[erlang-questions] Erlang meets physics

[Jared Kofron]

Hi Pablo-
My first project was on embedded hardware, and basically consisted of NIFs and Webmachine dispatching.
Pretty fun stuff.  

My current project is a little more involved, but is also pretty interesting:

Right now the bus over which communication takes place is abstracted away by having hardware modules
which translate API functions into their appropriate wire representation and then transmit those representations
over a handle that they have to the correct bus.  

In essence what I've done is taken instruments and buses and given them something like behaviors.  
Hardware can perform read,write,or configure, for example.  So if I want to read the center frequency of a sweeper,
I might say

hp8340b:read(high_frequency_sweeper,<<"cw_freq">>).

which the device module (in this case hp8340b) translates into the GPIB command "OPCW", and based on the
device address (which is governed by the atom in the first argument), dispatches it over a bus handle that it owns
in a state variable via bus:send_query/2.  

At the moment, we only communicate with things over GPIB via ethernet using prologix devices to do the translation
for us - basically glorified telnet - but it gets the job done.  Everything is in a very alpha stage right now for this project,
but it is working really nicely so far.

JK

On Mar 13, 2012, at 12:20 PM, Pablo Platt wrote:

> How do you interact with the hardware?
> Do you use GPIB C libr and wrap it with a NIF?
> 
> From: Joe Armstrong <erlang@REDACTED>
> To: Jared Kofron <jared.nance@REDACTED> 
> Cc: Erlang Questions <erlang-questions@REDACTED> 
> Sent: Tuesday, March 13, 2012 12:34 PM
> Subject: Re: [erlang-questions] Erlang meets physics
> 
> Great news - spread the word !
> 
> Just for the record Erlang programmers numbers 1 and 2 (ie myself and
> Robert Virding)
> are both ex physicists.
> 
> When I lecture I often point out the similarity between causality and
> message reception.
> You don't know that something has happened until you get a message
> telling that it has happened.
> 
> (In physics it's a ray of light, or a photon, or something -
> forgetting entanglement for the moment)
> 
> In computing it's the reception of a message.
> 
> As a ex physicist I know that we can't say anything about simultaneous
> events occurring
> at different places in space-time - turn this into computer science
> and the same arguments
> apply to things like making sure replicated data is consistent on
> remote sites - well you can't
> - at least if you want to change it - Brewer's CAP theorem applies -
> which for a physicist makes
> perfect sense.
> 
> Also as an ex physicist I realize that things do actually happen in
> parallel in the real world,
> so modelling them in a sequential programming language (if I wanted to do that)
> is big time crazy - just describe the parallel stuff in a concurrent
> language and the program
> writes itself. Wait a few years till we have million core computers
> and the parallel problems
> can be solved 1:1 on parallel computers - and programming simulations
> and so on will be
> really easy - but don't even think about doing it in a sequential language...
> 
> Cheers
> 
> /Joe
> 
> 
> On Mon, Mar 12, 2012 at 2:34 AM, Jared Kofron <jared.nance@REDACTED> wrote:
> > Hi All,
> > I've been using Erlang at work for a few years now, and I thought I'd throw my experience out there, as
> > my application is a little different than what you usually see on the list - I am a graduate student at the
> > Center for Nuclear Physics and Astrophysics at the University of Washington, and use Erlang extensively
> > in my work.
> >
> > In my experience, something that Erlang is really great at but doesn't receive much attention for these days
> > is managing and interacting with hardware.  In any physics experiment of even modest sizes, you wind up
> > having to keep track of the state of various pieces of equipment, often modify that state, and constantly
> > interrogate particular values.  For example, we might want to change the current in a magnetic trap, turn
> > that trap off altogether, or simply read back the voltage drop across our superconducting magnet.
> >
> > So far, I have deployed Erlang in this zone for two separate experiments (SNO+, a large particle physics
> > experiment in Canada) and Project 8 (a small nuclear physics experiment here in Seattle).  Both times have
> > been great successes, and I have found the reception of Erlang in this market to be great.  In general, what
> > I have done is wrap a hardware management layer with some kind of outside world interface. For SNO+, we
> > used Webmachine and RESTful control, and for Project 8 we actually conduct all communication
> > by using CouchDB as a message passing interface.
> >
> > Physicists are suspicious creatures, but once you demonstrate the feature set that you get for practically
> > free with OTP, they see the advantage pretty quickly.  On top of that, the development cycle for sophisticated
> > applications can be greatly reduced - more than once it made my group float to the top in terms of meeting
> > goals.
> >
> > In short, as far as I am concerned, Erlang has found a new niche in the world of Physics, and I intend to
> > spread the word as much as I can!
> >
> > Jared Kofron
> > _______________________________________________
> > erlang-questions mailing list
> > erlang-questions@REDACTED
> > http://erlang.org/mailman/listinfo/erlang-questions
> _______________________________________________
> erlang-questions mailing list
> erlang-questions@REDACTED
> http://erlang.org/mailman/listinfo/erlang-questions
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20120313/37cc0c7e/attachment.htm>