3D visualisation in notebook in couple years
Ondrej Certik
here is an interesting page about O3D API from Google:
http://code.google.com/apis/o3d/
watch the video there, they want each browser to be able to show
advanced 3D things. So I think that it's clear that
we should be able to interact with hermes completely within a browser
in couple years. And the hermes itself
can be running on a local computer, or remote on our server, that
doesn't make a difference.
Ondrej
Pavel Solin
some doubts even inside our group. But it is very important
to see the future before everyone else does, then it will be
late. Let's not get frustrated by technical details even though
most of the time they are very difficult and there is no
immediate benefit to see. I would like to ask everyone
to be supportive of this effort. Let's keep this goal in mind
with everything we do in Hermes.
For example Pavel's nice GUI will need to be translated to
become a web application. Of course this does not mean
that Pavel will do it, but we will have to do it somehow.
Best,
Pavel
Pavel Solin
University of Nevada, Reno
http://hpfem.math.unr.edu/people/pavel/
Hermes project: http://hpfem.org/
FemHub project: http://femhub.org/
Jakub Cerveny
> some doubts even inside our group. But it is very important
> to see the future before everyone else does, then it will be
> late. Let's not get frustrated by technical details even though
> most of the time they are very difficult and there is no
> immediate benefit to see. I would like to ask everyone
> to be supportive of this effort. Let's keep this goal in mind
> with everything we do in Hermes.
Perhaps you could share some thoughts on the potential benefits of
having a web-based solver, to motivate the group and show this is
worth the effort. Maybe I am too conservative, but to me the whole
endeavor seems like a lot of mundane computer-science drudgery with an
uncertain effect. Why would I want to run Hermes (or Autocad, or 3D
Studio) in the browser when it's simpler and faster to run it
directly? If you need to run stuff remotely, there are readily
available solutions for that (Remote Desktop, VNC, ...). Yes it's cool
to run Google Maps on an iPod but this does not necessarily generalize
to other applications.
This really reminds me of the Cloud Computing hype we are seeing
lately. Cloud Computing has a number of high-profile proponents
(Google, Amazon), but also a crowd of critics (headed by R. M.
Stallman, the founder of GNU). I personally hope the whole idea will
disappear with time, or end up like the rest of the technologies that
were once supposed to become the future of computing (touchscreens
were supposed to replace the mouse, voice recognition was supposed to
completely replace the keyboard, etc., etc.). I may be wrong.
Jakub
Pavel Solin
http://alamos.math.arizona.edu/~rychlik/JOde/index.html.
I am using it in the classroom to teach
ordinary differential equations, and so do
many other people. It is extremely popular.
We can achieve the same with Hermes.
Even though this is not our only goal,
it is an important one. I already proposed
to the NSF to create an interactive finite
element course in this way. It will be very
different from how finite elements are taught
today. If they decline it, I am sure we will get
the support from the Department of Education.
And engineers have many courses involving
PDE, such as electrical fields, heat transfer,
fluid dynamics, etc.
There are other obvious benefits of a web
interface such as being able to use large CPU
power remotely, not having to install the SW
and deal with different platforms, always have
the latest version, etc., but these have been
discussed before.
Pavel
Jakub Cerveny
> http://alamos.math.arizona.edu/~rychlik/JOde/index.html.
> I am using it in the classroom to teach
> ordinary differential equations, and so do
> many other people. It is extremely popular.
> We can achieve the same with Hermes.
Yes this is cute and I agree a simple web-based PDE solver would also
be nice, for education purposes. But I doubt this is useful for more
serious stuff like CFD, and for the development of the software. I
would go crazy if I had to use the browser to test the library (unless
the web interface was extremely well done, which is seldom the case).
I believe the option to run the software locally is still very
important.
Anyway, good luck to you all.
Jakub
Pavel Solin
I am glad to hear other opinions, of course things are
not B&W.
> But I doubt this is useful for more
> serious stuff like CFD
problem definition through a web-based GUI is not
the main problem. Obviously if someone is paid
full time to do high-level 3D CFD simulations in
a national lab or so, he will not do it through a web
browser. The web interface is meant for average users,
students, etc. Since I do not know what is possible,
I cannot say whether the development work in our
group could later go through the web interface as
well. That would be great but I simply do not know.
> I believe the option to run the software locally is still very
> important.
Pavel
Ondrej Certik
> Hi Jakub,
> I am glad to hear other opinions, of course things are
> not B&W.
>
>> But I doubt this is useful for more
>> serious stuff like CFD
>
> The difficulty of CFD is mainly in the equations. The
> problem definition through a web-based GUI is not
> the main problem. Obviously if someone is paid
> full time to do high-level 3D CFD simulations in
> a national lab or so, he will not do it through a web
> browser. The web interface is meant for average users,
> students, etc. Since I do not know what is possible,
> I cannot say whether the development work in our
> group could later go through the web interface as
> well. That would be great but I simply do not know.
>
>> I believe the option to run the software locally is still very
>> important.
>
> No one doubts that, at least for a few more years.
As I said in my first email, that option will be here forever --- you
start hermes on your laptop, the only question is how you interact
with it. If it's a QT based gui, then you have to install it on all
computers, something like a google earth. If you interact with a
browser, you get much wider user base and you can get hermes into
every classroom, so this is like google maps in the browser.
So as you can see, even Google does both, but you can ask yourself,
how often you use google earth and how often google maps in the
browser (at least I use almost always maps in the browser, because I
am lazy to install google earth over and over again).
Ondrej
Ondrej Certik
Also -- at the SIAM conference, I had all examples neatly prepared on
my linux laptop and when our presentation was due, we realised my
laptop doesn't work with the screen projector... So we had to quickly
move our slides to Shengtai's laptop, but what about hermes? If we
didn't have the notebook, we would be totally finished. Fortunately, I
just started Safari on Mac, connected to our server and the
presentation proceeded without a glitch and people in the audience got
an idea what can be done with hermes, when I showed them live in the
notebook.
Pavel will go to couple conferences soon, so that's why I think it's
very important to have hermes working in the notebook.
Ondrej
Brian Granger
the sage/spd notebook for both hermes and other things.
> Perhaps you could share some thoughts on the potential benefits of
> having a web-based solver, to motivate the group and show this is
> worth the effort. Maybe I am too conservative, but to me the whole
> endeavor seems like a lot of mundane computer-science drudgery with an
> uncertain effect. Why would I want to run Hermes (or Autocad, or 3D
> Studio) in the browser when it's simpler and faster to run it
> directly? If you need to run stuff remotely, there are readily
> available solutions for that (Remote Desktop, VNC, ...). Yes it's cool
> to run Google Maps on an iPod but this does not necessarily generalize
> to other applications.
* Web app doesn't imply remote. I use the Sage notebook regularly
running on my local machine.
* Most potential users of hermes (even ones who are brilliant
scientists) won't be able to install it. Period. Remember, most of
these users are not computer scientists, hackers or even applied
mathematicians. They are domain scientists (physics, chemistry,
etc.). For these users having something that is easy to install (SPD)
and easy to use (web notebook) makes all the difference in the world.
Many of them only use Mathematica or Matlab. Without something like
the web notebook, they won't touch hermes.
* Many potential users of hermes (professors like myself) spend a lot
of time teaching as well. Without a web based notebook, I wouldn't
even think about using hermes in a teaching context.
* Funding is very competitive. Because most funding agencies in the
US are mandated by congress to look at educational outcomes, the web
notebook increases the chances of getting hermes funded for years to
come.
* The web notebook provides a great way of presenting hermes based
things in talks and lectures.
* The web notebook opens the door for interesting new types of
collaboration. You can share notebooks.
* It is already written so even though it does involve "CS drudgery"
someone else already did it.
I agree that there are somethings (serious software development,
algorithm development, etc.) that are not appropriate for the
notebook. From this perspective, hermes *developers* might care less
about the notebook than hermes *users* - at least when it comes to
writing C++ code.
Cheers,
Brian