Friam

Is Robert a Deterministic System?

Classic

List

Threaded

13 messages Options

Nick Thompson

Is Robert a Deterministic System?

Al --

I found this so interesting, I wanted to pull it out and put it at the head of my message, where it wont get lost in the dross at the end of my message. (By the way, is there anyway to deal with that problem? I often can barely FIND then new text amongst all the quotations.) Reading this passage, I am struck again with the fact that, even though I mucked around with you guys for a better part of a year, I never thought to ask these questions. Anyway, here it is.

Nope, I really do mean deterministic. First up, you can make a complex
system out of a bunch of deterministic relations but that doesn't mean you
have to. You can make a complex system out of stochastic relations as well.
Secondly, you don't always get a sensitivity with respect to initial
conditions. For example the logistic equation x <- kx(1-x) is only chaotic
for k > 3.57. Below that it has fixed point attractors or shows
period-doubling.

So what does this have to do with my computer? Well I'm not sure that my
computer is a complex system, but even if it is when I'm just playing around
with it (like now) it's close enough to an attractor that it gives
consistent answers to the question of (say) what is your CPU loading. It
doesn't matter whether I opened Word first or Firefox, my CPU still says
that it's running around 10%. (Actually that's a pretty convincing proof
that a my PC's CPU performance isn't well-described as a complex system. It
doesn't display much sensitivity to initial conditions at all.) Anyway, to a
pretty good approximation it's a (non-complex) deterministic system.

And I'm not. The answer to the question "how are you?" is highly stochastic
and is affected (but not determined) by my temporal distance from my last
coffee, how well I slept last night, who I've just seen on CNN, whether that
irritating pain in my back is still there etc etc.

So, I still want to know.... I didnt get where I am today without asking the same stupid question over and over again -- when you ask your cpu howzit performing, what actually gets measured?

But a more important question raised for me by what you say here has to do with complexity: having heard it defined at the SFISS as sensitivity to initial conditions, I am not sure where I stand in the world any more. Owen? Steve?

Also, that message from Giles in that funny misspelled french looked pretty interesting. I think the dependency of such questions on language actually makes the case for the simularity of computers with people in the sense that it is true of both our relationships with computers AND our relatinships with people that there has to be a series of conventions in place before anybody can talk to anything or anybody. And as to Robert's last point, wouldnt a sloppily built computer be a better model by which to explore self awareness than a well engineered one?

You all raise deeper questions than I am prepared for. I just wanted to know what the Crocodile has for dinner, so to speak.

Nick

Nicholas Thompson
nickthompson at earthlink.net
http://home.earthlink.net/~nickthompson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://redfish.com/pipermail/friam_redfish.com/attachments/20051129/76accb65/attachment.htm

Stephen Guerin

Self-knowledge

> when you ask your cpu howzit performing, what actually gets measured?

Nick, are folks reading more into what you're really asking? When you ask the
question "Hey CPU, how are you performing?" Do you just mean you're looking at
the Performance tab in the Windows Task Manager (I know you're on windows) and
wonder what it's measuring? The primary top window shows how busy the CPU is as
a time series graph. The percentage measure is how many cycles per second your
computer is being used to compute by an application divided by the number of
cylces per second it is idle.

> But a more important question raised for me by what you say here has to do
with complexity:
> having heard it defined at the SFISS as sensitivity to initial conditions, I
am not sure
> where I stand in the world any more. Owen? Steve?

Sensitivity to initial conditions is a description of deterministic chaos not
necessarily complexity. Complex systems can be insensitive to initial conditions
if they are in a stable attractor -- different initial conditions would get
wiped out as the system iterates. That said, Complex systems are sensitive to
initial conditions near phase transitions.

-S

Roger Critchlow-2

Is Robert a Deterministic System?

In reply to this post by Nick Thompson

On 11/29/05, Nicholas Thompson <nickthompson at earthlink.net> wrote:

>
> So, I still want to know.... I didnt get where I am today without asking the
> same stupid question over and over again -- when you ask your cpu howzit
> performing, what actually gets measured?
>

Your computer has a list of processes to run. Some of these are
system services and some are user processes. If you type ctrl-alt-del
on Windows, you'll get the Windows Task Manager (assuming your
computer isn't currently locked up) and under the Processes tab you'll
get the list of processes.

Under the column labelled CPU you get the percent of the CPU time
which each process is using. Click on the column label a few times
and you'll get the list of processes sorted in descending order of CPU
usage. The list will resort itself every second or two as the usages
change.

Most of the processes are using 0% of the CPU. This is because
they're sleeping, waiting for user input or for a service request from
elsewhere. When the event they're waiting on occurs, they will wake
up, service the event, consume some CPU cycles, and then go back to
sleep.

Near the top of the CPU usage sorted list there will be a process with
the Image Name of "System Idle Process". This is a process which does
nothing. It gets scheduled to run whenever there's nothing else to
do. It's currently consuming about 80-90% of the CPU time on my
computer, so my computer is running a 10-20% capacity load.

How does the computer know this? There's another process near the top
of the list with the Image Name "taskmgr.exe". It's running 0-5% of
the CPU on my machine. This process is reading the list of processes
to run and deciding which process to run next. When it schedules a
process to run it notes that the process was allocated a time slice.
When the running process yields the CPU, either by going to sleep
waiting for an event or because its time slice finished, the taskmgr
will note how much of the time slice was actually used. Add up all the
time slices used over a second and you get 100% of the CPU usage, less
the context switch and interrupt service overhead. So the computer
doesn't know what it's doing, but it knows what it did over the last
1000 time slices and gives you a statistical summary of that activity.

That's the first order story. The second order story gets into
allocating memory for the process, filling the memory with the right
bits so the process can run its program on its data, and saving the
memory bits back to more permanent storage. This happens at several
levels, each of which has pathologies.

-- rec --

Luciano Oviedo

Self-knowledge

In reply to this post by Stephen Guerin

I second that question - what are you really asking? Is it even a question?
;-)

Luciano

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf
Of Stephen Guerin
Sent: Tuesday, November 29, 2005 9:37 AM
To: nickthompson at earthlink.net; The Friday Morning Applied Complexity Coffee
Group
Subject: Re: [FRIAM] Self-knowledge

> when you ask your cpu howzit performing, what actually gets measured?

Nick, are folks reading more into what you're really asking? When you ask
the
question "Hey CPU, how are you performing?" Do you just mean you're looking
at
the Performance tab in the Windows Task Manager (I know you're on windows)
and
wonder what it's measuring? The primary top window shows how busy the CPU is
as
a time series graph. The percentage measure is how many cycles per second
your
computer is being used to compute by an application divided by the number of
cylces per second it is idle.

> But a more important question raised for me by what you say here has to do
with complexity:
> having heard it defined at the SFISS as sensitivity to initial conditions,
I
am not sure
> where I stand in the world any more. Owen? Steve?

Sensitivity to initial conditions is a description of deterministic chaos
not
necessarily complexity. Complex systems can be insensitive to initial
conditions
if they are in a stable attractor -- different initial conditions would get
wiped out as the system iterates. That said, Complex systems are sensitive
to
initial conditions near phase transitions.

-S

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at Mission Cafe
lectures, archives, unsubscribe, maps at http://www.friam.org

Stephen Guerin

Self-knowledge

In reply to this post by Stephen Guerin

oops, i gave the wrong ratio...I should have said the number of non-idle cycles
divided by the total available cycles.

> -----Original Message-----
> From: Stephen Guerin [mailto:stephen.guerin at redfish.com]
> Sent: Tuesday, November 29, 2005 9:37 AM
> To: nickthompson at earthlink.net; The Friday Morning Applied Complexity
> Coffee Group
> Subject: Re: [FRIAM] Self-knowledge
>
>
> > when you ask your cpu howzit performing, what actually gets measured?
>
> Nick, are folks reading more into what you're really asking? When you ask the
> question "Hey CPU, how are you performing?" Do you just mean you're looking at
> the Performance tab in the Windows Task Manager (I know you're on windows) and
> wonder what it's measuring? The primary top window shows how busy the
> CPU is as
> a time series graph. The percentage measure is how many cycles per second your
> computer is being used to compute by an application divided by the number of
> cylces per second it is idle.
>
> > But a more important question raised for me by what you say here has to do
> with complexity:
> > having heard it defined at the SFISS as sensitivity to initial conditions, I
> am not sure
> > where I stand in the world any more. Owen? Steve?
>
> Sensitivity to initial conditions is a description of deterministic chaos not
> necessarily complexity. Complex systems can be insensitive to initial
> conditions
> if they are in a stable attractor -- different initial conditions would get
> wiped out as the system iterates. That said, Complex systems are sensitive to
> initial conditions near phase transitions.
>
> -S
>
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at Mission Cafe
> lectures, archives, unsubscribe, maps at http://www.friam.org
>
>

Giles Bowkett

Is Robert a Deterministic System?

In reply to this post by Nick Thompson

> Also, that message from Giles in that funny misspelled french looked pretty
> interesting. I think the dependency of such questions on language actually

fran?ais?! sacre bleu! mon cire d'oreille n'est pas fran?ais! c'est fromage!

--
Giles Bowkett = Giles Goat Boy
http://www.gilesgoatboy.org/

Russell Standish

Is Robert a Deterministic System?

Ah bien, une langue que je peux comprendre. Particulierement a propos
parler des sujets de demander a la Lune, et de la Lune enceinte. Qui
dit la Lune est enceinte? Pas toi je pense - seulement une erreur de
traduction.

Hasta la vista, baby!

On Tue, Nov 29, 2005 at 11:11:57AM -0700, Giles Bowkett wrote:

> > Also, that message from Giles in that funny misspelled french looked pretty
> > interesting. I think the dependency of such questions on language actually
>
> fran?ais?! sacre bleu! mon cire d'oreille n'est pas fran?ais! c'est fromage!
>
>
> --
> Giles Bowkett = Giles Goat Boy
> http://www.gilesgoatboy.org/
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at Mission Cafe
> lectures, archives, unsubscribe, maps at http://www.friam.org

--
*PS: A number of people ask me about the attachment to my email, which
is of type "application/pgp-signature". Don't worry, it is not a
virus. It is an electronic signature, that may be used to verify this
email came from me if you have PGP or GPG installed. Otherwise, you
may safely ignore this attachment.

----------------------------------------------------------------------------
A/Prof Russell Standish Phone 8308 3119 (mobile)
Mathematics 0425 253119 (")
UNSW SYDNEY 2052 R.Standish at unsw.edu.au
Australia http://parallel.hpc.unsw.edu.au/rks
International prefix +612, Interstate prefix 02
----------------------------------------------------------------------------

Russell Standish

Self-knowledge

In reply to this post by Stephen Guerin

On Tue, Nov 29, 2005 at 09:37:25AM -0700, Stephen Guerin wrote:
> > when you ask your cpu howzit performing, what actually gets measured?
>
> Nick, are folks reading more into what you're really asking? When you ask the
> question "Hey CPU, how are you performing?" Do you just mean you're looking at
> the Performance tab in the Windows Task Manager (I know you're on windows) and
> wonder what it's measuring? The primary top window shows how busy the CPU is as

Nick is that what you're asking? Why didn't you say so? The correct
term is asking your computer for its "load average". Here's how you
ask mine:

mpi>top -b -n1
top - 10:11:10 up 1:16, 3 users, load average: 1.05, 1.09, 0.81
Tasks: 52 total, 2 running, 49 sleeping, 1 stopped, 0 zombie
Cpu(s): 25.7% user, 4.3% system, 0.0% nice, 70.0% idle
Mem: 223660k total, 219464k used, 4196k free, 2556k buffers
Swap: 393584k total, 18908k used, 374676k free, 32696k cached

The load average on mu computer is currently 1.05 or 105% to you. I'm
running a simulation in the background.

Cheers

--
*PS: A number of people ask me about the attachment to my email, which
is of type "application/pgp-signature". Don't worry, it is not a
virus. It is an electronic signature, that may be used to verify this
email came from me if you have PGP or GPG installed. Otherwise, you
may safely ignore this attachment.

----------------------------------------------------------------------------
A/Prof Russell Standish Phone 8308 3119 (mobile)
Mathematics 0425 253119 (")
UNSW SYDNEY 2052 R.Standish at unsw.edu.au
Australia http://parallel.hpc.unsw.edu.au/rks
International prefix +612, Interstate prefix 02
----------------------------------------------------------------------------

Giles Bowkett

Is Robert a Deterministic System?

In reply to this post by Russell Standish

Mais c'?tait une question enceinte, non?

On 11/29/05, Russell Standish <r.standish at unsw.edu.au> wrote:

> Ah bien, une langue que je peux comprendre. Particulierement a propos
> parler des sujets de demander a la Lune, et de la Lune enceinte. Qui
> dit la Lune est enceinte? Pas toi je pense - seulement une erreur de
> traduction.
>
> Hasta la vista, baby!
>
> On Tue, Nov 29, 2005 at 11:11:57AM -0700, Giles Bowkett wrote:
> > > Also, that message from Giles in that funny misspelled french looked pretty
> > > interesting. I think the dependency of such questions on language actually
> >
> > fran?ais?! sacre bleu! mon cire d'oreille n'est pas fran?ais! c'est fromage!
> >
> >
> > --
> > Giles Bowkett = Giles Goat Boy
> > http://www.gilesgoatboy.org/
> >
> > ============================================================
> > FRIAM Applied Complexity Group listserv
> > Meets Fridays 9a-11:30 at Mission Cafe
> > lectures, archives, unsubscribe, maps at http://www.friam.org
>
> --
> *PS: A number of people ask me about the attachment to my email, which
> is of type "application/pgp-signature". Don't worry, it is not a
> virus. It is an electronic signature, that may be used to verify this
> email came from me if you have PGP or GPG installed. Otherwise, you
> may safely ignore this attachment.
>
> ----------------------------------------------------------------------------
> A/Prof Russell Standish Phone 8308 3119 (mobile)
> Mathematics 0425 253119 (")
> UNSW SYDNEY 2052 R.Standish at unsw.edu.au
> Australia http://parallel.hpc.unsw.edu.au/rks
> International prefix +612, Interstate prefix 02
> ----------------------------------------------------------------------------
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at Mission Cafe
> lectures, archives, unsubscribe, maps at http://www.friam.org
>

--
Giles Bowkett = Giles Goat Boy
http://www.gilesgoatboy.org/

Parks, Raymond

Is Robert a Deterministic System?

In reply to this post by Roger Critchlow-2

Roger Critchlow wrote:

> Near the top of the CPU usage sorted list there will be a process with
> the Image Name of "System Idle Process". This is a process which does
> nothing. It gets scheduled to run whenever there's nothing else to
> do. It's currently consuming about 80-90% of the CPU time on my
> computer, so my computer is running a 10-20% capacity load.
>
> How does the computer know this? There's another process near the top
> of the list with the Image Name "taskmgr.exe". It's running 0-5% of
> the CPU on my machine. This process is reading the list of processes
> to run and deciding which process to run next.

This leads right into one of my pet peeves with Winders. When I boot
up my home system (AMD XP1800 running W2KPro), I nearly always get
impatient while waiting for all the startup programs to run. So, I
start up the Task Manager, sort by CPU and watch. Over a period of
minutes, the startup programs each intermittently take 5-10% of the CPU
for a cycle or so of Task Manager display. Most of that time that I'm
waiting for the machine to boot up System Idle Process has more than 90%
of the CPU load. What's up with that? Why doesn't Winders just run the
programs until they're done? Yes, I know about waiting on resources,
etc, but this type of inefficiency clearly points to a really bad
context switching and scheduling algorithm.

--
Ray Parks rcparks at sandia.gov
IDART Project Lead Voice:505-844-4024
IORTA Department Fax:505-844-9641
http://www.sandia.gov/idart Pager:800-690-5288

Russell Standish

Is Robert a Deterministic System?

It's quite possible the processes are waiting on I/O, so you can get
it to boot faster with faster disk. Another trick (available to Linux
users at least) is to background tasks that aren't essential for
interactive use - eg if you run sendmail on your laptop (as I happen
to do), you don't need to wait for sendmail to start before you log
in. My understanding was that Windows already did this by default -
but perhaps that only came in with WinXP.

Slashdot have had a number of discussions about tuning boot times -
you might like to search for other people's comments.

Cheers

On Wed, Nov 30, 2005 at 03:49:14PM -0700, Raymond Parks wrote:

> Roger Critchlow wrote:
> > Near the top of the CPU usage sorted list there will be a process with
> > the Image Name of "System Idle Process". This is a process which does
> > nothing. It gets scheduled to run whenever there's nothing else to
> > do. It's currently consuming about 80-90% of the CPU time on my
> > computer, so my computer is running a 10-20% capacity load.
> >
> > How does the computer know this? There's another process near the top
> > of the list with the Image Name "taskmgr.exe". It's running 0-5% of
> > the CPU on my machine. This process is reading the list of processes
> > to run and deciding which process to run next.
>
> This leads right into one of my pet peeves with Winders. When I boot
> up my home system (AMD XP1800 running W2KPro), I nearly always get
> impatient while waiting for all the startup programs to run. So, I
> start up the Task Manager, sort by CPU and watch. Over a period of
> minutes, the startup programs each intermittently take 5-10% of the CPU
> for a cycle or so of Task Manager display. Most of that time that I'm
> waiting for the machine to boot up System Idle Process has more than 90%
> of the CPU load. What's up with that? Why doesn't Winders just run the
> programs until they're done? Yes, I know about waiting on resources,
> etc, but this type of inefficiency clearly points to a really bad
> context switching and scheduling algorithm.
>
> --
> Ray Parks rcparks at sandia.gov
> IDART Project Lead Voice:505-844-4024
> IORTA Department Fax:505-844-9641
> http://www.sandia.gov/idart Pager:800-690-5288
>
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at Mission Cafe
> lectures, archives, unsubscribe, maps at http://www.friam.org

Parks, Raymond

Is Robert a Deterministic System?

Russell Standish wrote:

> It's quite possible the processes are waiting on I/O, so you can get
> it to boot faster with faster disk. Another trick (available to Linux
> users at least) is to background tasks that aren't essential for
> interactive use - eg if you run sendmail on your laptop (as I happen
> to do), you don't need to wait for sendmail to start before you log
> in. My understanding was that Windows already did this by default -
> but perhaps that only came in with WinXP.

It does, to some extent, but these background tasks seem to get
either an equal or higher priority than my foreground tasks. Actually,
to say this is during boot up is incorrect. I'm talking about the
period after I login until the various systray tasks are started. While
this is happening, I can do what I like, but my request is thrashed
along with the systray tasks (I've watched this in task manager).

The way in which the systray tasks (and whatever I attempt to start)
swap back and forth, with significant periods when none of them are
getting cpu cycles, implies to me that there is a problem with context
switching and task scheduling. If one task is waiting on the disk, then
another task must be getting disk service. If they are all waiting on a
service (disk or otherwise) then the OS has a serious problem with
resolving these types of conflicts.

A friend (not on this list) suggested that the programs themselves
could be causing the problem. Apparently, some Winders programmers of
these types of applications (instant messengers, preloaders, printer ink
watchers) actually program in wait states in order to avoid conflicts
with other programs during start up. Which, in itself, is a comment
about the stability of the underlying operating system. Just as one
should ask nurses about the best doctors, one should ask the real
programmers about the operating system.

> Slashdot have had a number of discussions about tuning boot times -
> you might like to search for other people's comments.

I just don't understand why I should have to tune boot times. I used
to do that back in the day with 16-bit minicomputers and Phoenix drives
(the kind you used to have to physically flip over to get the other
side). Why is this required for a "modern" operating system and hardware?

--
Ray Parks rcparks at sandia.gov
IDART Project Lead Voice:505-844-4024
IORTA Department Fax:505-844-9641
http://www.sandia.gov/idart Pager:800-690-5288

Russell Standish

Is Robert a Deterministic System?

On Wed, Nov 30, 2005 at 05:27:45PM -0700, Raymond Parks wrote:
> The way in which the systray tasks (and whatever I attempt to start)
> swap back and forth, with significant periods when none of them are
> getting cpu cycles, implies to me that there is a problem with context
> switching and task scheduling. If one task is waiting on the disk, then
> another task must be getting disk service. If they are all waiting on a
> service (disk or otherwise) then the OS has a serious problem with
> resolving these types of conflicts.

Don't forget that the disk itself takes time to return the data
requested of it. While this happens, tasks will usually wait (block)
until the data appears. I notice this quite a lot on my laptop which
has I suspect a rather slow disk. But this shouldn't cause thrashing...

I cannot comment about Window's abilities in task switching, as I
really haven't had much experience with Windows. All I can comment is
that Windows ME absolutely sucks at it, and Windows XP home edition is
not too bad (only slightly more sluggish than the Linux running on the
same laptop). But then - I've never punished a Windows box, unlike a
few Linux boxes around here...

>
> I just don't understand why I should have to tune boot times. I used
> to do that back in the day with 16-bit minicomputers and Phoenix drives
> (the kind you used to have to physically flip over to get the other
> side). Why is this required for a "modern" operating system and hardware?
>

Linux is designed to be a general purpose operating system. The sorts
of daemons (sorry "services") one runs on a laptop are likely to be
rather different from a departmental file server, or an ISP mail
server. But only one distribution is available on CD, which is
generally a compromise for all the uses that the distribution author
thought of. Consequently, tuning your OS by removing unecessary
daemons, and or backgrounding them, or running them on different
priority levels can make a substantial difference to the machine's
performance (or perception thereof). But, no it is not required!

I have no idea how Microsoft deal with this, as Windows does not seem
to have anything like the configurability of Linux.

Cheers

--
*PS: A number of people ask me about the attachment to my email, which
is of type "application/pgp-signature". Don't worry, it is not a
virus. It is an electronic signature, that may be used to verify this
email came from me if you have PGP or GPG installed. Otherwise, you
may safely ignore this attachment.

----------------------------------------------------------------------------
A/Prof Russell Standish Phone 8308 3119 (mobile)
Mathematics 0425 253119 (")
UNSW SYDNEY 2052 R.Standish at unsw.edu.au
Australia http://parallel.hpc.unsw.edu.au/rks
International prefix +612, Interstate prefix 02
----------------------------------------------------------------------------