Now I have added a comment

[More]

Updated to 10.5.1. First impressions are

 

1.  Back-to-my-mac (including both file sharing and screen sharing) now appears to work more consistently than before.
2.  Internet sharing giving out IP addresses via DHCP seemed not to work.
3. I note that the firewall now tells more of the truth, but that is cosmetic.

Other point to make is that I am really liking QuickLook, which works well in several different contexts. I would add that since this is linked to CoverFlow, I actually realise that CoverFlow is useful for browsing (particularly when pruning out old useless files) because you can see what you are looking at without having to open it.

 

[More]

[More]

I agree that it is not as impressive as I had hoped and this has led to a feeling of disappointment.

There are some things about Spaces that are better than VirtueDesktop, in my opinion, and some things that were better IN VirtueDesktop. For example, I think using the exposé-like interface for Spaces is better than the VirtueDesktop interface, but VirtueDesktop did not tie an application to a window if you docked it [I found that was a useful way of moving applications from one window to another - shrink application into dock, move to other window, unshrink from dock] whereas with Spaces it seems the only way to move an application from one window to another is to press F8 to present the set of shrunken windows and drag the application from one space to another. Once you have opened an application in a window it is tied to that window even if you have docked it. I find this a particular irritation with iTunes - I want it to be playing but out of the way most of the time - with VirtueDesktop I could just unshrink it from the dock quickly and then put it back again and get on with whatever I was doing, but with Spaces I get switched to another window (whichever one I opened iTunes up in) and then have to switch back to where I was working - this is not something huge, but we're supposed t be saying "it's the little things that Apple gets right that make the whole experience better".

The new dock, with stacks, is also a disappointment - I couldn't care less about the glass shelf because I always have my dock on the side of the screen and never see it. The look for the dock on the side - a translucent dark grey - is fine if you have a light coloured desktop but awful (in my view) if you have a dark desktop (as in the new default desktop image). Hence my first reaction to the new dock was "yugh!". I'm over that, but stacks is still an issue. I don't like the way the icon for a folder changes to the icon for the first item in whatever sort order you have chosen. It is no longer obvious what is a folder and what is a file. When you open a stack with many files the grid display is unusable. It is "cute" to see the thumbnails of the first pages of some documents and the quite pleasant images for some generic file types, but I am well past the stage of limiting my file names to 12 characters, which means that most of the names of my files are truncated and illegible until I call up a proper finder window. Two clicks instead of one. One step forward two steps back.

The biggest plus, I think, is that Leopard seems to me (subjectively) to be a little quicker. 

I think iCal is supposed to be much more usable for sharing calendars than before - I will check that out and hope it is true.

Mail has a few tweaks - the auto-detection of names and dates and the integration with iCal will, I think, be quite useful - but this was not an obvious thing - I stumbled on it by accident when my cursor was floating above the word "tomorrow" in a mail message and a little drop-down menu appeared. I can imagine some people using the new Notes and To Do list features, but I don't think I will be one of them - they are not bad features, it's just that I have a way of working already and they don't fit in.

Similarly, I don't think I need iChat to share files. In the middle of a chat I might use it but it doesn't solve a problem I had.

Time Machine is obviously a good idea, but I have not tried it out and probably won't in the near future - I don't need another backup option on the desktop, and my laptop really does spend most of it's waking life on my lap, and not connected to a massive external disk. The initial promise of being able to do this through one of the new Airport base stations was appealing, but was withdrawn just before release of Leopard.

The interface to Spotlight is better than it was before, but still does not give users access to the full power of the underlying engine. 

In a sense, I am frustrated by this upgrade because I think it really IS better than Tiger, but as far as the user experience is concerned it is mostly by small increments. And the new features seem promising but do not fulfil the promise yet. I wouldn't willingly downgrade to Tiger, but I find it hard to pinpoint why someone using Tiger should rush to upgrade.

 

[More]

test

[More]

I attach plots of my comparison of lattice parameters computed by CASTEP and GULP:

Here black = CASTEP, red = GULP (my standard model without shells or three-body forces).

There are several discrepancies, including the different slopes (ie different elastic constants) and GULP not getting the behaviour in c at low-pressure.

Next picture shows the phonon frequencies of the set of lowest-frequency modes for wave vectors along [110]:

Note that this computes imaginary frequencies (negative squares of frequencies, shown as negative values here). This is what you would expect for a phase transition, and interestingly there is a dip around a value of [1/3,1/3,0] as you would expect for experiment.

The negative here though is that the imaginary modes persist to high pressures, so we conclude that the model is unstable against a phase transition but that a phase transition back to this as a stable structure doesn't happen. 

 

[More]

So, we discussed using your initial empirical pots to do MD with DLPOLY3, but isn't the danger here that we'll simply get the same behaviour we saw with GULP? What I'm saying is that if these pots want to deform the real structure with GULP, surely we'll see something similar with MD?

[More]

Look at the structure. When I added 3-body terms, it helped in some ways but didn't manage to stop the O atom from moving into the Ca layer.

[More]

Martin, I can stop it from collapsing along the c-axis by putting in a 3-body O-Si-O term for the tetrahedra. The question then becomes what's an acceptable error to the cell parameters?

[More]

Mark, if you include a line in the input file (at the end) of the form "output cssr <name>", where <name> is the name of the file (no extension), gulp will create <name>.cssr which CrystalMaker can read. Then you can see the sructure.

[More]