Posts by Dimitris Hatzopoulos

41) Message boards : Number crunching : Current and Future (Message 19363)
Posted 27 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
Keith, since the science team might have a hard time reading every message during the CASP7 period, you might want to ask again in Discussion on Dr baker's journal or some other appropriate topic in Science subforum.

I've asked this question in the past myself.

Obviously a lot depends on the protein topology itself. Some are particularly hard (e.g. 1tul), which is why the RMSD-Energy chart distribution of red dots is so different from one protein to another.

AFAIK the energy function is quite accurate, but the problem is sampling the conformational space.
42) Message boards : Number crunching : Report Problems with Rosetta Version 5.24 (Message 19350)
Posted 27 Jun 2006 by Profile Dimitris Hatzopoulos
Post:

Ideally ofcourse the BOINC-server should be smarter and noticing your PC can't handle the ultra-big WU, send you one of the regular, smaller jobs. But this feature is not yet available in BOINC server code unfortunately.


Actually it is. There is only enough space in the feeder queue for 1,000 workunits. When the scheduler connects up to the feeder queue to get work it cycles through all 1,000 slots looking for available work. When all 1,000 queue slots are filled up with large jobs that is what the server returns.

Splitting the queue up equally is supported with different applications. If this is really a big problem we could set things up in such a way that the project believes it has more than one application and 50% of the queue is saved for each application.


Rom, thanks for the feedback. Although I don't know how the current system works (what are the "groups" of jobs sent, e.g. jobs needing 256, 512, 768, 1G? memory ) it seems it'd help to split the queue as you suggest to make sure there are always small jobs available.

Apparently many people get this message "there was work, but your PC has less RAM than needed", see e.g. posts by Carlos (a very small percentage of users posts here).
43) Message boards : Cafe Rosetta : About the price of oil (Message 19322)
Posted 26 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
One more article on the same subject (less technical, as it's aimed at the general audience):

source

Interview: Oil Expert Explains What's Driving Record High Prices
By Jeffrey Donovan
Azerbaijan -- Caspian oil derricks, pollution
(AFP)
Oil prices have climbed steadily since 2004, and today they rose again to new record highs of more than $74 a barrel. What's driving prices -- and can they be brought back down? Government officials and oil executives from more than 65 nations will be asking those same questions when they gather in Qatar on April 22-23 at a meeting of the International Energy Forum. Among the factors driving prices up are fears over possible supply disruption in major oil-producing states such as Iran, Iraq, and Nigeria, where a rebel offensive has cut off one-quarter of output.

PRAGUE, April 20, 2006 (RFE/RL) -- RFE/RL spoke with Leo Drollas, the deputy director and chief economist of London's Center for Global Energy Studies.

RFE/RL: The price of oil continues to climb to record highs. Analysts say fears over the Iranian nuclear crisis is one of only several issues driving up prices. Iranian President Mahmud Ahmadinejad only strengthened those fears on April 19, when he said oil prices are still below their true levels. What's your take?

Leo Drollas: The first thing we must say is that these high prices are really paper prices, in a sense for paper oil, they're not the prompt market -- that is, for real wet barrels, as we call them. So they're obviously driven a lot by news and by sentiment, and they refer to oil that is bought and sold months ahead through paper contracts. And these prices move quite quickly on news, and the news is bad at the moment for oil because of the Iranian standoff, because of problems continuing in Nigeria, and many other factors in the market.

RFE/RL: So you're saying that, beyond fears over Iran or supply disruptions in Nigeria, the futures market is driving up prices?

Drollas: There's a lot of money that has come into the oil market over the last few years. The money that is now tracking commodity indices has increased from about $8 billion in 2001 to about $70 billion today. So we've got a huge influx of money into commodity-tracking indices, and a large part of those indices of course refer to oil. So we've got a lot of speculative money or hedge-fund money or other kinds of investors coming into oil, thinking they're on a roll now and that oil prices will forever increase. And in a sense, this tends to fulfill the prophecy, as long as the money keeps coming in.

RFE/RL: How does that drive prices, exactly?

Drollas: What tends to happen is that the futures prices, especially for months further out, tend to rise, and they create a difference between future prices for outer months and spot prices for oil, especially for "wet" barrels. And this differential encourages people to buy physical oil, and therefore, the pressure from the futures market is transmitted to the actual spot market for oil.

RFE/RL: But isn't capacity part of the problem as well, both in terms of the actual supply of oil and the fact that, particularly in the United States, refining capacity is down, with some refining facilities still not fully recovered from last summer's Hurricane Katrina?

Drollas: Well, this is the huge, unanswerable question as to what component of this price run-up is due to speculation, as we call it, or due to real fundamental factors. My own gut feeling is that maybe $15 [per barrel] or so at the moment is due to these kind of pressures from the futures market, from speculation if you like. In other words, the price should have been quite a bit lower than that, because there isn't actually a physical shortage of oil at this very moment.

RFE/RL: So this gets us back, perhaps, to emotions -- to fears over possible supply disruptions due to crises in key producing countries, such as Iran, Nigeria, Iraq.

Drollas: Of course, the fear is there because of these factors: Iran and Nigeria in particular, and also Venezuela to some extent in the background. But one must bear in mind that there is a physical problem too, in the sense that the spare capacity to produce crude oil in the world is under 3 percent of global demand for oil. And a business that runs with such little spare capacity is vulnerable to all kinds of events that make people worry or fear whether the system is able to cope with the pressures that are put on it in, let's say, the autumn or the winter that we have ahead of us.

rest of interview in the link above
44) Message boards : Number crunching : Changes to robots.txt to avoid Boinc DB overhead by bots (Message 19293)
Posted 25 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
I was reading over SIMAP about excluding bots from accessing BOINC dbs (for performance reasons) -here- and checking R's robots.txt noticed that it needs to be changed by adding the "/rosetta/" path to the URLs excluded from bot (Googlebot, Yahoo Slurp etc) visits.

ie.
http://boinc.bakerlab.org/rosetta/robots.txt

User-agent: *
Disallow: /account
Disallow: /add_venue
Disallow: /am_
Disallow: /bug_report
Disallow: /edit_
Disallow: /host_
Disallow: /prefs_
Disallow: /result
Disallow: /team
Disallow: /workunit

should be:

User-agent: *
Disallow: /rosetta/account
Disallow: /rosetta/add_venue
Disallow: /rosetta/am_
Disallow: /rosetta/bug_report
...
etc

as the default examples are relevant for project URLs which don't include a path.
45) Message boards : Rosetta@home Science : WorldCommunityGrig - HPF2 (uses Rosetta) Phase 2 started (Message 19287)
Posted 25 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
As for the redundency on the results, well I don't think they changed it from the default BOINC settings ;-), also since it goes into a database and they are actually generating 'results' (rather than R@H's more developmental setting) it needs to be double checked at least. Also that's the way it works on Grid.org software...


Fluffy, CASP7 is actually very much "production" (vs "development" / "experimental") and yet R@H doesn't need to run redundant WUs.

To re-iterate the points in the validation thread, looking at the predictions of 2chf (from Top-Predictions page)



we're all looking for a needle in a haystack: the lowest energy structure highlighted in blue at the bottom-left of the graph.

If R@H were sending out the same WU 5 times, i.e. re-calculating every red dot plotted above 5 times, we'd be able to calculate overall 1/5 as many dots (unique predicted protein structures) and quite possibly miss the elusive blue dot. One can also see that there is no clustering of results in the bottom-left corner of the RMSD-Energy graph.

Just because grid.org and WCG run it this way, doesn't make it right. The only reason would be for calibrating credits, but not for the science IMHO.
46) Message boards : Rosetta@home Science : WorldCommunityGrig - HPF2 (uses Rosetta) Phase 2 started (Message 19261)
Posted 25 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
I try to refrain from being critical, but take a good look at WCG's initial replication and quorum settings.

Then, have a look at our discussion in Validation (not for credits, but for scientific reasons)

While I certainly believe HPF is a very important project, I am not happy with the level of redundancy WCG (IBM) runs it. Maybe FAAH really needs this level of redundancy and they can't differentiate between WUs/apps? (again I need to look at BOINC-server code sometime)

Unless ofcourse I'm missing something obvious.
47) Message boards : Rosetta@home Science : Rosetta@home page at Wikipedia (Message 19259)
Posted 24 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
I just noticed a fellow update the info in Rosetta@home page at Wikipedia (which is great as I don't always have the time to update things) and just wanted to note a couple of things:

In the "The project already has ~XX,000 active PCs (Jun-2006) for ~YY TeraFLOPS " paragraph, I use the number of active PCs (hosts) from http://www.boincstats.com/stats/project_graph.php?pr=rosetta which currently shows ~65K hosts (not to be confused with the ~70K users / people).

Also, I noticed the latest official project requirements now suggest 256MB RAM minimum. The avg WU I get is still 75-140MB (with the big ones ~300MB), so if we allow for 40-80MB for O/S (not including e.g. antivirus etc) doesn't it leave a very little RAM left for apps before system starts paging to disk (and affects performance)?
48) Message boards : Cafe Rosetta : About the price of oil (Message 19238)
Posted 24 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
Have you studied the effect of oil price on the other commodities? I mean if price of oil goes up, it will cost more to run a tractor across a field and harvest sugar cane, or cotton or oranges... and electricity costs will increase which has a great impact on smelting of metals, precious or otherwise. I mean how much of the increase in the prices of these other commodities is directly attributable to the increase in oil prices?


This is a very good question and it sounds logical.

In my opinion (being a participant in the markets for many years) is that it's primarily a FINANCIAL phenomenon, initiated by the tsunami of paper money and credit, unleashed by the world's central banks, starting back in 2002. Basically it's been just monetary inflation via an explosion of paper money and credit. Official reported inflation was masked mostly by changes in the CPI (if CPI were calculated as in 1980s, inflation would stand at 8% yoy) but also in part by the deflationary forces of globalisation.

In the process, we experienced a "Flucht in die Sachwerte" by more savvy investors.

Now, to answer your question about whether energy costs cause the price of other goods to rise, I'll send some more charts of commodities, which attempted to rally along with the rest but couldn't hold those gains:









So WHY do some commodities rally but others not e.g. cotton or corn or wheat or coffee? By now you'll probably have put the pieces in the puzzle, by noticing that the commodities which enjoyed big price increases (and held onto those price gains) were the ones which are either STORABLE (base and precious metals) or had a BROKEN price discovery mechanism (as in my opinion has oil, with no link between futures and spot and priced off futures).

So, basically in all cases the rally in prices has been due to "investor" ("speculator") money inflows. Obviously for those commodities where storage is too cumbersome and/or rot/decay/decompose etc it didn't work so well.


Btw, orange juice is in fact "frozen concentrated orange juice", which is why it's storable and it worked.

Also, don't think that e.g. cocoa supply caught up with "demand" and brought the price down (you know, from the billionz of Chinese and Indians who suddenly found out the joys of drinking hot chocolate as the media of mass deception would say :-), because a cocoa-tree needs about 5yr to grow.

Finally, you'll also notice, that e.g. natural gas, which is FUNGIBLE for oil in many applications, has dropped quite a bit (-65%) since Dec-05, as storages are full and they have no room to put the extra production! Yet oil has held to its entire gains.


49) Message boards : Number crunching : Report Problems with Rosetta Version 5.24 (Message 19234)
Posted 24 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
I have been finding lately that when I check on my computer running version 5.24 I see the following message:

6/24/2006 1:31:16 PM|rosetta@home|Message from server: Your computer has only 266780672 bytes of memory; workunit requires 233219328 more bytes
6/24/2006 1:31:16 PM|rosetta@home|Message from server: No work sent
6/24/2006 1:31:16 PM|rosetta@home|Message from server: (there was work but your computer doesn't have enough memory)
6/24/2006 1:31:16 PM|rosetta@home|No work from project


Please see my earlier comments in this thread about encountering a really big WU. Some CASP7 WUs require lots of memory, so knowing that your PC can't cope, BOINC-server at Rosetta@home won't send it to your BOINC-client/PC. And gives you this message instead.

The Rosetta team (see answers to my question by David Kim and Bin) said they have mixed small and large jobs in the queue, so if you try again ("manual update") you'll probably get a WU that can be handled by your PC.

Ideally ofcourse the BOINC-server should be smarter and noticing your PC can't handle the ultra-big WU, send you one of the regular, smaller jobs. But this feature is not yet available in BOINC server code unfortunately.
50) Message boards : Rosetta@home Science : CASP7 confusion (Message 19233)
Posted 24 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
CASP is a "blind test" performed every 2 years, where teams working on "protein prediction" from all over the world, submit their "predictions" (via computational methods) for target proteins (targets are the ones we see T301, T319, T320) which are currently being solved EXPERIMENTALLY (in the lab, via Xray crystallography).

So the target proteins structures of CASP7 are still unknown, but will become known in a few weeks / months.

One of the CASP7 targets T294 Xray structure was published early ("the exam question was leaked" so to say), so it was cancelled and they won't take mathematical predictions for it.


E-Mail from CASP team 16/06/2006

Bad news first ...
It was brought to our attention that a paper was published on CASP7
structure T0294 ( http://dx.doi.org/10.1016/j.jmb.2006.05.018 ).
As this happened less than 3 weeks after the target release, we are
canceling this target according to our rules. No models will be assessed
on this target. This information leak is not a fault of the SG center
that provided the target, as the publication is from another group that
solved the same protein.
51) Message boards : Cafe Rosetta : About the price of oil (Message 19177)
Posted 23 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
One more week gone by, with oil closing above the $70/bar price (it's been 2 months already). Doesn't make headlines anymore and people seem to be "accepting" it (?).

A poster at SETI's forum (here) asked me if I have an idea when the "bubble" will burst. I don't, primarily because I see incredible complacency by those being victimised (i.e. consumers) to the point I have to wonder about a hidden agenda (see my article on oil) and also because unlike stockmarket bubbles which deflate out of their own weight (nobody is obligated to buy a stock if he thinks it's overvalued), the KEY difference with a runup in vital commodities like energy is that there is almost inelastic, price-inensitive buying by the victims who need it.

The runup in price of commodities due to "investor" inflows is not limited to oil and it's (IMHO) NOT because everything is suddenly running out (peak XYZ). It's just that it's the one which most affects most westerners' everyday life. And because people notice the increase in its price (unlike many of the other commodities below).

For comparison, I'm attaching monthly charts of various commodities. You'll notice that those which enjoy almost inelastic, price-insensitive demand by consumer-victims held up their price quite well.

Oil:


Copper:


Platinum:


Gold/Silver



Orange Juice:


Sugar:
52) Message boards : Number crunching : Report Problems with Rosetta Version 5.24 (Message 19120)
Posted 22 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
@Dimitri

The big WU are only sent to machines with 512 MB RAM and above. I'm glad they finally decided to use this BOINC-Feature to send larger jobs to higher-spec-machines.


Mea culpa, I thought it worked as I described below. I haven't looked into the sources for quite some time.
53) Message boards : Number crunching : Report Problems with Rosetta Version 5.24 (Message 19100)
Posted 22 Jun 2006 by Profile Dimitris Hatzopoulos
Post:

We've marked these WUs as high memory jobs and they will only be sent to machines with more than 512M memory.


I've not kept up with BOINC-server developments, but I think that currently such a high-memory job will still be sent to e.g. a PC with only 256MB RAM. And only AFTER that WU has been downloaded (a procedure which might last 45min-1hr for a dialup guy) the local BOINC client will notice it's not suitable and dump it.

Personally, I don't mind, as I have fast Internet and a few months ago I had upgraded all my PCs with extra RAM "for Rosetta".

But from the perspective of a dialup user... waiting 45min-1hr to download a 12MB WU and then see it aborted. He wouldn't be very happy.
54) Message boards : Number crunching : Report Problems with Rosetta Version 5.24 (Message 19093)
Posted 22 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
Not a bug, but I noticed a v5.24 WU FRA_t298_hom001_5_IGNORE_THE_REST_dec22.pdb_747_47_0

which has a working set (RAM usage) of almost 300MB (298) and 837MB virtual.

Plus the data files for this WU are ~11.5MB.

It reminds me that Rosetta could really use a BigWU flag and my concern that woth such WUs we might lose folks with 512MB RAM (who run other software in addition to crunching for Rosetta) and/or dialup.
55) Message boards : Number crunching : Memory requirement in the future (Message 19089)
Posted 21 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
Hmm, a WU FRA_t298* I've running with v5.24 right now shows a "Working Set" (i.e. physical memory of 285MB) and virtual memory of 838MB!

And the WU data itself two main files are 8.5MB + 3MB (gzipped)!

The BigWU switch in BOINC is BADLY needed... or we'll lose many folks with medium RAM PCs and dialup.
56) Message boards : Cafe Rosetta : About the price of oil (Message 18945)
Posted 19 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
Much as one would like to blame the OPEC cartel, we can't ignore the fact that the markets that create this broken pricing system are in NY (NYMEX) and London. Just what else can the Arabs do, other than say "the market is oversupplied and over-priced" as they've been saying since 2005, or "find me a buyer and I'll give him 2 million barrels/day" ?

If the consuming nations want to break this vicious circle, there are several options, country-to-country contracts being one of them.

As for being a market bubble, I agree it is the case. But the issue here is that unlike a stock bubble, which one can simply ignore, we're all forced to be "long energy" and pay the price dictated by this broken pricing system! It's been THREE (3) YEARS of it already.
57) Message boards : Cafe Rosetta : About the price of oil (Message 18944)
Posted 19 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
I'll copy here some comments I made in reply of questions/comments posted in different forums:

--

To repeat the issue with broken arbitrage in oil:

If arbitrage in oil would work as in almost every other market, then I could e.g. buy those extra 400,000 barrels/day from Saudis (which recently cut, citing lack of buyers) at prevailing price $70 and immediately sell short 400 contracts at $70.

If I held my shorts to delivery (of physical), then the "investors" in oil would have to build storage tanks to accumulate the extra 400Kbpd, i.e. 12 million barrels per month. If I kept doing it for a few months, investors would have to keep building tanks to accept new oil or at some point choke on the supply of new oil and stop initiating new longs (buying new paper barrels of oil). Yet I would still be buying 400Kbpd from Saudis and selling short 400 contracts per day, which -in absence of "investor" buyers- would pressure the price down.

And when that happens, the "oil investors" would incur a big loss, because they'd have accumulated their hundreds of millions barrels inventory held in their tanks, at higher prices.

The abovementioned scenario ofcourse assumes REAL "wet barrel" oil demand is stable (e.g. China's oil imports were down -2.2% in 2005).

But, apparently it works differently for oil, because unless those 400Kbpd from Saudis are the exact type specified in the futures contract (as e.g. Brent is 0.4% of world's oil production, but is used to price 60% of world's oil), I can't do this kind of arbitrage at a scale that it would PROMPTLY affect price.

That's probably why oil price is going up while producers claim they can't find buyers for their real "wet barrels" of oil.


This "feeding of barrels" is happening and "investors" are hoarding it, which is why you see inventories rise, but it's an extremely slow process and meanwhile gullible consumer victims are forced to pay these obscene prices.
58) Message boards : Cafe Rosetta : Illegal Borging and the Impact on Distributed Computing (Message 18905)
Posted 18 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
This is a very good argument in favor of Folding@home offering a BOINC-version, BOINC being the DC defacto standard, so one has to maintain a single DC platform.

I know that BOINC still doesn't support some features F@H is using (e.g. opt-in to run BigWUs etc), and I hope BOINC developers include them asap.

Similar problems arise when people can't run re.g. WCG and CureCancer on the same PC, even though both use the very same agent sw (UD's)
59) Message boards : Cafe Rosetta : About the price of oil (Message 18825)
Posted 16 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
One more week ended with oil at $70/bar (it has "stabilised" at 70-75 for the past 2 months). I'm copying some comments from mainstream media, on the issue of speculation in oil. He makes a good point, that very often people discuss the irrelevant "fixes", not the problem.

Although, as I explained many times, the real problem isn't speculation (which is a useful transfer-of-risk service), but the FAULTY pricing system for oil which is being taken advantage of. So we have a situation where oil price goes up and up and at the same time oil producers can't find buyers for their oil (even light sweet crudes, just in case one wonders).

Recently (Apr-06) Saudi Arabia reduced production by -400,000 barrels/day (from 9.5Mbpd to 9.1Mbpd) because it could find no buyers for it.


The Danger of Speculation
Commentary by Mike Norman for FOX Fan Central

Mike Norman
It’s time to speak the truth. No more disingenuous questioning and wondering. No more exasperated resignation. We know the reason why oil prices are high, and it’s time to admit it and do something about it.

Oil prices are high because of speculation, pure and simple. That’s not an assertion, that’s a fact. Yet rather than attack the speculation and rid ourselves of the problem, we flail away at the symptoms. High gasoline prices? Oh, let’s use hybrid cars, or drill in the Rockies or off the California coast. How about doubling the use of ethanol, even though it costs more to produce than the energy you get out of it? Then again, we can go to Alaska, or build more refineries, or triple the number of nuclear power plants. Sound good?

What if we just stopped the speculation?

No, you can’t do that! That would be interfering with the “free” market.

Hey, the “free” market is starting to get awfully expensive.

Tell me, how is it free when speculators rule the roost? It’s one thing when they do what they do with pieces of paper called stocks, but it’s another when they do it with a vital commodity like oil. We saw the devastation their behavior wrought in the 1990s, and we’re witnessing it again right now.

rest of article: http://www.foxnews.com/story/0,2933,166038,00.html
60) Message boards : Cafe Rosetta : About the price of oil (Message 18560)
Posted 13 Jun 2006 by Profile Dimitris Hatzopoulos
Post:
Physical ("wet barrel") market is drowning in oil.

A few days ago (5-Jun-06), WSJ reported that the Oil Minister of Saudi Arabia said they had trouble finding buyers for crude oil and hence reduced production from 9.5Mbpd (where it stood for the past 2yr) to 9.1Mbpd:


In an interview after a meeting here of the Organization of Petroleum Exporting Countries, Ali Naimi said other cartel members are having trouble finding buyers for all the crude they are producing, at a time when global stores are near full and many refiners have closed facilities for routine maintenance. One Saudi official said an estimated three million barrels a day of refining capacity is out of action and unable to process crude, at a time when the world is using some 84 million barrels a day of oil products like gasoline and jet fuel.

"It's not just heavy oil. Even light oil is having problems" finding buyers, Mr. Naimi said, referring to premium grades of crude known as light crude that are highly prized by refiners because they have high gasoline yields.

Asked if the kingdom was easing up on supply because of concern about the buildup of inventories in the U.S. and other importing countries, Mr. Naimi rejected such a motive, replying: "At $70 a barrel?" Mr. Naimi suggested that producers will sell all the oil they can at such high prices.

The implication of Mr. Naimi's remarks is that Saudi Arabia would again open its oil spigots when buyers ask for more oil. For the past two years, the Saudis say, their policy has been to sell as much oil as buyers want, to the limit of the kingdom's production capacity.


Previous 20 · Next 20



©2024 University of Washington
https://www.bakerlab.org