translate algo 1 to julia pseudo code

[floswald]

• the pseudo code in algo 1 is general
• copy this and replace the generic functions with their julia equivalents
• Of course much depends on what kind of datatypes we settle on in the end, but just the get a start regarding the control flow
• use while, if else end etc
• post solution here below

[mariecases]

Here's the code to prevent from needing to open in a PDF each time

[mariecases]

There's something I don't understand about the basics. In the paper, M is the number of courses offered and K is the number of courses a student has to take. So function N takes the initial guess of vector p (of size Mx1) and generates K neighbors ? If so, why K ?

[mariecases]

So far, at line 14 with a lot of functions to be replaced. Issues I set aside that will need to be addressed :

• The big while loop that keeps track of computation time and stops when computation time exceeds the allotted time
• l.13 : If d is not in tau. I imagine this may be a source of excess computation time. Does anyone know of a function that finds the intersection of two arrays by colum ?

[mariecases]

So it's pretty close. There are just a couple of issues, but the two big ones from above remain.

As an idea for the do until currenttime > t, we could do something like the following :

c=0
rightnow = Dates.Time(now())
t = 10 #in seconds for now
while Dates.second((Dates.Time(now()) - rightnow)) < t
    #Above, (Dates.Time(now()) - rightnow) is of type Nanosecond. We need this 
    #to be a number type in seconds (or minutes, seconds to see if the loop works).
    c = c + 1
end

We aren't sure this would work, but as mentioned, we're having an issue with type here so we can't even test. Any ideas ?

[mdesevricourt]

What I propose for the while loop with current time:

c=0
starttime = Dates.Time(now())
t = 10 #in seconds for now
while (now() - starttime).value < t *1000 #as now() - starttime is in milliseconds
    c = c + 1
end

Marie, is that helping?

[mdesevricourt]

For the second issue, it really is a matter of what type \tau and d are gonna be.

I have an example here, but maybe there's something I am not seeing:
Let's say we have two courses and that for the moment, we have two demand vectors in our tabu list

d1 = [0,1]
d2 = [1,0]
tau = [d1, d2]
#we want to do something if another demand vector generated by a price vector, let's say d3, is not in tau
d3 = [1,1]
if !(d3 in tau)
   #do something 
end