![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Call for Backup
Jun. 4th, 2007 12:27 pmSo, things are looking to be much less dire than in my previous post. As ![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif)
cheshirecatco pointed out, the worst-case situation probably would have been that I was wrong. Which would have still been mightily annoying, since that result was IMO my most important contribution so far, and others have been citing that paper in their own work.
Shijie never seemed particularly concerned about it. He was confident that my earlier results were ok, and the current version of the code has certainly been confirmed to be correct. I heard back from him and he was able to reproduce this calculation with one of my earlier codes. And from what I understand, he was able to get something similar with the current code, using the appropriate initial condition. What this likely means is that it's more sensitive to initial conditions than we thought. I'd done some testing of that, but not enough. This weakens our conclusions, but doesn't invalidate them. This is itself an interesting result, although only to a very few people.
I looked into using CVS to prevent this from happening again, but it seems to be a bit wonky on our network, and would be a hassle to use on the cluster. Since I'm the only one who will be using this code at this institution, I decided to just keep backing it up manually. Each working version gets its own directory tagged with the date. Every time I run a calculation, the code also gets copied into the data directory. The data files are so much larger than the code, it really doesn't matter to the disk space.
I've also got Francis to buy me an external hard drive, so's I can back up the whole drive regularly.
Thanks to all for your support on this. I think that's the most comments I've ever received to a posting.
cheshirecatco pointed out, the worst-case situation probably would have been that I was wrong. Which would have still been mightily annoying, since that result was IMO my most important contribution so far, and others have been citing that paper in their own work.Shijie never seemed particularly concerned about it. He was confident that my earlier results were ok, and the current version of the code has certainly been confirmed to be correct. I heard back from him and he was able to reproduce this calculation with one of my earlier codes. And from what I understand, he was able to get something similar with the current code, using the appropriate initial condition. What this likely means is that it's more sensitive to initial conditions than we thought. I'd done some testing of that, but not enough. This weakens our conclusions, but doesn't invalidate them. This is itself an interesting result, although only to a very few people.
I looked into using CVS to prevent this from happening again, but it seems to be a bit wonky on our network, and would be a hassle to use on the cluster. Since I'm the only one who will be using this code at this institution, I decided to just keep backing it up manually. Each working version gets its own directory tagged with the date. Every time I run a calculation, the code also gets copied into the data directory. The data files are so much larger than the code, it really doesn't matter to the disk space.
I've also got Francis to buy me an external hard drive, so's I can back up the whole drive regularly.
Thanks to all for your support on this. I think that's the most comments I've ever received to a posting.
