[julia-users] Re: Julia Summer of Code
Your link is to a local address on your machine... On Sunday, May 24, 2015 at 12:42:27 AM UTC-4, Miles Lubin wrote: > > We've put up a blog post (http://localhost:4000/blog/2015/05/jsoc-cfp/) > with some more details on Julia Summer of Code. There are still a number of > unfilled slots open; submission deadline is June 1st! > > On Friday, May 15, 2015 at 1:57:24 PM UTC-4, Viral Shah wrote: >> >> Folks, >> >> The Moore Foundation is generously funding us to allow for 6-8 Julia >> Summer of Code projects. Details will be published soon, but if you are >> interested, please mark your calendars and plan your projects. >> >> -viral >> >
[julia-users] Re: Julia Summer of Code
Sorry about that, here's the real link: http://julialang.org/blog/2015/05/jsoc-cfp/ On Sunday, May 24, 2015 at 12:42:27 AM UTC-4, Miles Lubin wrote: > > We've put up a blog post (http://localhost:4000/blog/2015/05/jsoc-cfp/) > with some more details on Julia Summer of Code. There are still a number of > unfilled slots open; submission deadline is June 1st! > > On Friday, May 15, 2015 at 1:57:24 PM UTC-4, Viral Shah wrote: >> >> Folks, >> >> The Moore Foundation is generously funding us to allow for 6-8 Julia >> Summer of Code projects. Details will be published soon, but if you are >> interested, please mark your calendars and plan your projects. >> >> -viral >> >
[julia-users] Re: Julia Summer of Code
We've put up a blog post (http://localhost:4000/blog/2015/05/jsoc-cfp/) with some more details on Julia Summer of Code. There are still a number of unfilled slots open; submission deadline is June 1st! On Friday, May 15, 2015 at 1:57:24 PM UTC-4, Viral Shah wrote: > > Folks, > > The Moore Foundation is generously funding us to allow for 6-8 Julia > Summer of Code projects. Details will be published soon, but if you are > interested, please mark your calendars and plan your projects. > > -viral >
[julia-users] Avoid memory allocations when reading from matrices
Reposting this from Gitter chat since it seems this is more active.
I'm writing a GloVe module to learn Julia.
How can I avoid memory allocations? My main function deals with a lot of
random indexing in Matrices.
A[i, :] = 0.5 * B[i, :]
In this case* i* isn't from a linear sequence. I'm not sure that matters.
Anyway, I've done analysis and I know B[i, :] is the issue here since it's
creating a copy.
https://github.com/JuliaLang/julia/blob/master/base/array.jl#L309 makes the
copy
I tried to do it via loop but it looks like that doesn't help either. In
fact, it seems to allocate slight more memory which seems really odd.
Here's some of the code, it's a little messy since I'm commenting different
approaches I'm trying out.
type Model{T}
W_main::Matrix{T}
W_ctx::Matrix{T}
b_main::Vector{T}
b_ctx::Vector{T}
W_main_grad::Matrix{T}
W_ctx_grad::Matrix{T}
b_main_grad::Vector{T}
b_ctx_grad::Vector{T}
covec::Vector{Cooccurence}
end
# Each vocab word in associated with a main vector and a context vector.
# The paper initializes the to values [-0.5, 0.5] / vecsize+1 and
# the gradients to 1.0.
#
# The +1 term is for the bias.
function Model(comatrix; vecsize=100)
vs = size(comatrix, 1)
Model(
(rand(vecsize, vs) - 0.5) / (vecsize + 1),
(rand(vecsize, vs) - 0.5) / (vecsize + 1),
(rand(vs) - 0.5) / (vecsize + 1),
(rand(vs) - 0.5) / (vecsize + 1),
ones(vecsize, vs),
ones(vecsize, vs),
ones(vs),
ones(vs),
CoVector(comatrix), # not required in 0.4
)
end
# TODO: figure out memory issue
# the memory comments are from 500 loop test with vecsize=100
function train!(m::Model, s::Adagrad; xmax=100, alpha=0.75)
J = 0.0
shuffle!(m.covec)
vecsize = size(m.W_main, 1)
eltype = typeof(m.b_main[1])
vm = zeros(eltype, vecsize)
vc = zeros(eltype, vecsize)
grad_main = zeros(eltype, vecsize)
grad_ctx = zeros(eltype, vecsize)
for n=1:s.niter
# shuffle indices
for i = 1:length(m.covec)
@inbounds l1 = m.covec[i].i # main index
@inbounds l2 = m.covec[i].j # context index
@inbounds v = m.covec[i].v
vm[:] = m.W_main[:, l1]
vc[:] = m.W_ctx[:, l2]
diff = dot(vec(vm), vec(vc)) + m.b_main[l1] + m.b_ctx[l2] -
log(v)
fdiff = ifelse(v < xmax, (v / xmax) ^ alpha, 1.0) * diff
J += 0.5 * fdiff * diff
fdiff *= s.lrate
# inc memory by ~200 MB && running time by 2x
grad_main[:] = fdiff * m.W_ctx[:, l2]
grad_ctx[:] = fdiff * m.W_main[:, l1]
# Adaptive learning
# inc ~ 600MB + 0.75s
#= @inbounds for ii = 1:vecsize =#
#= m.W_main[ii, l1] -= grad_main[ii] /
sqrt(m.W_main_grad[ii, l1]) =#
#= m.W_ctx[ii, l2] -= grad_ctx[ii] / sqrt(m.W_ctx_grad[ii,
l2]) =#
#= m.b_main[l1] -= fdiff ./ sqrt(m.b_main_grad[l1]) =#
#= m.b_ctx[l2] -= fdiff ./ sqrt(m.b_ctx_grad[l2]) =#
#= end =#
m.W_main[:, l1] -= grad_main ./ sqrt(m.W_main_grad[:, l1])
m.W_ctx[:, l2] -= grad_ctx ./ sqrt(m.W_ctx_grad[:, l2])
m.b_main[l1] -= fdiff ./ sqrt(m.b_main_grad[l1])
m.b_ctx[l2] -= fdiff ./ sqrt(m.b_ctx_grad[l2])
# Gradients
fdiff *= fdiff
m.W_main_grad[:, l1] += grad_main .^ 2
m.W_ctx_grad[:, l2] += grad_ctx .^ 2
m.b_main_grad[l1] += fdiff
m.b_ctx_grad[l2] += fdiff
end
#= if n % 10 == 0 =#
#= println("iteration $n, cost $J") =#
#= end =#
end
end
Here's the entire repo https://github.com/domluna/GloVe.jl. Might be
helpful.
I tried doing some loops but it allocates more memory (oddly enough) and
gets slower.
You'll notice the word vectors are indexed by column, I changed the
representation to that
seeing if it would make a difference during the loop. It didn't seem to.
The memory analysis showed
Julia Version 0.4.0-dev+4893
Commit eb5da26* (2015-05-19 11:51 UTC)
Platform Info:
System: Darwin (x86_64-apple-darwin14.4.0)
CPU: Intel(R) Core(TM) i5-2557M CPU @ 1.70GHz
WORD_SIZE: 64
BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Sandybridge)
LAPACK: libopenblas
LIBM: libopenlibm
LLVM: libLLVM-3.3
Here model consists of 100x19 Matrices and 100 element vectors, 19 words in
the vocab, 100 element word vector.
@time GloVe.train!(model, GloVe.Adagrad(500))
1.990 seconds (6383 k allocations: 1162 MB, 10.82% gc time)
0.3 has is a bit slower due to worse gc but same memory.
Any help would be greatly appreciated!
[julia-users] Re: What wrong in read(STDIN,Int) ?
`read` reads binary, not text. 939577809079766321 = d0a0d0a0d0a0d31 in
hex, notice the last two digits (31) is the ASCII code for '1' 0d is
carriage return and 0a is line feed. Those are the binary values of what
you typed.
you should use text IO to read the line, eg `readline()` and then convert
to int, eg `x=int(readline())`
Cheers
Lex
On Sunday, May 24, 2015 at 3:35:03 AM UTC+10, paul analyst wrote:
>
> What wrong ? after typing "1" var = 939577809079766321
> (one ENTER is to small )
>
> julia> println("type any digit and ENTER");x=read(STDIN,Int)
> type any digit and ENTER
> 1
>
> 939577809079766321
> julia>
>
> julia> x
> 939577809079766321
>
> julia> println("type any digit and ENTER");x=read(STDIN,Int64)
> type any digit and ENTER
> 1
>
[julia-users] performance problem with factor
See https://github.com/JuliaLang/julia/pull/11189
[julia-users] Re: Question about modules and overloaded functions
Thanks for the help guys! The pointer to the discussion helped! It seems the most "Juilesque" way to do this is to drop the modules altogether. All the solutions suggested didn't feel very appealing. Gabe
Re: [julia-users] Re: How to stop the program and enter the value of the variable from the keyboard (by user)?
You are asking for a return value of type "Int", but have typed in the
character "1". Try instead:
x=chomp(readline(STDIN))
On Sat, May 23, 2015 at 7:31 PM paul analyst wrote:
> What wrong ? after typing "1" var = 939577809079766321
> (one ENTER is to small )
>
> julia> println("type any digit and ENTER");x=read(STDIN,Int)
> type any digit and ENTER
> 1
>
> 939577809079766321
> julia>
>
> julia> x
> 939577809079766321
>
> julia> println("type any digit and ENTER");x=read(STDIN,Int64)
> type any digit and ENTER
> 1
>
>
>
> 939577809079766321
>
>
> Paul
>
>
> W dniu sobota, 23 maja 2015 13:02:54 UTC+2 użytkownik paul analyst napisał:
>>
>> How to stop the program and enter the value of the variable from the
>> keyboard (by user)?
>> in other words:
>> How to choose variants of the program with the keyboard?
>>
>> for example:
>>
>> a=zeros(10);
>> for i=1:10
>> x=rand(1);y=rand(1)
>> if x[].>y[] a[i]=x[]
>> else a[i]=SOMTHIG FROM KEYBOARD (USER DECISION)
>> end
>>
>> Paul
>>
>
[julia-users] Best practice in julia for sorting an array of UInt64 or byte strings, by frequency
I know *how* to write this myself, as I would in C, but I'd like to know if one of the built-in datatypes (or a package) in Julia would handle this for me without me having to write much code... I will be building up an array or dict or whatever with a key, where need to keep track of number of occurrences and a flag that indicates whether it is present in group 1, group 2, or both... I then need to sort it first by the group (group 1 first, then present in both, then group 2), and within the groups by frequency... I am curious both about which approach in Julia would be the easiest and simplest to write (for when performance is not important, where I build these tables just once), and which approach would perform the best (on 0.4, I'm not interested in 0.3.x). [Maybe it is the same approach... that would be nice!] Thanks for any advice on this!
[julia-users] What wrong in read(STDIN,Int) ?
What wrong ? after typing "1" var = 939577809079766321
(one ENTER is to small )
julia> println("type any digit and ENTER");x=read(STDIN,Int)
type any digit and ENTER
1
939577809079766321
julia>
julia> x
939577809079766321
julia> println("type any digit and ENTER");x=read(STDIN,Int64)
type any digit and ENTER
1
[julia-users] What wrong in read(STDIN,Int) ??? Re: How to stop the program and enter the value of the variable from the keyboard (by user)?
What wrong in read(STDIN,Int) ???
W dniu sobota, 23 maja 2015 19:31:47 UTC+2 użytkownik paul analyst napisał:
>
> What wrong ? after typing "1" var = 939577809079766321
> (one ENTER is to small )
>
> julia> println("type any digit and ENTER");x=read(STDIN,Int)
> type any digit and ENTER
> 1
>
> 939577809079766321
> julia>
>
> julia> x
> 939577809079766321
>
> julia> println("type any digit and ENTER");x=read(STDIN,Int64)
> type any digit and ENTER
> 1
>
>
>
> 939577809079766321
>
> Paul
>
> W dniu sobota, 23 maja 2015 13:02:54 UTC+2 użytkownik paul analyst napisał:
>>
>> How to stop the program and enter the value of the variable from the
>> keyboard (by user)?
>> in other words:
>> How to choose variants of the program with the keyboard?
>>
>> for example:
>>
>> a=zeros(10);
>> for i=1:10
>> x=rand(1);y=rand(1)
>> if x[].>y[] a[i]=x[]
>> else a[i]=SOMTHIG FROM KEYBOARD (USER DECISION)
>> end
>>
>> Paul
>>
>
[julia-users] Re: How to stop the program and enter the value of the variable from the keyboard (by user)?
What wrong ? after typing "1" var = 939577809079766321
(one ENTER is to small )
julia> println("type any digit and ENTER");x=read(STDIN,Int)
type any digit and ENTER
1
939577809079766321
julia>
julia> x
939577809079766321
julia> println("type any digit and ENTER");x=read(STDIN,Int64)
type any digit and ENTER
1
939577809079766321
Paul
W dniu sobota, 23 maja 2015 13:02:54 UTC+2 użytkownik paul analyst napisał:
>
> How to stop the program and enter the value of the variable from the
> keyboard (by user)?
> in other words:
> How to choose variants of the program with the keyboard?
>
> for example:
>
> a=zeros(10);
> for i=1:10
> x=rand(1);y=rand(1)
> if x[].>y[] a[i]=x[]
> else a[i]=SOMTHIG FROM KEYBOARD (USER DECISION)
> end
>
> Paul
>
[julia-users] performance problem with factor
The factor function seems to be slow.
julia> @time factor(147573952589676412927) # 2^67-1; Cole
elapsed time: 4.955545116 seconds (6 MB allocated, 6.82% gc time in 1
pauses with 0 full sweep)
Dict{Int128,Int64} with 2 entries:
761838257287 => 1
193707721=> 1
This is even worse if a bigint is taken as argument (e.g. big(2)^67-1).
In contrast, here is the timing for the unix factor utility (instant)
$ time factor 147573952589676412927
147573952589676412927: 193707721 761838257287
real0m0.043s
user0m0.000s
sys 0m0.004s
Is there a way to speed up the factorisation with julia?
[julia-users] Re: seaching on multiple columns
I'm also going to plug for DataFramesMeta, especially https://github.com/JuliaStats/DataFramesMeta.jl#where. For your example, you could use @where df (:donor .== "Patient") & (:pregnant .== "yes") See also this related issue: https://github.com/JuliaStats/DataFramesMeta.jl/issues/13 On Saturday, May 23, 2015 at 11:09:21 AM UTC-4, Martin Somers wrote: > > Was trying something like this earlier > > df[df[:donor] .== "Patient" & df[:pregnant] .== "yes", :] > > but doesnt seem to work is this even possible or am I going to have sub > divid the operations > what the what most efficient syntax > > M >
Re: [julia-users] seaching on multiple columns
Le samedi 23 mai 2015 à 08:09 -0700, Martin Somers a écrit : > Was trying something like this earlier > > > df[df[:donor] .== "Patient" & df[:pregnant] .== "yes", :] > > > > but doesnt seem to work is this even possible or am I going to have > sub divid the operations > what the what most efficient syntax You just need to add parentheses like this: df[(df[:donor] .== "Patient") & (df[:pregnant] .== "yes"), :] By the way, DataFramesMeta offers convenience macros that you may find useful for this kind of operations: https://github.com/JuliaStats/DataFramesMeta.jl Regards
[julia-users] Re: Question about modules and overloaded functions
One such recent, long discussion can be found at
https://groups.google.com/forum/?hl=en#!topic/julia-users/sk8Gxq7ws3w.
I will also take this opportunity to plug shamelessly for my "package"
https://github.com/davidagold/MetaMerge.jl that allows you to merge
functions. For your example, so long as you have a function f defined in
the module from which you call fmerge!, you can do
fmerge!(f, (A, f), (B, f))
and now the function f (as defined in the module in which fmerge! is
called) will have both methods.
On Saturday, May 23, 2015 at 10:56:56 AM UTC-4, Gabriel Goh wrote:
>
> Hey all,
>
> I'm wondering if you guys could educate me on the correct way to overload
> functions in modules. The code here doesn't work:
>
> # **
>
> module A
>
> type apple
> end
>
> export apple,f
>
> f(x::A.apple) = println("I am apple");
>
> end
>
> # **
>
> module B
>
> type orange
> end
>
> export orange, f
>
> f(x::B.orange) = println("I am orange");
>
> end
>
> using A
> using B
>
> f(apple())
> f(orange())
>
> I expect julia to print "I am apple" and "I am orange", but instead I get
>
> ERROR: `f` has no method matching f(::apple)
>
> Any clues?
>
[julia-users] seaching on multiple columns
Was trying something like this earlier df[df[:donor] .== "Patient" & df[:pregnant] .== "yes", :] but doesnt seem to work is this even possible or am I going to have sub divid the operations what the what most efficient syntax M
Re: [julia-users] Question about modules and overloaded functions
Inside B you have to do:
B.f(x::B.orange) = println("I am orange")
If you can be bothered you can read through some recent, long
discussions on the merits of this approach.
(BTW, I think in Julia 0.4, your example should give a better error.)
On Sat, 2015-05-23 at 06:43, Gabriel Goh wrote:
> Hey all,
>
> I'm wondering if you guys could educate me on the correct way to overload
> functions in modules. The code here doesn't work:
>
> # **
>
> module A
>
> type apple
> end
>
> export apple,f
>
> f(x::A.apple) = println("I am apple");
>
> end
>
> # **
>
> module B
>
> type orange
> end
>
> export orange, f
>
> f(x::B.orange) = println("I am orange");
>
> end
>
> using A
> using B
>
> f(apple())
> f(orange())
>
> I expect julia to print "I am apple" and "I am orange", but instead I get
>
> ERROR: `f` has no method matching f(::apple)
>
> Any clues?
[julia-users] Question about modules and overloaded functions
Hey all,
I'm wondering if you guys could educate me on the correct way to overload
functions in modules. The code here doesn't work:
# **
module A
type apple
end
export apple,f
f(x::A.apple) = println("I am apple");
end
# **
module B
type orange
end
export orange, f
f(x::B.orange) = println("I am orange");
end
using A
using B
f(apple())
f(orange())
I expect julia to print "I am apple" and "I am orange", but instead I get
ERROR: `f` has no method matching f(::apple)
Any clues?
Re: [julia-users] Using macroexpand for non-debugging
For @nref there is actually the function _nref which makes the
expression, so no need to use macroexpand. I would suspect that is the
preferred way.
On Sat, 2015-05-23 at 00:33, Peter Brady wrote:
> I've noticed that a common pattern in my code is to use macroexpand inside
> functions/macros to generate expressions which are then further manipulated
> into what I need. As an example, here's a short function I wrote to unroll
> a finite difference stencil at a wall (which is then spliced into a larger
> function to compute the requested derivatives)
>
> julia> function unroll_left(st, idx, N=3)
>pd = quote end
>for j=1:size(st,2)
>ex = Expr(:call, :+)
>for i=1:size(st,1)
>solref = macroexpand(:(@nref $N sol
> d->(d==$idx?$i:i_{d})))
>push!(ex.args, :($solref*$(st[i,j])))
>end
>derref = macroexpand(:(@nref $N der d->(d==$idx?$j:i_{d})))
>push!(pd.args, :($derref = $ex))
>end
>pd
>end
>
> julia> a = rand(4,2)
> 4x2 Array{Float64,2}:
> 0.0352328 0.342554
> 0.898077 0.489546
> 0.933981 0.558188
> 0.721357 0.0678564
>
> julia> unroll_left(a, 1)
> quote
> der[1,i_2,i_3] = sol[1,i_2,i_3] * 0.035232761679327096 + sol[2,i_2,i_3]
> * 0.8980770465285655 + sol[3,i_2,i_3] * 0.9339813198913125 + sol[4,i_2,i_3]
> * 0.7213570230346098
> der[2,i_2,i_3] = sol[1,i_2,i_3] * 0.34255367557827077 + sol[2,i_2,i_3]
> * 0.4895461571970048 + sol[3,i_2,i_3] * 0.5581879223691961 + sol[4,i_2,i_3]
> * 0.06785635982311189
> end
>
> julia> unroll_left(a, 2)
> quote
> der[i_1,1,i_3] = sol[i_1,1,i_3] * 0.035232761679327096 + sol[i_1,2,i_3]
> * 0.8980770465285655 + sol[i_1,3,i_3] * 0.9339813198913125 + sol[i_1,4,i_3]
> * 0.7213570230346098
> der[i_1,2,i_3] = sol[i_1,1,i_3] * 0.34255367557827077 + sol[i_1,2,i_3]
> * 0.4895461571970048 + sol[i_1,3,i_3] * 0.5581879223691961 + sol[i_1,4,i_3]
> * 0.06785635982311189
> end
>
>
> This works perfectly but using macroexpand in this way feels a bit hackish.
> Is there a better way?
[julia-users] How to stop the program and enter the value of the variable from the keyboard (by user)?
How to stop the program and enter the value of the variable from the keyboard (by user)? in other words: How to choose variants of the program with the keyboard? for example: a=zeros(10); for i=1:10 x=rand(1);y=rand(1) if x[].>y[] a[i]=x[] else a[i]=SOMTHIG FROM KEYBOARD (USER DECISION) end Paul
