Thu: gridding it up …

Ok, ok, it seems we’ve got the grid system working correctly. I’ve changed a number of things in the code:

• no print to screen to say that it is on line x of y results wise
• changed the directory to pump data out to
• changed the times to run to 1 — this is becase the queuing system will have multiple jobs submitted to it
• num_boids — originally set low for testing, but then increased it
• timesteps — set to 1000

… and then changed the controller code to only run the simulation. The remaining data analysis — e.g. go through each file and calculate hulls, etc — will need some tweaking, methinks.

I also needed to change the filename syntax — each file now has a random number added into it. It appears that 10 jobs would only produce 8 results, I presume to a number of jobs having the same ddmmmyyyhhmmss syntax in them. The addition of a 5 digit random number should clear this up.

But …. wow, is it quick. Each job takes ~2min on a remote computer. Therefore a 20min run = 2min.

I heart the grid. It is awesome.

Tue: notes from Fri with Will & Stats and general update …

Applied for Grid access: granted but can’t log-in like a noob. Email DRS to see what I’ve done wrong, no doubt.

Went to Stress Management course: met Helena/Elena and Julia.

Met with Paul C. of Stats Services — some notes:

• use summarisation to indicate results: namely the point at which the gradient plateaus (presum. the gradient), the size of the area produced (will prob. be a max. due to using -ve numbers), and the timestep when the max. area was identified.
• anticipate that quadratic result will be seen
• use regression testing to go forwards. Need to use about 99 data values.
• R code of use?: model<-lm(y~x1+x2+…+xn) summary(model)
• Regression (numbers only) and ANOVA (groups only) analysis are children of class General Linear Model (groups or numbers or any combination). t-test is a child of ANOVA (ANOVA with 1 factor with 2 groups)

General update

• good tutorial on ANOVA analysis: http://www.celiagreen.com/charlesmccreery/statistics/anova.pdf
• need some way to see if what is being produced by model is actually of sense …
• had a quick look on t’interwebs about physics (turning and momentum) — nothing great, but will need to look again methinks.

Monday: random action and T-tests …

Created a random boid module, so have stuff for null hypothesis. Now to go get data!

Look at http://www.stattutorials.com/EXCEL/EXCEL_TTEST1.html re: T-tests in Excel.

Saturday: unravelling coding issues …

I, world, am a plonker. Realised that apparent issues with results was due to the fact I was changing a variable, and when x = 0, nothing happens — as it should. Great eh? Sigh.

Now need to work out how to analyse all of my lovely results, woop!

Thursday: one bug down, one to go …

One bug has been resolved: there was an out of range array error because where the last line of data was being written to the output.

Next bug appears to be in mungCoordsAndVels somewhere — the first line of munged coords data is always the same. Look at the debug code to see if you are closer.

Thursday: num_boids, why do you frustrate me so? …

Wasted all of last night rebuilding my laptop. Thank you Norton. Only got the OS installed and patched, so now need to install apps — yay!

Today been trying to get the visualisations running, but spotted that the data output under certain conditions is identical — and I don’t know why. From looking at the code, I keep getting a range index error message when the num_boids submitted via the Controller exceeds 9 — I think it is something to do with double digits breaking the range. Anyway, one for tomorrow methinks.

Monday: wifi woes …

Today is a lack of coding. Although being snowed in means I should be rested and ready for action, an attempt to update my antivirus software has crippled my main PC, by preventing any network connections, either LAN or Wifi. So currently somewhere in driver hell, trying to get it back to where I was this morning.

Ho hum, at least it is secure at the moment … ! Although the dratted driver install now appears to have frozen … :-(. Hopefully inspiration will come to the rescue.