Basic Julia Syntax
- use basic syntax in Julia programming language
next week: stats
The goal today is to expose you to just those parts of Julia that you need to implement reservoir sampling for the most recent homework.
Announcements
- Iterative grading implemented
- See video describing how to access Canvas feedback
- Travis wins speed challenge with a time of 1:20! 👏
- I played around with some different instances, and wasn’t able to get past the network bottleneck of around 3:30 for previous homework.
123 GO - who is one person you admire?
Mistake in last lecture: The way I inferred a network bottleneck from looking at top cannot be exactly right.
If you have an explanation, I would love to hear it.
These things are still true:
- The network can be a bottleneck.
- Measure to see what the bottleneck is.
When I run the histogram pipeline on c4.xlarge with 4 vCPU’s it takes 3:30.
Here’s top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4049 ec2-user 20 0 44948 1236 1128 S 78.1 0.0 0:07.47 unpigz
4052 ec2-user 20 0 129340 8380 2060 R 60.8 0.1 0:05.94 sort
4050 ec2-user 20 0 114648 760 700 S 49.2 0.0 0:04.74 cut
4051 ec2-user 20 0 123636 1016 912 S 18.9 0.0 0:01.77 sed
4055 ec2-user 20 0 1255408 383336 11236 S 15.9 5.0 0:04.80 aws
2694 rngd 20 0 397100 4672 3832 S 0.3 0.1 0:24.59 rngd
When I run a simpler pipeline that simply counts lines (aws s3 cp ... - | unpigz | wc -l) on c4.xlarge it takes 1:30.
Here’s top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3829 ec2-user 20 0 44948 1320 1216 S 166.1 0.0 0:49.86 unpigz
3832 ec2-user 20 0 1262060 398784 11188 S 55.5 5.2 0:18.67 aws
3830 ec2-user 20 0 114644 724 660 R 17.9 0.0 0:05.39 wc
2694 rngd 20 0 397100 4672 3832 S 0.7 0.1 0:24.15 rngd
3852 ec2-user 20 0 171020 4484 3780 R 0.3 0.1 0:00.03 top
1 root 20 0 191044 5388 3980 S 0.0 0.1 0:01.56 systemd
Resources
Julia’s official website has tons of resources to help you learn. Find one that works for you and follow along with it, since we won’t be spending too much time exclusively on Julia in class. Share the ones you find helpful on Discord.
Content
Ask questions!
- Good: clarify conceptual differences between languages.
- Harder: really specific syntax or which function to use.
Modules organize objects in namespaces.
Use import X to use namespace lookups in module X.
import Random
Random.shuffle(1:5)
using X makes exported objects from the module X globally available.
using Random
shuffle(1:5)
Why import rather than using?
- Avoid name collisions if both modules export same name.
- Let’s you be more clear and specific - tells you what’s happening.
? enters the help menu to access builtin documentation.
rand(x) selects a random element from x.
rand([10, 20, 30])
rand(1:10)
When there are too versions of a function, f(x) and f!(x), the convention is that f!(x) mutates x, meaning it modifies x in place.
x = [10, 20, 30]
shuffle(x)
shuffle!(x)
Mutation is an example of a side effect. 123 GO - Do side effects make it easier or more difficult to understand code?
Answer- more difficult, particularly for parallel programming.
Arrays are the basic container in Julia.
x = [10, 20, 30]
typeof(x)
We can select and assign elements in arrays based on indexes.
Julia uses 1 based indexing.
x[2]
# update
x[2] = 200
1:n represents a sequence of integers from 1 to n.
y = 1:10
When used inside brackets [, end represents the last element of an array.
x[length(x)]
x[end]
x[2:end]
x[1:end]
# syntactic sugar
x[end-1]
x[length(x) - 1]
x[-1]
x[end-2]
end closes the end of syntactic blocks, like if statements.
if 1 < 2
println("sweet")
else
println("uh oh")
end
Simple functions return the value of the last expression.
function foo(x)
y = x + 1
y * 2 # <-- last expression
end
foo(10)
function foo2(x)
println("starting")
y = x + 1
return y * 2
println("all done")
end
Office hours
function set_first_to_100!(x)
x[1] = 100
end
x = [10, 20, 30]
set_first_to_100!(x)
Exercise
See Canvas quiz