a fresh subject
19:53 < kanzure> you might appreciate this rant about debugging and engineering biology http://gnusha.org/logs/html/2016-09-08.log.html#t19:11-111
19:54 < kanzure> "... there is no source, the bytecode has multiple reentrent abstractions, is unstable and has a very low signal to noise ratio, the runtime is unbootstrappable, the execution is nondeterministic, it tries to randomly integrate and execute code from other computers... multiple reentrant and self-modifying abstractions. absolutely everything has subtle side effects."
19:54 < fltrz> I've been trying to solve mysql 'datadir' problem all day long, it just refuses to use a different directory for storing my databases
19:55 < kanzure> alright well first don't use mysql. for small projects use sqlite. for anything else then use postgresql.
19:56 < fltrz> its a bit like debugging an analog circuit (programmers tools for debugging like debugger are not applicable)
19:56 < fltrz> only the circuit is huge in schematic form, but tiny in physical form
19:57 < fltrz> and we don't start out with any schematics
19:57 < fltrz> and its full of johnson noise
19:57 < fltrz> and we dont understand most of the components
19:58 < fltrz> kanzure: its for the sqldump of scimag
19:59 < fltrz> its huge, and I don't have space on internal drive, and mysql on debian jessie is retentive about its datadir
20:00 < kanzure> with sqlalchemy + sqlite you can do really rapid prototyping of things https://gist.github.com/alvesjnr/1005481
20:01 < kanzure> also you can safely split up your data set and only load up a portion of the data at the time (since it's all separate rows anyway
20:01 < kanzure> )
20:01 < fltrz> I actually don't even need the sql functionality, just a linear parse and extracting the columns I need, opencsv choked on the TSV format
20:02 < kanzure> use python generator to read each line using the csv module. it will not read the entire file at once, just line by line.
20:02 < kanzure> like https://stackoverflow.com/a/17444799/687783
20:03 < kanzure> https://docs.python.org/3.7/library/csv.html
20:09 < fltrz> hm, I don't feel like trying another CSV library, I tried opencsv yesterday, .. some of the 65M titles contain weird characters and opencsv chokes at 28% of the db
20:15 < kanzure> while trying to find that rant link i ran into this old thing:
20:15 < kanzure> .tw https://twitter.com/QuantumG/status/43472344756387840
20:15 < yoleaux> haha "Andrew Hessel is the Justin Bieber of synthetic biology" http://www.youtube.com/watch?v=bRbCIduMGb0 (via @kanzure) (@QuantumG)
21:56 -!- Urchin [~urchin@unaffiliated/urchin] has quit [Ping timeout: 258 seconds]
22:23 -!- hehelleshin [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has joined ##hplusroadmap
22:27 -!- Guest7910 [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has quit [Ping timeout: 240 seconds]
22:30 < nmz787> kanzure: I really want that text-to-images neural network, along with a really good text-to-speach program running in-sync... I'd have tons of articles to
22:30 < nmz787> "watch"
22:49 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 240 seconds]
22:51 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
22:55 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Client Quit]
22:55 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
23:52 -!- esmerelda [~mabel@174-24-228-29.tukw.qwest.net] has joined ##hplusroadmap
23:52 -!- esmerelda [~mabel@174-24-228-29.tukw.qwest.net] has quit [Changing host]
23:52 -!- esmerelda [~mabel@unaffiliated/jacco] has joined ##hplusroadmap
23:54 -!- cevi_ [~zeb@pool-74-103-254-184.prvdri.fios.verizon.net] has joined ##hplusroadmap
--- Log closed Sat Aug 19 00:00:53 2017