Archive | March, 2013

Single(ish) Serve Chocolate Mousse

26 Mar

2013-03-20 14.53.43

Today I got around to making the chocolate mousse I was craving on Sunday It is roughly based of this recipe. Only roughly.

2013-03-20 14.47.05


I thirded it, used less stevia instead of sugar and used a sugar free chocolate. After all of that I decided it was too chocolately (yes, I know, odd) and so put the rest of the cream in.

2013-03-20 14.53.07

Folding in cream

Then it was a bit one noted, so I put in some sherry. Normally I would go for something like Tia Maria or Grand Mariner, but my choice was limited to sherry and bourbon.  Left it in the fridge for a few hours. It is really more than a single serving, but still something you could eat in a single sitting without feeling completely overwhelmed.


  • 90g chocolate
  • 200ml cream
  • 1 egg
  • 1 tsp stevia
  • Some extra flavouring (coffee, alcohol) optional, about a table spoon.

Melt the chocolate in the microwave. I used the bowl that I would eventually keep the chocolate mousse in to save washing up. In a separate bowl whip the cream a bit. In yet another bowl mix egg and stevia until light and creamy. Mix egg into chocolate (make sure that the chocolate isn’t too hot as other wise you will curdle the egg). Fold in the cream. If at this point you feel it could do with some extra flavour, add some. I added in some sherry. Chill until you want to eat it.


Review: Organ Gluten Free Multigrain Pasta with Quinoa Penne

19 Mar

IMAG0263 I normally stay well away from gluten free pasta. My experience with rice or corn pastas is that they are overly mushy, grainy, tasteless and expensive. I’ve also gotten use to not eating much starchy food, so I rarely if ever feel like eating pasta. However, yesterday was a day for comfort food so I bought some Organ Gluten Free Multigrain Pasta with Quinoa Penne. What made me decide to buy it was that it contained quinoa and millet, which I thought may improve the texture somewhat. It was still an expensive buy at $5 for 250g.


I cooked it according to direction: chuck it in boiling water for 6.5 minutes, drain serve with sauce. When I tried it at 6 minute mark, it didn’t really seem al dente, more just hard. At 6.5 minutes it was a bit softer, but still different from what I would call al dente. At least it wasn’t mush or grainy. It also wasn’t tasteless. Being gluten free, it is never going to taste like a durum wheat pasta, but it was fairly good. Something you could eat fairly plainly with just some good olive oil and cheese.

I quickly realised that I’m really not use to eating much starch. While I was full, I really didn’t feel much different if I had just eaten a bag of doritos. My normal standard of eathing pasta sauce on top of a salad suits me much better (but then it isn’t really comfort food). So while the Organ gluten free pasta is a fairly good pasta substitute, the starch along with the expense means it isn’t going to  part of my regular diet.


Comfort Foods

18 Mar

I’m sick. A busy week, lack of sleep and increase in tutoring made me initially think that all I was needed was a good night sleep and a quiet morning enjoying myself. But as it turns out, I needed a quiet weekend, bed and a day or two off work.

So I’m home now, obstinately doing work, really thinking of ways to cheer myself up and stop my throat from burning.

In terms of cheering well myself up. well I find that impossible to uncouple from food. Well I could get a good book and curl up in bed read, but it is even better if I have some nice tea. So really, I’m looking at comfort food.

So far, my best comfort food has been Italian Sausage and Lentil Soup from Smitten Kitchen. I made this last week, so now I have a stockpile in the freezer. Warm soup is soothing and it is spicy enough to clear out my sinuses a bit.

My other default comfort food is porridge. I’m very inexact with my porridge: put some traditional oats in a bowl, poor milk over it, add a splash of water. Chuck the whole thing in the microwave for 2-3 minutes, or until the oats are mushy.  Add some golden syrup, or maybe some fruit. I haven’t actually had any porridge with this bout of sickness. Most grains (other than rice) upset me as does milk, so I don’t keep them in the house. I need to do some research of quinoa porridge: I’m not a huge fan of quinoa, but maybe in porridge while I’m sick will make all the difference.

What I’m really craving now is chocolate mousse. Some cool and velvety sounds perfect for my throat. With an added bonus of chocolate. I don’t really have a default recipe for chocolate mousse (well if I have a Stephanie Alexander book handy, I will make her fudgy chocolate mousse), but I do have a criteria. A good chocolate mousse should have proper chocolate, cream, raw egg and a bit of liqueur or coffee.  Something like this. Unfortunately no chocolate in the house currently, but it may almost be worth an adventure to the shops. The other problem is my sugar-free March. Do I break it if I’m sick needing chocolate mousse? Do I compromise and get sugar free chocolate with suspect artificial sweeteners? Maybe make a mousse just using cocoa?

What are your favourite comfort foods?

Mint Cinnamon’s second chance

13 Mar

This was going to be a post about how I liked mint/cinnamon, but couldn’t actually work in it. I couldn’t work in it as I couldn’t get the windows network share I store all my work on to automatically connect and couldn’t find an editor that I liked. Well I do like Kate, but it wouldn’t easily open (read drag and drop open) the files that were on the network drive that I couldn’t easily connect to.

But while writing, I thought “Why don’t I give it one more go”. And I magically got the network  drive to automatically connect. And now that it is connecting, Kate is opening the files I want. So linux is getting a good second chance now. Well even more than that, currently I have no real annoyances with it. For anyone else, and my own future reference, this is how I got it to work:

These are the instruction I followed to get the mount working. The issues I was having the first time is that I wasn’t using the GUI to first connect to the drive so that the password would be remembered. I wasn’t using the GUI as it wasn’t laid out in the way they explained. For me, I open up a file browser, File ->Connect to Server. From there you get a nice screen like:


You don’t want an FTP connection, you want a Windows Server one. When using gvfs-mount my command line was: smb://<domain>;<user name>@<server>/<server name>, so put those bits of information into the relative fields and hit conntect. With any luck it will connect. Then unmount and follow the rest of the instructions.



Bike Rage

12 Mar

I’m generally a bike rider or pedestrian. Not one of those lycra clad going out for some exercise riders, more I want to get places without taking an hour to walk there riders. Most of my riding is to and from uni, a nice and easy 10 minute ride.

For the most part I’m on a bike path. Pedestrians occasionally walk 2-3 abreast taking up the entire path, but when ringing my bell they move out of the way. On the parts where I’m on the roads, there cars are mostly fine. There are a few they over take me slightly closely, but it is rarely. These things don’t worry me.

What worries me is cars trying to be nice. They see me waiting to cross the road (this happens as both a pedestrian and cyclist), and think “Oh, I’ll be nice and let the person go through”. Here, in NSW, Australia, cars have right of way unless it is a marked pedestrian zebra crossing or the car is turning into a side street that has a pedestrian already crossing (well obviously they can’t run over pedestrians on the road, but the pedestrian shouldn’t be crossing). Bicycles are considered vehicles and should not be on foot paths, only marked bicycle paths. Obviously slightly different rules if you are on a major road. The problem with being “nice” is the cars behind the “nice car” has no idea that the person is trying to be “nice”, so they swerve around the “nice” car and nearly run into the pedestrian/cyclist.

Or what happened to me this morning. I was half way crossing a road waiting in the refuge island. The path on the opposite side of the road isn’t wheel accessible, so I turn into the road and then turn onto the path. A car stopped waiting for me to cross. Initially I didn’t move as they had right of way. The lady driving was waiving me through and I could see cars behind her getting annoyed (not enough room to over take here) I decided just to go. Just a bit further down the road where there is sort of room to over take (only sort of) she decided to over take me. Not safe, not a good idea.

So if you are car, please remember who has right of way. Being “nice” does no one any favours. If you are a cyclist or pedestrian,  don’t accept favours from cars, as there are other cars around that won’t appreciate it.


Introverts and Extroverts

8 Mar

I went to a Toastmasters meeting the other day. It was my first experience there. The table topics session was enjoyable (you are given a word and you need to talk for 50s to 60s on that word. The words were inspired by “My County” by Dorothea Mackellar). After some admin, there came the longer (5m to 7m) prepared speeches. The first was a research talk. Here the the talker has to research an area and give an informative talk based on what they had learnt. This talk was titled “Introverts and Extroverts”.

In this talk he kept to the traditional stereotypes of introverts (quiet, don’t like to talk to others) and extroverts (loud, want to talk to everyone). Then compared introverts to the stereotypical asperger (person? It isn’t a disease, so patient isn’t the right word, and I wouldn’t say they suffer. I digress). The example he kept on coming back to was that you were at a party. There is someone sitting in the corner not really interacting with  anyone. You go up to him, he doesn’t seem that interested in talking to you. That person is apparently an introvert. And it isn’t his fault that he doesn’t want to talk to you, he is an introvert, he can’t help himself and there is nothing wrong with that. But maybe if you talk to him about, say physics, he will suddenly talk to you and become the most enthusiastic person you have ever met.

This pissed me off

I do strongly identify as an introvert. But that doesn’t make me so socially inept that I can’t find someone to talk to at a party and hold a decent conversation at a party. I feel that an introvert is some one who gains energy when alone, loses it when surrounded. Conversely, an extrovert gains energy when surrounded by others, by loses it when by themselves. This really does fit me. I actually quite like being around other people, but my upper limit is about 3 social evenings each week. Maybe a bit higher if the people I’m socialising with are good friends. Of course, introversion/extroversion are just different ends of a continuum, not everyone is one or the other.

The fantastic article, Caring for Your Introvert, points out that for many introverts its not so much an issue with having a conversation, its just vapid conversations. Whats really the point? Yes, there is the social ritual you need to go through (“how are you going, plans for the weekend?”), but once you’ve acknowledged the person shown that you care about what they are doing, what else is there? Whats the point of participating in a circle jerk where you are all whinging or congratulating each other about the same thing? Go out and fix it and then you can talk about how you fixed it. If you have all done it, it isn’t worth the extended congratulations.

So just remember, introverts aren’t necessarily antisocial. If some one is not interested in talking to you, maybe they actually want to talk about SOMETHING, not your vapidity or just plain don’t like you. Introversion is not a social fault. Introverts aren’t all shy or socially inept. We probably are happier with our own company than with you.

Performance Profiles

7 Mar


You’ve made all your test data, ran all your tests and now all you have to do is say what solver or what set of parameters are the best (or at the least the best in what circumstances. Faced with a slab of data, albeit nicely put into a spreadsheet,  its not so easy to draw those conclusions. Humans, well at least this human, are fairly visual creatures. We don’t like looking at long lists of numbers but given a nice graph or diagram we can grasp the concept quite quickly.

Performance Profiles are one way to turn these slab of data into a pretty picture. On the horizontal profile you have some sort of performance metric, like time taken to run or the value of the best solution found by a solver. On the vertical axis you have the percentage of tests. You read the performance profile as saying “This solver found a solution in x time y% of the time” or “This solver found a solution with this objective gap, this amount of the time”. The best solution would look like a vertical line on the far left hand side of the graph, the worse will be a vertical on the far right hand side of the graph. For the most part, your best solution would be on the line that is in the top left hand corner the most.

The paper I linked to earlier says something slightly different to what I just did: on the horizontal axis it has the ratio to the best solution in each test case. Either one works, both have their problems. With what I described, if one test has a performance metric value around one, while another test has a value around 100, the performance profile is going to look somewhat stretched out. This is more of a matter of how do you interpret it. Just be careful. With comparing ratios, you will run into issues if your best value happens to be zero (cause you can’t divide by zero). To get around this you can add an epsilon to each value, but its a bit dodgy and as if you are comparing values 0 and 100 in one test, well you are going to get a crazy big value.

How to make performance profiles in a spreadsheet

Currently I’m booted up in Windows (Linux transition doesn’t happen when I haven’t worked out how to make latex tables out of Libre Office Calc yet), so this will be written for Microsoft Excel, but I imagine it would translate easily to other spreadsheets.

Your data

1) Get your data. Nothing special, just choose one performance profile you want to compare over. Each different solver should have their own column, and each test case should be in the same row

2) (optional, for ratios) Do you have any zeroes in your data? This is problematic as we are going to divide by the smallest or largest value in each row, and well, dividing by zero is a bad idea. If you have zeroes, replace them with a similarly small number, how small will depend on what the rest of your data looks like. I use the formula =IF(A2=0, 0.00001, A2) and a new set of columns to make life easier for me.

2) (optional, for ratios) We are now going to  find the ratios with the best solution. For each row (i.e. each test case), divide each value by the best value on that line. The best value could be the smallest value or the largest. As I’m looking for the solver that produces the smallest objective values, I want the best value is the smallest. To find the ratios, I use the formula =F2/MIN($F2:$I2) (change the cell references as needed) and create a new set of columns with the ratios.

3) We now want a list of each ratio/value that appears.To make this happen, I just copy and paste the values (not formulas) into one big long column, sort it and delete any duplicates I see. I like to them sort the data so that I can check for any obvious mistakes later. Normally I delete the ones as there are heaps of them. Duplicates don’t really matter in this case as they won’t affect our final result.

Performance Profile final data list

Performance Profile final data list

4) Now we want to find what percentage of tests have achieved a value of at least each ratio. I simple countif statement (to get the cumulative frequency) divided by the total number of tests (to get the percentage) to do the work. I use =COUNTIF(K:K, “<=”&$P2)/38. If you sorted the column in the previous step, you should see that each column is monotonically non-decreasing (i.e. getting bigger of staying the same)

5) We can now graph! Simply highlight your data columns and make a line graph. Now you have a performance profile from which you can easily make some observations about each of your solvers. From this one I can easily rank the solvers from solver 1 being the best, to solver 4 being the worse. Yes the graph I have is not a brilliant one: it doesn’t have a title or  any of the axis labelled, but it is good enough to get results. If I was publishing it I would tidy it up (especially that x-axis)

6) (optional) you may notice in the performance profile that when each line steps up it does so at an angle. Really we want it to go straight up. To do this, we add in some dummy points. So for each x value there should be two y values: one for the percentage of cases that reach a value of x (what you already have below) and the other percentage being the previous percentage. Admittedly I could get excel to play nicely with me here.


Really dodgy excel produced performance profile

Performance Profiles in Gnuplot

Yesterday when I started to write this, I initially tried to do it all in excel. In the end I couldn’t get the x axis to be on a sensible scale and couldn’t get step 6 working (the performance profile would look a bit like a heart rate monitor). So I turned to gnuplot. Really, so much easier.

1) Organise your data. Start with steps 1-4 in the spreadsheet instructions. When you have all your percentages organised copy and paste into a text file (although conventionally called a .dat file). Each column should be the percentage reached by the different solvers, with the first column being the x value reach. Well really it doesn’t matter where each of the columns are, but it is nice to keep to convention.

2) Make your .plt file. This is the file that contains all your plot commands. Mine looks like:

#Always put these two at the beginning as otherwise gnuplot might remember something without you realising

#set terminal png 
set terminal pdfcairo font "Gill Sans,9" linewidth 4
set output 'timePP.pdf'

set key bot right    #This puts the legend on the bottom right hand corner of the plot
#set logscale x 10    #This gives the x axis an log base 2 scale
set key autotitle columnheader #title columnhead take the name of the graph from the top of the y-axis column 
set termoption dashed    #I had to include this to get the differnet line styles sorted

set ylabel "%of instances"    #x-axis label
set xlabel "Time"    #y axis label

#This block sets different line styles, so later when I plot, I can just say "ls 1" (using linestyle 1) instead of writting all the style criteria
#set style line x ==  define line style x, which I'll later call using ls x
# lt y == set the line type as y. 1 is a solid line, 2 a dashed line, 3 a dotted line, 4 a dash dot line
#lc rgb "black"  == line colour is black. You can put in other line colours (like red, orange, yellow etc.), I'm not sure what the limits are
#lw set the line width
set style line 1 lc rgb "gray" lw 1
set style line 2 lc rgb "black" lw 1 
set style line 3 lc rgb "black" lw 1 
set style line 4 lc rgb "black" lw 1 

#Plot them. Each line is a sepearte line on the one plot. 
#DPL.dat is the data file we are getting the data from
#using a:b means data from column a is the x-axis, data from column b is the y axis
#With steps means we want vertical/horizontal steps, no diagonal lines between points
#ls x : use linestyle x
plot "time.dat" using 1:2 with steps ls 1,\
       "time.dat" using 1:3 with steps ls 2,\
       "time.dat" using 1:4 with steps ls 3,\
       "time.dat" using 1:5 with steps ls 4 
set output

And that gives me:


Of course it is easy enough to add in colours, but as I’m publishing this, no colours allowed.


So that is all it takes. Performance Profiles are convenient way to compare many cases to see which one is the best. Of course profiles don’t let you see why different solvers perform the way they do and in which circumstances, but they are a good first step.