e-meetup tomorrow? about klara.jl HOT 14 CLOSED

I'd be open to doing something like this, but need a lot more advance warning to make appointments (like a week at least).

from klara.jl.

True, that was very short notice. Let's postpone the group e-meeting till next week (@spencerlyon2 we can have a short informal chat tomorrow). I will send out a doodle poll later on today to see what time would suit most of us next week.

from klara.jl.

Sounds good to me. Thanks for organizing.

from klara.jl.

Works for me. Any time after 7am West Coast works.

Regards,
Rob J. Goedman
[email protected]

On Nov 4, 2014, at 10:14 AM, Spencer Lyon [email protected] wrote:

Sounds good to me. Thanks for organizing.

—
Reply to this email directly or view it on GitHub #73 (comment).

from klara.jl.

I am glad about the prospect of e-meeting up, it would be great to chat about Julia MCMC and to get in touch in real time. To give us some time to coordinate, I set a doodle poll here:

http://doodle.com/2ggw3zef5kpgcs9e#table

Doodle doesn't give the option of Atlantic time zone (i.e. New York time) and I found it partial to choose UTC (London) time. For this reason, I ended up choosing the east US time zone, which is -5 hours UTC, i.e. -1 hour New York time (@goedman, this covers the time lower bound you set I think).

from klara.jl.

@scidom ,

Would you mind adding Brian to the this list? I expect [email protected] mailto:[email protected] should work.

Regards,
Rob J. Goedman
[email protected]

On Nov 4, 2014, at 1:48 PM, Theodore Papamarkou [email protected] wrote:

I am glad with the prospect of e-meeting up, it would be great to chat about Julia MCMC and to get in touch in real time. To give us some time to coordinate, I set a doodle poll here:

http://doodle.com/2ggw3zef5kpgcs9e http://doodle.com/2ggw3zef5kpgcs9e
Doodle doesn't give the option of Atlantic time zone (i.e. New York time) and I found it partial to choose UTC (London) time. For this reason, I ended up choosing the east US time zone, which is -5 hours UTC, i.e. -1 hour New York time (@goedman https://github.com/goedman, I think this covers the time lower bound you set I think).

—
Reply to this email directly or view it on GitHub #73 (comment).

from klara.jl.

@scidom

Unless I am not seeing something properly, New York is UTC -5: http://time.is/facts/New_York.

from klara.jl.

Sure @goedman, add @brian-j-smith if he would like to join the conversation.

Ah, sorry, I got confused @spencerlyon2, you are right :) Ok, so the doodle is in New York time.

from klara.jl.

Hi all, based on the doodle poll, shall we have the google hangout on Wed 12th of Nov, 1pm New York time?

After further discussions with @spencerlyon2 we decided to postpone the Gibbs sampler elaboration for now; the focus has been put on adding the Gibbs code of @spencerlyon2 to MCMC.jl. From my side, I am trying to set up some sort of universal link btwn the Gibbs sampler and factor graphs (by turning deterministic factors-nodes to probabilistic ones using the dirac delta function). If this works it will make it unnecessary to reduce the factor graph, but I need to let this thinking develop for a couple more days.

Instead, let's have a short introductory meeting for 30 mins or so, to get to know each other and to discuss the possibility of future agenda items.

from klara.jl.

How does one join the Google Hangout - it's my first time. Thanks!

from klara.jl.

Hmm, same here. No clue.

Rob J. Goedman
[email protected]

On Nov 12, 2014, at 10:03 AM, Brian J Smith [email protected] wrote:

How does one join the Google Hangout - it's my first time. Thanks!

—
Reply to this email directly or view it on GitHub #73 (comment).

from klara.jl.

Follow this link:

https://plus.google.com/hangouts/_/gxc3iw7gzd5yqpbmeavtj4edaea

from klara.jl.

it was nice to meet you all today. To update this public issue, @spencerlyon2, @brian-j-smith, @goedman and I had a first Google Hangout today about the future of MCMC in Julia. It has been a promising meeting. We will be joining forces to combine Mamba.jl, Jugs.jl, Stan.jl and MCMC.jl in order to create a unique package that will bring together the best features of these existing packages.

The idea is to provide a generic MCMC engine in Julia (as generic as JuMP for those familiar with the latter) with an interpreter-compiler that will borrow ideas among else from the PGM.jl proposal. There will be a rather long incubation period allowing for compiling a wish list of features and for a joint software engineering effort.

I will follow up with some minutes and with making some e-space for this public endeavour.

from klara.jl.

I will now close this issue. We can give ourselves some time to think about model specification and we can have the next meeting in about 3 weeks' time.

from klara.jl.