An Introduction to the Julia Programming Language¶
Varialbles and Types¶
In [1]:
56*8
Out[1]:
In [2]:
x=7
Out[2]:
In [3]:
x+1
Out[3]:
In [4]:
x^2
Out[4]:
In [5]:
typeof(x)
Out[5]:
In [6]:
x/3
Out[6]:
In [7]:
cos(pi)
Out[7]:
In [8]:
typeof(ans)
Out[8]:
In [9]:
im^2
Out[9]:
In [10]:
typeof(ans)
Out[10]:
In [11]:
"Hello" * " world"
Out[11]:
In [12]:
typeof(ans)
Out[12]:
In [13]:
70//15
Out[13]:
In [14]:
true
Out[14]:
In [15]:
12 ≠42
Out[15]:
In [16]:
2 ∈ [6,3,5]
Out[16]:
In [17]:
typeof(ans)
Out[17]:
In [18]:
"Like in perl you may write x=$x"
Out[18]:
In [19]:
"... but also $(exp(x))"
Out[19]:
Type Hierarchy¶
The type hierarchy for numbers looks like:
Functions + Multiple Dispatch¶
One can define functions simply like that:
In [20]:
f(x) = x ^ 3
Out[20]:
In [21]:
f(3)
Out[21]:
In [22]:
f("ha ")
Out[22]:
In [23]:
function f(x::AbstractString)
println("\nnow in f(x::AbstractString) with x=$x of type $(typeof(x))")
"Hello $x"
end
function f(x::AbstractFloat)
println("\nnow in f(x::AbstractFloat) with x=$x of type $(typeof(x))")
x+1.0
end
Out[23]:
In [24]:
methods(f)
Out[24]:
In [25]:
@show f(2.5);
In [26]:
@show f( Float16(2.5) );
In [27]:
@show f("abc")
@show f("αβγ");
In [28]:
methods(print)
Out[28]:
In [29]:
methods(println)
Out[29]:
Composite Types¶
In [30]:
type Point
x::Int
y::Int
name::ASCIIString
end
In [31]:
p = Point(2,3,"Siegfried")
Out[31]:
In [35]:
function f(p::Point)
println("\nnow in f(x::Point) with p=$p of type $(typeof(p))")
p.x = p.x + p.y
return p.name
end
Out[35]:
In [36]:
methods(f)
Out[36]:
In [37]:
f(p)
Out[37]:
In [38]:
p
Out[38]:
parametric types and methods¶
In [39]:
type Line{T}
x0::T
y0::T
x1::T
y1::T
name::ASCIIString
end
In [40]:
L1=Line(23,3,4,44,"Julia")
Out[40]:
In [41]:
L2=Line(23.777,3.5,4.1,4.0,"Rolf")
Out[41]:
In [42]:
function f{T}(x::Line{T})
println("\nnow in f(x::Line{T}) with x=$x of type $(typeof(x))")
x.name
end
Out[42]:
In [43]:
f(L1),f(L2)
Out[43]:
Arrays¶
In [44]:
A=[1,2,9]
Out[44]:
In [45]:
A == 2
Out[45]:
In [46]:
A .== 2
Out[46]:
In [47]:
A+2.1
Out[47]:
In [48]:
B=[1 2 3; 6 7 4; 6 2 1]
Out[48]:
In [49]:
C=[ i*j for i in 1:3, j in 1:4 ]
Out[49]:
In [50]:
B * A
Out[50]:
In [51]:
#B =rand(500,500)
@time eigvals(B)
Out[51]:
In [52]:
@time svd(B)
Out[52]:
Dictionary¶
In [53]:
d = Dict( "A"=>6, "B"=> 20)
Out[53]:
In [54]:
d["A"]
Out[54]:
In [55]:
for (key,value) in d
print("$key ----> $value\n")
end
Regular Expressions¶
In [56]:
my_regex = r"[ABC]"
@show ismatch(my_regex,"Foo")
@show my_regex("Bar")
@show my_regex("boo");
In [57]:
my_regex =r"(\d+)g"
match(my_regex,"take 500g sugar")
Out[57]:
Flow Control¶
In [58]:
for i in 1:5
println(i)
end
In [59]:
for i in 1:10
if i % 3 == 0
println(i)
end
end
Packages¶
Lots of packages are available - see http://pkg.julialang.org for a list.
To view all locally installed packages:
In [60]:
Pkg.status()
To add a package (and its dependencies):
In [ ]:
Pkg.add("DataFrames")
To update installed packages to the last release:
In [ ]:
Pkg.update()
To stay on the top of the master branch:
In [ ]:
Pkg.checkout("DataFrames")
In [ ]:
#Pkg.generate("someName")
DataFrames¶
In [61]:
using DataFrames
d = DataFrame(x = 21:30, y = rand(10))
Out[61]:
In [62]:
mean(d[:x]), mean(d[:y])
Out[62]:
In [63]:
using StatsBase
cov(d[:x], d[:y])
Out[63]:
In [64]:
using RDatasets
iris = dataset("datasets", "iris")
@show size(iris)
head(iris)
Out[64]:
In [65]:
by(iris, :Species, size)
Out[65]:
In [66]:
using DimensionalityReduction
iris_S = convert(Array,DataArray(iris[:,1:4]))
Xpca = pca(iris_S)
Out[66]:
Graphics - PlotlyJS¶
In [67]:
using PlotlyJS
using Colors
using Blink
#using Rsvg
#init_notebook()
In [68]:
nms = unique(iris[:Species])
colors = [RGB(0.89, 0.1, 0.1), RGB(0.21, 0.50, 0.72), RGB(0.28, 0.68, 0.3)]
data = GenericTrace[]
for (i, nm) in enumerate(nms)
sc = Xpca.scores[iris[:Species] .== nm, :]
x=sc[:,1]
y=sc[:,2]
trace = scatter(;name=nm, mode="markers",
marker_size=10, marker_color=colors[i], marker_line_width=0,
x=x, y=y)
push!(data, trace)
end
layout = Layout(width=800, height=550, autosize=false, title="Iris dataset",
xaxis=attr(title="PC1"),
yaxis=attr(title="PC2")
)
p1=PlotlyJS.plot(data, layout)
Out[68]:
In [69]:
typeof(p1)
Out[69]:
In [ ]:
#PlotlyJS.savefig(p1, "output_filename.pdf")
In [70]:
function clustering_alpha_shapes()
@eval using DataFrames, RDatasets, Colors
# load data
iris = dataset("datasets", "iris")
nms = unique(iris[:Species])
colors = [RGB(0.89, 0.1, 0.1), RGB(0.21, 0.50, 0.72), RGB(0.28, 0.68, 0.3)]
data = GenericTrace[]
for (i, nm) in enumerate(nms)
df = iris[iris[:Species] .== nm, :]
x=df[:SepalLength]
y=df[:SepalWidth]
z=df[:PetalLength]
trace = scatter3d(;name=nm, mode="markers",
marker_size=3, marker_color=colors[i], marker_line_width=0,
x=x, y=y, z=z)
push!(data, trace)
cluster = mesh3d(;color=colors[i], opacity=0.3, x=x, y=y, z=z)
push!(data, cluster)
end
# notice the nested attrs to create complex JSON objects
layout = Layout(width=800, height=550, autosize=false, title="Iris dataset",
scene=attr(xaxis=attr(gridcolor="rgb(255, 255, 255)",
zerolinecolor="rgb(255, 255, 255)",
showbackground=true,
backgroundcolor="rgb(230, 230,230)"),
yaxis=attr(gridcolor="rgb(255, 255, 255)",
zerolinecolor="rgb(255, 255, 255)",
showbackground=true,
backgroundcolor="rgb(230, 230,230)"),
zaxis=attr(gridcolor="rgb(255, 255, 255)",
zerolinecolor="rgb(255, 255, 255)",
showbackground=true,
backgroundcolor="rgb(230, 230,230)"),
aspectratio=attr(x=1, y=1, z=0.7),
aspectmode = "manual"))
PlotlyJS.plot(data, layout)
end
clustering_alpha_shapes()
Out[70]:
In [71]:
using RCall
In [72]:
R"rnorm(10)"
Out[72]:
In [73]:
L = 8
@rput L
R"v <- rnorm(L)"
@rget v
v
Out[73]:
In [74]:
rcopy(R"t.test(rnorm(1000))")
Out[74]:
In [75]:
ccall( (:clock, "libc"), Int32, ())
Out[75]:
Calling external programs¶
In [76]:
run(`echo Hello`)
In [77]:
file = "HelloWorld.jl"
run(`cat $file`)
BioJulia¶
have a look here https://github.com/BioJulia
In [78]:
using Bio.Seq
seq = dna"ACGTTT"
Out[78]:
In [79]:
reverse_complement(seq)
Out[79]:
In [80]:
seq = dna"ACAGCGTAGCT";
@show approxsearch(seq, dna"AGGG", 0)
@show approxsearch(seq, dna"AGGG", 1)
@show approxsearch(seq, dna"AGGG", 2);
Vectorization¶
In [81]:
L = 100000000
A = rand(1:1000,L)
B = rand(1:1000,L)
function sum1!(A,B)
for i in 1:length(A)
A[i] = A[i]+B[i]
end
A
end
function sum2!(A,B)
@simd for i in 1:length(A)
@inbounds A[i] = A[i]+B[i]
end
A
end
@time A+=B;
@time sum1!(A,B)
@time sum2!(A,B)
;
Parallel Computing - Multiple Processors¶
In [82]:
procs()
Out[82]:
In [83]:
addprocs(2)
procs()
Out[83]:
In [84]:
r = remotecall(2, rand, 2, 2)
Out[84]:
In [85]:
fetch(r)
Out[85]:
In [ ]:
#addprocs(fill("ergophobie.molgen.mpg.de",10),dir="/home/arndt",exename="/usr/bin/julia")
nheads() = @parallel (+) for i=1:2000000000
Int(rand(Bool))
end
@time nheads()
Graphics - Plots¶
In [86]:
using Plots
#plotlyjs()
gr()
#unicodeplots()
Out[86]:
In [87]:
Plots.plot(rand(10,5),w=4)
Out[87]:
Probabilistic Programming¶
Tasks¶
offer python's yield like behavior
In [ ]:
function producer(max::Int = 4)
for i in 1:max
produce(i)
end
end
for i in Task(producer)
println(i)
end
In [ ]:
for i in Task(() -> producer(2))
println(i)
end