r)[1:end]
> Aptrs = Array{Ptr{eltype(ar)}}(sz)
> for I in CartesianRange(sz)
> Aptrs[I] = pointer(ar, sub2ind(size(ar), 1, I.I...))
> end
> println(ccall(("test","test"),Float64,(Ptr{Ptr{Ptr{Ptr{Float64,),
> Aptrs))
>
>
> something is wrong here...
>
>
&
ccall(sym, Void, (Ptr{Ptr{T}},), Aptrs)
When you pass an array via `ccall`, it takes the pointer of the data, which
is why this produces a Ptr{Ptr{T}} from an Array{Ptr{T}}.
Best,
--Tim
On Fri, Nov 4, 2016 at 9:14 AM, Alexander Lyapin wrote:
> but in C i have
If you need the pointers to each lower-dimensional slice, you should be
able to build it like so (warning: untested):
```
sz = size(A)[2:end] # if you need type-stability, use Base.tail(size(A))
instead
Aptrs = Array{Ptr{eltype(A)}(sz)
for I in CartesianRange(sz)
Aptrs[I] = pointer(A,
Hmm, that's surprising. Looks like we're using generic broadcasting
machinery for that operation (check out what @which P.*P returns). Might be
good to add .* to this line:
https://github.com/JuliaLang/julia/blob/b7f1aa7554c71d3759702b9c2e14904ebdc94199/base/arraymath.jl#L69.
Want to make a pull
(i.e., the loss of a simple model
> of how things are laid out). Maybe my fears are unfounded. Others don't
> seem concerned it would seem.
>
> I'll check out those packages that you mention.
>
> Thanks,
> Bob
>
> On Sun, Oct 30, 2016 at 2:29 PM, Tim Holy <tim.h...@g
abstract tree.
>
> Does this help?
>
> Bob
>
> ps I don't have any concrete examples yet because OffsetArrays are so
> scarce in the julia ecosystem. I'm just looking ahead to a world where it
> will no longer be possible to assume AbstractArrays are OneTo. This seems
>
I should add, abstract discussions are all well and good, but I fully
recognize you may have something very specific in mind and my answer simply
fails to understand your concern. If you can give one or more concrete
examples of where the current system is either bug-prone or a pain in the
At least with the non-1 arrays I know of now, you can always get the OneTo
array form with parent(A). However, that's not something I feel comfortable
saying will be guaranteed forever.
In terms of the "automatic conversion" part of your proposal, I'm a bit
uncomfortable because in my view the
Yes, it should be pretty responsive now---ImageView doesn't have to go
through a Python layer, so I think it's a bit snappier. It's been a while
since I've used Matlab, but at least when I first compared them ImageView
was considerably better than Matlab for interactive display of movies, etc.
But
> Why not use dispatch instead?
Because subtyping isn't powerful enough for all needs. For example:
julia> using Unitful
julia> const mm = u"mm"
mm
julia> isa(3.2mm, AbstractFloat)
false
You'd probably like to use the fancy logic of `FloatRange` if you're
constructing a range
+1. We need number traits for a variety of circumstances; I was also
contemplating them as a step in generalizing the FloatRange/StepRange
distinction, for example to Unitful numbers (numbers with physical units).
You need type-stability for the created range object, so I think a trait is
the only
That document is aimed at developers to tell them how to make their package
ready for arrays that have custom indices. As a user, the key line is:
> Such array types are expected to be supplied through packages.
I recommend the OffsetArrays package.
Great to see someone interested in trying
Don't look at me, I swear it was just a simple change in the return type of
a function---they should have used a parametric type definition. Another
$1billion down the drain...
Best,
--Tim
On Wed, Oct 19, 2016 at 11:50 PM, Michele Zaffalon <
michele.zaffa...@gmail.com> wrote:
> Tim Ho
julia> a = bitrand(3,5)
3×5 BitArray{2}:
true false false true true
false true true true false
true true true true true
julia> Int.(a)
3×5 Array{Int64,2}:
1 0 0 1 1
0 1 1 1 0
1 1 1 1 1
julia> convert(Array{Int}, a)
3×5 Array{Int64,2}:
1 0 0 1 1
0 1 1
Hmm, I fixed this on the images-next branch but forgot to fix it on master.
Will be fixed in https://github.com/timholy/Images.jl/pull/564.
If you're just getting started with Julia & Images, you might want to
consider using the upcoming version of Images. See
For simple cases there's also the IndirectArrays package:
https://github.com/JuliaArrays/IndirectArrays.jl
Best,
--Tim
On Fri, Sep 30, 2016 at 10:51 PM, wrote:
> OK! many thanks for your fast reply
> --a
>
>
> On Friday, September 30, 2016 at 3:53:43 PM UTC+2,
No real clue about what's happening, but my immediate thought was that if
your algorithm is iterative and uses some kind of threshold to decide
convergence, then it seems possible that a change in the accuracy of some
computation might lead to it getting "stuck" occasionally due to roundoff
error.
Try explaining both indexing behaviors to a newcomer and you'll see the
difference.
Old behavior: `3:3` causes the dimension to be retained; `3` causes the
dimension to be dropped if it's a 'trailing dimension' (all the later
indices are also scalars) but retained if it's a 'leading dimension'
You don't have to do this in the type definition, you can do it at
introspection time:
fieldtype(Foo{Float64,3},:a) returns Array{T,N}
fieldtype(Foo{Float64,3},:b) returns Array{Float64,3}
More generally:
julia> T, N = TypeVar(:T, true), TypeVar(:N, true)
(T,N)
julia> fieldtype(Foo{T,N},
See ntoh and hton. Perhaps even better, see StrPack.jl.
--Tim
On Sunday, September 18, 2016 1:13:43 AM CDT Femto Trader wrote:
> Hello,
>
> I'd like to read this file
> http://www.dukascopy.com/datafeed/EURUSD/2016/02/14/20h_ticks.bi5
> using Julia.
>
> It's a LZMA compressed file.
>
> I can
On Sunday, September 18, 2016 2:55:46 AM CDT K leo wrote:
> I have been using simply A=[0; A[1:end-1]], but found it to be somehow
> quite expensive. I saw that there is unshift! but it has to be followed up
> with deleteat! to make the array the same size, i.e. there need to be two
> operations.
Yes, the idea of wrappers that reinterpret or convert makes sense.
On Friday, September 16, 2016 9:29:33 AM CDT Páll Haraldsson wrote:
> type Color
> r::Uint8
> g::Uint8
> b::Uint8
> a::Uint8
RGBA{U8} is bitwise identical to this, so you can use reinterpret as needed.
--Tim
https://github.com/JuliaLang/julia/issues/18524
On Wednesday, September 14, 2016 9:15:45 AM CDT Marius Millea wrote:
> Is this the expected behavior?
>
> julia> mapslices(x->tuple(x), [1 2; 3 4], 1)
> 1×2 Array{Tuple{Array{Int64,1}},2}:
> ([2,4],) ([2,4],)
>
> julia> mapslices(x->tuple(x...),
Which version of julia? If you're not using 0.5, try it and you might be
pleased.
You can also launch `julia --inline=no`, which occasionally still remains
useful.
--Tim
On Tuesday, September 13, 2016 8:55:58 AM CDT Tim Wheeler wrote:
> Hi Julia Users,
>
> So I was looking at
nt? (Note that for both of them, the result is
2-dimensional.)
--Tim
On Monday, September 12, 2016 6:47:04 PM CDT Neal Becker wrote:
> I haven't studied it, but I guess that newaxis increases the dimensionality,
> while specifying 0 for the stride. Can reshape do that?
>
> Tim Holy wrote:
I'm not certain I understand what `np.newaxis` does, but doesn't `reshape` do
the same thing? (newaxis does look like a convenient way to specify shape,
though.)
Best,
--Tim
On Monday, September 12, 2016 3:28:56 PM CDT Neal Becker wrote:
> Some time ago I asked this question
>
There are some great resources at http://julialang.org/learning/
Best,
--Tim
On Monday, September 12, 2016 7:56:46 AM CDT
rishucod...@gmail.com wrote:
> Thanks for the help. Can you suggest me what should I learn to
work in
> Julia?
Had not seen that, thanks for sharing.
For this month's survey at least, passing Clojure, within sight of Rust, and
nearly 1/3 of the way to Fortran. Pretty impressive. Nice work, Julia
community!
Best,
--Tim
On Sunday, September 11, 2016 12:52:21 AM CDT Viral Shah wrote:
> Perhaps many of
Every time julia allocates memory, it adds the number of bytes to a running
tally; once that
tally gets big enough, it does a garbage-collection pass (and resets the
tally).
Consequently GC is triggered only when you allocate, but even the smallest
allocation can
trigger GC if it pushes the
More likely you just need to run Pkg.update(). For example, you can test with a
"dummy"
package repository:
$ mkdir /tmp/pkgs
* _*
* _ _ _(_)_ | A fresh approach to technical computing*
* (_) | (_) (_)| Documentation: http://docs.julialang.org*
* _ _ _|
Try putting a semicolon at the end of that line.
Thanks for the bug report! https://github.com/timholy/Images.jl/issues/554
--Tim
On Friday, September 9, 2016 5:18:45 PM CDT Drew Mahedy wrote:
> In Julia, if I try and concatenate two images together of type:
>
>
https://github.com/JuliaLang/julia/issues/18419
--Tim
On Wednesday, September 7, 2016 12:59:05 AM CDT Lutfullah Tomak wrote:
> A reduced case that also makes multiplication of the same Irrational an
> error.
>
>_
>_ _ _(_)_ | A fresh approach to technical
On Thursday, September 8, 2016 6:31:37 AM CDT Páll Haraldsson wrote:
> Not sure I understand ("this case"), in another thread Tim Holy cautioned
> that C++ would not be inlined into Julia functions, as Julia functions can
> be.
My caution was for the opposite direction: you
Keep in mind that statements like "very little overhead" depend entirely on
what you're comparing against. Your TestFunc is quite expensive, so it's not
surprising that how it's called adds little overhead. If you called a much
cheaper function, you might come to a rather different conclusion.
https://github.com/JuliaIO/JLD.jl/blob/master/doc/jld.md#custom-serialization
On Wednesday, August 31, 2016 1:56:25 AM CDT Kristoffer Carlsson wrote:
> I am sure you know but I just want to point out that it is possible to
> overload the way JLD saves and loads types. You can write a custom save
On Tuesday, August 30, 2016 3:57:36 PM CDT Yichao Yu wrote:
> And even then, this completely looses the advantage of using tuple
> (inferrable size and element types) so you shouldn't do this in general
> unless you are going to do a lot of work with the tiny tuple afterwards.
Right. If you want
I'm rather surprised that building from source gave you version 0.4.7; I don't
think that makes sense (it should have been master). Are you sure you know
what you're running? (Try `.julia` from the directory in which you built
julia, rather than just `julia`.)
Best,
--Tim
On Monday, August
I confess I'm not quite sure what the right answer is here. It would seem
overkill to have both `I` and something that's the same thing, except sized.
OTOH, I see the attraction.
Maybe if it were part of a general mechanism, e.g., SizedArrayOperator{O,N}
(for performing an operation `O` on
eads
> @timeit test(3000,40)
> 1 loops, best of 3: 3.84 s per loop
>
> # without @threads
> @timeit test(3000,40)
> 1 loops, best of 3: 2.33 s per loop
>
> On Monday, August 29, 2016 at 6:50:15 PM UTC+2, Tim Holy wrote:
> > Very quickly (train to catch!): try this
>
Very quickly (train to catch!): try this https://github.com/JuliaLang/julia/
issues/17395#issuecomment-241911387
and see if it helps.
--Tim
On Monday, August 29, 2016 9:22:09 AM CDT Marius Millea wrote:
> I've parallelized some code with @threads, but instead of a factor NCPUs
> speed
On Monday, August 29, 2016 10:40:10 AM CDT Júlio Hoffimann wrote:
> Why would one want dense identity matrix?
Because you might want it to serve as an initialization for an iterative
optimization routine (e.g., ICA) that updates the solution in place, and which
assumes a dense matrix?
We could
To rephrase what Steven and Tony said, for some things you won't need a macro.
For example, `unsafe_wrap` didn't exist on Julia 0.4, but Compat contains an
implementation of `unsafe_wrap` for use on Julia 0.4. It's just a plain-old
function call, so you don't need `@compat`---just use it in
Or perhaps make the package Deconvolution.jl, and have wiener be the first (and
currently only) method?
Best,
--Tim
On Sunday, August 28, 2016 12:44:44 PM CDT Mosè Giordano wrote:
> Hi all,
>
> I wrote a very simple implementation of the Wiener deconvolution
>
If you can't find another home, I'd be happy to have it in Images, but to me
DSP seems like the (slightly) better choice.
That said, I also don't think it's terrible to have small packages (they are
easier to document and faster for newcomers to come to grips with), so it
could also stay a
For unstructured grids, it depends a lot on what you want. I'm a fan of
piecewise linear polyhedral interpolation, but there are many other choices:
https://en.wikipedia.org/wiki/Multivariate_interpolation#Irregular_grid_.
28scattered_data.29. One of the (many) Voronoi/Delaunay packages should
julia> qn = :(:n)
:(:n)
julia> ex = :("My docstring about n, which has value $($qn))")
:("My docstring about n, which has value $(:n))")
julia> dump(ex)
Expr
head: Symbol string
args: Array{Any}((3,))
1: String "My docstring about n, which has value "
2: QuoteNode
value:
Good catch. Looks like the edge-handling in `findlocalmaxima` needed to be a
bit more refined---it was discarding results from the first and last sigma-
values supplied by the user.
I may have fixed this in https://github.com/timholy/Images.jl/commit/
7336f35c824b15de9e4d0def8e739bdeb6ed3b3d,
You must be using Julia 0.5, where `view` means something different than
you're expecting (it creates a SubArray). Maybe try `display(img)`.
Best,
--Tim
On Friday, August 26, 2016 1:31:46 PM CDT Bill Maier wrote:
> I have the following code in an IJulia notebook(). The image is saved
>
enyEdge}()
>confidence!(x, -1.0)
>return x
>end
> end
>
> And define getters and setters in such a way that type assertions make
> things certain for the compiler?
> I saw that Jeff proposed a similar solution in julia issue #269 to handle
> circular type de
What have you tried so far? See http://docs.julialang.org/en/latest/manual/
performance-tips/#tools, esp. @code_warntype.
--Tim
On Wednesday, August 24, 2016 1:55:23 PM CDT Venil Noronha wrote:
> The following code seems to consume a lot of memory. Could anyone spot what
> I'm doing wrong here?
I don't think that's type-stable. Since each node of each tree will also be a
different type, I also think you'll end up hating life due to compile times.
There's some (peripherally) relevant discussion at http://docs.julialang.org/
I haven't inspected anything in detail, but I agree with Kristoffer that my
first impression is that I really like the new look. Nice work, folks.
Best,
--Tim
On Monday, August 22, 2016 1:30:39 PM CDT Kristoffer Carlsson wrote:
> Just from looking at the generated docs for the Documenter
Perhaps https://github.com/JuliaComputing/NDSparseData.jl?
--Tim
On Sunday, August 21, 2016 8:14:48 AM CDT Kristof Cools wrote:
> Just wondering whether there have emerged any packages for this in the
> meantime. I need a rank 3 sparse matrix to implement a retarded potential
> integral equation
Not an answer to your question (I don't have time to check now), but in case
it helps perhaps FFTViews may be a useful model. (The code is shorter than the
README :-).)
https://github.com/timholy/FFTViews.jl
I think getting its tests to pass requires what will become julia-0.5+rc3
(currently
For what it's worth, long ago I was caught by that once, too.
To fix it, I think all one would have to do is make the doc badges 2x larger
(if possible).
--Tim
On Sunday, August 14, 2016 1:22:26 AM CDT Kristoffer Carlsson wrote:
> What did you expect them to be? They are right under a big
AFAICT, it remains possible to do dynamic type generation if you (1) print the
code that would define the type to a file, and (2) `include` the file.
function create_type_dynamically{T}(::Type{T})
type_name = string("MyType", T)
isdefined(Main, Symbol(type_name)) && return nothing
I'd recommend starting by picking a very small project. For example, fix a bug
or implement a small improvement in a package that you already find useful or
interesting. That way you'll get some guidance while making a positive
contribution; once you know more about julia, it will be easier to
There are several previous threads on this, hopefully one of those will help.
Best,
--Tim
On Monday, August 8, 2016 8:13:42 PM CDT Chang Kwon wrote:
> Is there a way to precompile all packages at once? Each time that I run
> Pkg.update(), I would also like to precompile all packages so that when
import Base: !
also works (and you can include more operators).
--Tim
On Monday, August 8, 2016 9:59:25 PM CDT Fengyang Wang wrote:
> On Monday, August 8, 2016 at 10:26:46 AM UTC-4, Kevin Squire wrote:
> > Try
> >
> > import Base.(!)
> >
> > Cheers,
> >
> > Kevin
>
> Why do import
Aside from checking out the package's master, a glance at the list of open PRs
and recent commits usually tells you a lot.
Best,
--Tim
On Thursday, August 4, 2016 7:23:11 PM CDT Chris Rackauckas wrote:
> Not always. Some repos have a dev branch or a v0.5 compatibility branch,
> etc. I think
On Wednesday, August 3, 2016 6:34:51 AM CDT Yichao Yu wrote:
> I guess I could register it after some clean up later, given that we had
> FastAnonymous package (which was in a very similar situation, if not
> slightly more undefined than this one, sorry @timholy ;-p )
Hey, no sweat. I agree that
module MyMetaPackage
using Reexport
@reexport using PackageA
@reexport using PackageB
...
end
Best.
--Tim
On Monday, August 1, 2016 1:48:47 AM CDT Steven Sagaert wrote:
> is more than just a webpage with a list of packages... for starters the
> concept of metapackage.
>
> On Monday, August
t that a bit un-Julianic? Or
> are you worried that people will then start using (e.g.) 1:len(A) instead
> of 1:length(A), resulting in an even bigger mess?
>
> On Tuesday, July 26, 2016 at 1:47:57 PM UTC+2, Tim Holy wrote:
> > On Tuesday, July 26, 2016 2:31:10 AM CDT Oliver Schulz
OK, new package here:
https://github.com/timholy/EndpointRanges.jl
It might make life easier even for folks who aren't using custom-indexed
arrays. Give it a whirl and see what breaks!
Best,
--Tim
On Tuesday, July 26, 2016 2:31:10 AM CDT Oliver Schulz wrote:
> Hi Tim,
>
> I should have known
On Tuesday, July 26, 2016 2:31:10 AM CDT Oliver Schulz wrote:
> I should have known you already have a (very elegant) solution in your
> pocket. :-) I really like your proposed first:last syntax!
As discussed in that issue, the problem is that it may not be workable (it's
also very unlikely to
Hi Oliver,
On Tuesday, July 26, 2016 12:21:27 AM CDT Oliver Schulz wrote:
> Will there by something like
>
> A[start+1:end-1]
That would be lovely. See discussion in https://github.com/JuliaLang/julia/
pull/15750.
> length(linearindices(A))
>
> instead of length() in generic code. This
Given the apparent interest in the topic and the decisions that people seem to
be making, it seems worth pointing out that folks are still using apples-to-
oranges comparisons on this benchmark.
There are at least two important differences:
- in the other languages, `linspace` allocates a
As long as you don't mind preserving exactly what's between the {},
(t.parameters...)
is an easy way to get this.
--Tim
On Sunday, July 24, 2016 5:21:56 AM CDT Kristoffer Carlsson wrote:
> Maybe;
>
> type MyType end
>
> function f(t::DataType)
> I = ()
> for t in t.types
> if isa(t,
On Wednesday, July 20, 2016 4:01:00 PM CDT Gabriel Gellner wrote:
> Is there a standard (Base or common package) that implements something akin
> to matlab's `findpeaks`. It is easy enough to make something like this, but
> I imagined that this would be something that already exists but that I
They can mean "real" boxing and consequent performance problems, but sometimes
these get auto-removed during compilation. I see this all the time when
writing array code, for example this function which takes an input tuple and
adds 1 to each element:
julia> @inline inc1(a) = _inc1(a...)
inc1
Perhaps just
f = (U,V) -> flipdim(conv(flipdim(U, 1), V), 1)
if su < sv
return f([u; zeros(eltype(u), sv - su)], v)
elseif sv < su
return f(u, [v; zeros(eltype(v), su - sv)])
else
return f(u, v)
end
?
Worth pointing out that this is not specific to SubArrays; any AbstractArray
I never could tell my shapes apart. Nearly failed kindergarten because all my
lines had 3 corners.
--Tim
On Saturday, July 16, 2016 4:20:04 AM CDT Patrick Kofod Mogensen wrote:
> On Thursday, July 14, 2016 at 10:14:38 PM UTC+2, Tim Holy wrote:
> > *Until we get "diagonal dispa
Hi Madeleine,
Is there any chance you have two versions of julia installed, and are running
the wrong one? (Confession: I make that mistake surprisingly frequently!) The
reason I ask is that on current master:
julia> methodswith(Array, kron)
0-element Array{Method,1}
In master (which is what
Until we get "diagonal dispatch," I think the only way to do this is to expand
the number of arguments in the function:
myfunc(vec::Vector) = _myfunc(eltype(vec), vec) # internal function
_myfunc{D<:Dict{ASCIIString}}(::Type{D}, vec) = 1
_myfunc{D}(::Type{D}, vec) = 2
julia>
What might be missing is to note the important difference between abstract and
concrete types. Otherwise it might indeed be interpreted as advice to use
curly-brackets in your type definitions.
Best,
--Tim
On Tuesday, July 12, 2016 12:36:28 PM CDT Yichao Yu wrote:
> On Tue, Jul 12, 2016 at
The documentation page seems to be down at the moment, but when it's back up
see the "performance tips" page.
In particular,
- never benchmark anything in global scope, always wrap it in a function
- always run once before you benchmark it (to force JIT compilation)
--Tim
On Sunday, July 10,
much of a coding noob to figure out what to do
> (even with your precious hint).
>
> On Saturday, 9 July 2016 14:15:29 UTC+1, Tim Holy wrote:
> > Looks like xcorr has a type-instability. You can see this from
> >
> > x = rand(10 * 192000); u = rand(10 * 192000, 3);
>
Looks like xcorr has a type-instability. You can see this from
x = rand(10 * 192000); u = rand(10 * 192000, 3);
su = sub(u, :, 1);
@code_warntype xcorr(x, su)
Would you mind filing an issue? https://github.com/JuliaLang/julia/issues/new
Alternatively, if you feel up to fixing it, the hint is
Ambiguities are often a bit tricky. Two tips I've adopted:
- Have as few methods as possible, and declare types on their arguments only
when absolutely necessary. These measures greatly reduce your exposure to the
risk of ambiguities. To achieve this, it sometimes takes a fair bit of thought
On Friday, July 8, 2016 6:43:36 PM CDT Daniel Carrera wrote:
> For that matter, will there be upper case functions for every
> concrete type? ... I'm just curious. I wouldn't actually use that feature.
Yes, it's just the constructor. In most cases you don't have to define them
manually, they are
`has` or `exists`
--Tim
On Friday, July 8, 2016 8:01:40 AM CDT Davide Lasagna wrote:
> I have read the available documentation but I cannot seem to get it.
>
> How do I test whether an existing .jld file, opened as
>
> jldopen(filename, "r") do handle
> # test whether handle contains the
The string unification is already in julia-0.5.
There are functions called String(), Int(), and Float64(). In some cases there
are lowercase variants, and these often "do more" (e.g., `float` will parse a
string and return an AbstractFloat). The uppercase versions are the minimalist
On Thursday, July 7, 2016 4:41:10 PM CDT Islam Badreldin wrote:
> Maybe
> this means PyPlot.jl needs to add better support for ColorTypes?
That sounds like a very reasonable solution. I don't really know PyPlot at
all, so I don't have any advice to offer, but given how well you seem to
Well, you've just written that convenience function yourself :-). It's not
more complicated than that.
The reason there isn't an exported function is because the general case is not
quite as simple as it may seem. For example, BGR and RGB both mean "RGB
colors," but with different internal
both. This way I can easily the profiling via
>
> julia> Profile.print(format=:flat, sortedby=:count)
>
> Sorry for my dumb question.
>
> On Tuesday, July 5, 2016 at 12:49:37 PM UTC+1, Tim Holy wrote:
> > I don't know what "profiling a module" means. You profil
I don't know what "profiling a module" means. You profile running code,
wherever
that code happens to live---and that's all there is to say. To profile the code
in a module, you just have to write code that exercises the functions in the
module.
The meaning of the numbers is described here:
Duplicated in
http://stackoverflow.com/questions/38147946/julia-how-to-store-tuples-in-different-files
--Tim
On Friday, July 1, 2016 7:48:34 AM CDT Ahmed Mazari wrote:
> Hello,
>
> l want to store each tuple X[p] in a different file
> # X[1] in mini_batch1.jld X[2] in mini_batch2.
> # but
ay, July 1, 2016 5:27:03 PM CDT Rainer J. Engelken wrote:
> 2016-07-01 11:41 GMT+02:00 Tim Holy <tim.h...@gmail.com>:
> > My guess is that you could do better by doing a "batch update" of the
> > queue,
> > so that you don't rebalance the heap each time.
> >
Read about cache here:
http://julialang.org/blog/2013/09/fast-numeric
and add @inbounds @simd in front of the inner loop.
--Tim
On Friday, July 1, 2016 8:10:33 AM CDT baillot maxime wrote:
> Thank you but don't worry in general I do it like 3 or 4 times before
> looking at the time :)
>
> I
To translate:
[a] concatenation is deprecated; use collect(a) instead
in depwarn at ./deprecated.jl:73
while loading In[2], in expression starting on line 9
The first line implies you have some object `a` and you're putting brackets
around it, `[a]`. Julia is telling you to switch
Hi everyone,
I'm writing to call your attention to a new page on the julialang website,
explaining the mission of a newly-created group called "Julia Stewards":
http://julialang.org/community/stewards/
This is a group that we hope will have very little to do, but which seems
important
In Base there's mapslices, which may already do what you want.
Something like what you're proposing (but not yet with a nice "high level" API
wrapper, since the low-level stuff is still in flux) is already here:
https://github.com/timholy/ArrayIteration.jl
That's aiming at a very general
I'd try TensorOperations.jl or AxisAlgorithms.jl
Best,
--Tim
On Sunday, June 26, 2016 4:24:17 AM CDT Alex Williams wrote:
> I'm trying to code an efficient implementation of the n-mode tensor
> product. Basically, this amounts to taking a dot product between a vector
> (b) and the mode-n fibers
On Sunday, June 26, 2016 3:27:57 AM CDT Nils Gudat wrote:
> Oh well that would explain why it had no effect then - is there some
> version that contains this fix then? Would building the current 0.4.7+
> master from source do? (Sorry, I've stuck strictly to stable versions so
> far so am not well
On Saturday, June 25, 2016 6:45:05 AM CDT jw3126 wrote:
> @Tim: Thanks for the insights! On my machine slowness starts to kick in at
> size 9 already.
Depends on which version of julia you're running (I'm running a recent
julia-0.5-dev).
> I tried to read the llvm code, but did not really
>
It might scale just fine, because LLVM might discover the pattern of your
"giant switch statement" and essentially compute a jump location. Even if not,
if you're calling this repeatedly with a tuple of consistent length, the
branches become predictable and then they kind of don't count. So in
On Sunday, June 19, 2016 at 10:20:11 PM UTC-5, Tim Holy wrote:
>
> Presumably if someone sat down and implemented similar operations in
> pure-julia using, say,
> tuples, it would blow away what we have now.
Proof-of-principle in this
gist: https://gist.github
On Sunday, June 19, 2016 5:51:35 PM CDT Chris Rackauckas wrote:
> But as a user, I do find it troubling that the only reliable high-precision
> number type is clearly not aiming for performance.
More like "it's hard to make it high performance" because the interaction with
the C library means
Hmm, that needs fixing. Thanks for mentioning it.
--Tim
On Fri, Jun 17, 2016 at 4:04 PM, Kaela Martin wrote:
> JLD worked! Thanks!
>
> As a side note, I typed in the wrong command and got lots of errors.
>
> save("temp.jld","t",t)
> load("temp.jld","t",t) # t is
>
>
Can you use JLD.jl? It also writes hdf5 files, but adds annotation to allow
julia to know what it's dealing with.
Best,
--Tim
On Friday, June 17, 2016 10:51:07 AM CDT Kaela Martin wrote:
> Let's say I have a composite type of:
>
> type TEST
> numbers
> times
> measurement
> end
>
1 - 100 of 1531 matches
Mail list logo