[julia-users] Re: Transpose view of a matrix

The plan is that the transpose function will return this eventually and within the 1.0 time frame but it's not done yet. It will probably not be a PermutedDimsArray though because it wouldn't do the right thing for the conjugate transpose of a complex matrix. On Wednesday, October 12, 2016 at

[julia-users] Re: Hermitian or Symmetric + UniformScaling

It's not on purpose. It is just that it hasn't been implemented yet. It would be great if you could open a pull request with such a method. You might also want to define a special type for C+λI such that you can avoid creating a new matrix but it is probably better to experiment with such a

[julia-users] Re: linear algebra question

The documentation should be expanded with more examples. Many of the linear algebra functions work for arbitrary input types so if you construct a matrix with rational or integer inputs then many of the functions will still work. We don't have much support in base for fancy math on such

Re: [julia-users] Re: Why is BLAS.dot slower than BLAS.axpy!?

Maybe I’ll try 0.5 and OpenBLAS for comparison. > > On 10 Sep 2016, at 2:34 AM, Andreas Noack <andreasnoackjen...@gmail.com> > wrote: > > Try to time it again with threading disabled. Sometimes the > threading heuristics can cause unintuitive performance. > > On Friday, Sept

[julia-users] Re: Error in call to eigs--0 index in arpack wrapper

It would be helpful if you could provide a self-contained example. Also, would it be possible to try out the release candidate for 0.5. We have made a few changes to the ARPACK wrappers so it would be useful to know if is only happening on 0.4. Thanks. On Saturday, September 10, 2016 at

[julia-users] Re: Why is BLAS.dot slower than BLAS.axpy!?

Try to time it again with threading disabled. Sometimes the threading heuristics can cause unintuitive performance. On Friday, September 9, 2016 at 6:39:13 AM UTC-4, Sheehan Olver wrote: > > > I have the following code that is part of a Householder routine, where > j::Int64, > N::Int64,

Re: [julia-users] Re: Julia equivalent to Matlab's [Q, R, P] = qr(A)?

My memory is too short. I just realized that I implemented a generic pivoted QR last fall so if you try out the prerelease of Julia 0.5 then you'll be able to compute the pivoted QR of a BigFloat matrix. On Wednesday, September 7, 2016 at 9:20:12 AM UTC-4, Andreas Noack wrote: > >

Re: [julia-users] Re: Julia equivalent to Matlab's [Q, R, P] = qr(A)?

We use LAPACK's QR with column pivoting. See http://www.netlib.org/lapack/lug/node42.html. LAPACK uses blocking for BLAS3 but that is not necessary for a generic implementation. So it's the task is just to sort the columns by norm at each step. If you want to try an implementation you can look

Re: [julia-users] Re: Return type of eye()

Evan, this is exactly where you should use I, i.e. m = m + λ*I The reason is that eye(m) will first allocate a dense matrix of size(m,1)^2 elements. Then * will do size(m,1)^2 multiplications of lambda and allocate a new size(m,1)^2 matrix for the result. Finally, size(m,1)^2 additions will be

Re: [julia-users] Re: Return type of eye()

Julia is developing over time. Originally, eye was probably implemented to mimic Matlab. Later we realized that the type system allowed us to define the much nicer UniformScaling which has the special case const I = UniformScaling(1) which is almost alway better to use unless you plan to modify

Re: [julia-users] Re: Return type of eye()

We could deprecate eye. Then the users would get a warning directing them to use `I` instead. On Mon, Aug 29, 2016 at 6:29 AM, Júlio Hoffimann wrote: > I still think that having a "global variable" named "I" is not robust. > I've read so many scripts in matlab that do

Re: [julia-users] Re: Return type of eye()

If we could somehow make `I` more visible, wouldn't you think that B = I*A is better than B = eye(1)*A ? Small side note: the best we can hope for is probably performance similarly to B = copy(A) because it wouldn't be okay to alias A and B when B has been constructed from *. On Mon, Aug

[julia-users] Re: is there an rcond (reciprocal condition number of a matrix) function to complement cond

No. We are only exposing `cond` but as you can see in https://github.com/JuliaLang/julia/blob/master/base/linalg/lu.jl#L235 we are actually getting `rcond` from LAPACK and then calling `inv`. I can see the usefulness of working with a number in [0,1] instead of [1,inf) but it seems superfluous

[julia-users] Re: Return type of eye()

You can also overwrite eye Could you elaborate on the "90% of the users won't be aware of these internal details in their day-to-day coding" part? If we ignore the name for a while, why is `I` not what you want here? It is as efficient as it can possibly be. On Sunday, August 28, 2016 at

Re: [julia-users] Re: chol() more strict in v0.5?

ntry like > this in Compat.jl. > > On Sunday, August 7, 2016 at 10:02:44 PM UTC+2, Andreas Noack wrote: >> >> It would be great with an entry for this in Compat.jl, e.g. something like >> >> cholfact(A::HermOrSym, args...) = cholfact(A.data, A.uplo, args...) >>

Re: [julia-users] Re: chol() more strict in v0.5?

It would be great with an entry for this in Compat.jl, e.g. something like cholfact(A::HermOrSym, args...) = cholfact(A.data, A.uplo, args...) On Sun, Aug 7, 2016 at 2:44 PM, Chris <7hunderstr...@gmail.com> wrote: > mmh, could you explain your comment a little more? > > David, thanks for the

Re: [julia-users] Re: eigs fails silently

mail.com > > wrote: > >> Andreas, thanks for the investigation. I'll use 0.5 for now, and hope the >> real problems I encounter are within the capabilities of ARPACK. >> >> It's embarrassing to be bested by Matlab... >> >> On Fri, Aug 5, 2016 at 9:23 P

[julia-users] Re: eigs fails silently

I've looked a bit into this. I believe there is a bug in the Julia wrappers on 0.4. The good news is that the bug appears to be fixed on 0.5. The bad news is the ARPACK seems to have a hard time with the problem. I get julia> eigs(A,C,nev = 1, which = :LR)[1] ERROR:

[julia-users] Re: chol() more strict in v0.5?

Yes. We are stricter now. LAPACK doesn't check for symmetry at all which we used to mimic. It might seem pedantic but it will capture programming errors and it is more in line with how check things elsewhere and it's in line with the behavior for the sparse Cholesky. For now, you'd need to use

[julia-users] Re: Linear-Equation-System Solver("\") does not work reliably for the "Rational" Type

A couple of things go wrong here. Right now, Julia tries to use the QR factorization to solve underdetermined systems. If things had worked the way I'd planned, your matrix would have been promoted to floating point elements (I know that it's not what you want so keep reading). The promotion

[julia-users] Re: caching the pivot ordering for a sparse cholesky factorization?

The for Cholesky and sparse LDLt there are methods for reusing an existing symbolic factorization. They are cholfact!/ldltfact!(Factorization,Matrix) so you can e.g. do ``` julia> A = sprandn(100,100,0.1) + 10I; julia> F = cholfact(Symmetric(A)); julia> cholfact!(F, Symmetric(A - I)); ```

Re: [julia-users] Re: testing positive definiteness via Cholesky?

We could consider not to throw in cholfact but only when the factorization is applied. This is what we do for LU and BunchKaufman. On Saturday, June 11, 2016 at 5:02:55 PM UTC-4, Tony Kelman wrote: > > For now, you can just manually make the same cholmod calls that cholfact > does, then use the

[julia-users] Re: Pivoting when inverting a sparse matrix

To start with the conclusion, the easiest solution here is to use the lufact, i.e. lufact(sparse(A))\ones(3). The explanation is that when a sparse matrix is symmetric, we first try to use a sparse Cholesky factorization and when that fails we try a sparse LDLt factorization. Both use Cholmod,

[julia-users] Re: is the L in (L,U,p) = lu(A) in guaranteed to have 1's on the diagonal when A is not full rank?

Yes. The way out Ls are constructed will ensure that it has 1s in the diagonal. On Tuesday, June 7, 2016 at 6:20:26 PM UTC-4, Gabriel Goh wrote: > > A simple question as posed in the title, this is guaranteed by LAPACK > > The factorization has the form > A = P * L * U > where

Re: [julia-users] Re: linreg strangely picky about inputs

tot_xy += x[i]*y[i] > end > b1 = (tot_xy - mx*tot_y)/tot_dx # a 1\n cancels in the top and bottom > b0 = tot_y/n - b1*mx > return [b0, b1] > end > > If anyone has any comments on what could be made better. > > Thanks, > Gabriel > > On Mon

[julia-users] Re: linreg strangely picky about inputs

return [b0, b1] > end > > Which I find speeds up around 3x, or do you mean writing a custom cov > function that is smarter about memory? (I am returning an array as I like > to be able to do vector math on the coefficients ... but if I return a > tuple it isn't much faster for me)

[julia-users] Re: linreg strangely picky about inputs

I don't think that linreg has received much attention over the years. Most often it is almost as simple to use \. It you take a look at linreg then I'd suggest that you try to write in terms of cov and var and return a tuple instead of an Array. That will speed up the computation already now

[julia-users] Re: JuliaCon birds of a feather

The Hackathon on Saturday is quite flexible so if somebody volunteers to organize such session then I'm sure we can make it happen. There should be people around that are knowledgeable about the topics. On Tuesday, May 17, 2016 at 10:26:00 AM UTC-4, Josef Sachs wrote: > > Will there be any

[julia-users] Re: Matrix multiplication performance with BitMatrix, Matrix{Bool}, Matrix{Int8}

I think there'll necessarily be some overhead in the decompression of the packed Booleans in a BitArray. The difference between Bool and Int8 is that the Int8 is promoted to a Float64 whereas a Bool is not. It appears that the Bool multiplication in

[julia-users] JuliaCon 2016: Keynote speakers

has co-designed the programming language Scheme, which has greatly influenced the design of Julia, as well as languages such as Fortress and Java. Andreas Noack JuliaCon 2016 Local Chair Postdoctoral Associate Computer Science and Artificial Intelligence Laboratory Massachusetts Institute

[julia-users] Re: hypot question

It's to ensure that the return type doesn't depend on the value of x. If x and y are integers then the return type of hypot1 will be Int if x==0 and Float64 otherwise.

Re: [julia-users] Re: git help needed

> CONFLICT (content): Merge conflict in src/MixedModels.jl > Auto-merging src/GLMM/PIRLS.jl > CONFLICT (content): Merge conflict in src/GLMM/PIRLS.jl > Automatic merge failed; fix conflicts and then commit the result. > > > > > On Tuesday, March 8, 2016 at 2:29:08

[julia-users] Re: git help needed

Provided that you have pushed all your changes, I think the easiest solution is to "remove" the commit in which you add the chksqr fix. You can do that with git reset --hard c0b5c41d136013a8e2cd57f5bedd8c96f5d2e3c6 # the commit right before the chksqr changes git cherry-pick

Re: [julia-users] Re: ANN: CppWrapper C++ wrapping package

the .so) > > We currently have a C++ project with Python bindings done using > Boost.Python, which also uses the latter approach, so translating this is > more natural using CppWrapper. > > Maybe I should clarify this in the docs? > > Cheers, > > Bart > > On

[julia-users] Re: ANN: CppWrapper C++ wrapping package

Hi Bart, Are you aware of https://github.com/Keno/Cxx.jl? What are the reasons for a separate package? Best Andreas On Sunday, February 14, 2016 at 6:11:07 PM UTC-5, Bart Janssens wrote: > > Hi all, > > The CppWrapper package is meant to facilitate exposing C++ libraries to > Julia, allowing

[julia-users] Re: Factors in Sparse QR factorization

Eventually, we should support multiplying with Q in the same way it possible with dense QR. It will require that we introduce a Q type for sparse QR for which we can overload the usual matrix multiplication functions by appropriate calls to qmult. Notice however that qmult is only defined for

[julia-users] Re: Cholmod Factor re-use

Sorry for the confusion here. We have moved the CHOLMOD module to SparseMatrix.CHOLMOD in 0.4 and SparseArrays.CHOLMOD in 0.5. However, update! will probably be an exported function in 0.5 so this should become much easier. On Tuesday, December 1, 2015 at 7:08:12 AM UTC-5, Matthew Pearce

[julia-users] Re: drandn sometimes crashes

Please file this issue at DistributedArrays.jl On Monday, November 23, 2015 at 8:13:06 AM UTC-5, Antonio Suriano wrote: > > addprocs(3) > > @everywhere using DistributedArrays > > function tony(N) > return sum(drandn(N,N)) > end > > > function pesante(N) > a=zeros(N,N) > for i = 1:N > for j=1:N >

[julia-users] Re: low rank matrix-vector multiplication and order of operations

The order of operations is from left to right so the parentheses can be important here. We have discussed ways of executing more efficiently for matrix products, see https://github.com/JuliaLang/julia/issues/12065, but so far nothing has been implemented. In that issue, you can also see the

[julia-users] Re: VAR estimation in Julia

Hi Michael Author of Civecm.jl here. The repo has code that I developed during my PhD and I haven't spend much time on it lately, but please ask questions if you'd like to use it. Best Andreas On Friday, October 16, 2015 at 5:59:44 PM UTC-4, Michael Wang wrote: > > Cool, thank you! I will

[julia-users] Re: Type stability (or not) in core stats functions

I think you are right that we should simply remove the mean keyword argument from cov and cor. If users want the efficient versions with user provided means then they can use corm and covm. Right now they are not exported, but we could consider doing it, although I'm in doubt if it is really

[julia-users] Re: Is some Autocorrelation function in Julia?

Take a look in StatsBase.jl On Sunday, August 9, 2015 at 8:17:02 AM UTC-4, paul analyst wrote: *Is some Autocorrelation function in Julia?Paul*

[julia-users] Re: Time series current state of the art

For the example you describe, you can simply use the tools in base, but unfortunately I don't think our reader can handle continental style decimal comma yet. However, that is easy to search/replace with a dot. Something like cov(diff(log(readdlm(prices.csv, ';' should then do the job.

[julia-users] Re: How to make (sort of) pointers to an array subsection ?

You can use sub or slice for this, e.g. W = slice(Wall, 2:size(Wall, 1), 2:size(Wall, 2)) Den onsdag den 15. juli 2015 kl. 06.14.14 UTC-4 skrev Ferran Mazzanti: Hi folks, I have a little mess with the way arrays are being handled in Julia. I come from C and fortran95 and I know I can do

[julia-users] Re: Efficient way to compute X' diag(w) Y

No such BLAS routine exists, but for larger matrices the calculation will be dominated by the final matrix-matrix product anyway. Den tirsdag den 7. juli 2015 kl. 18.24.34 UTC-4 skrev Matthieu: Thanks, this is what I currently do :) However, I'd like to find a solution that is both memory

[julia-users] Re: Condition Number of a Matrix

Hi Jared The short answer is yes. Different algorithms are used in `svdvals` and `svd`/`svdfact`. In both cases, we are using the divide and conquer routine xGESDD from LAPACK, but internally in the routine uses two different algorithms depending on the choice for the vectors. Are requested or

Re: [julia-users] Re: Efficient way to compute X' diag(w) Y

You could, but unless the matrices are small, it would be slower because it wouldn't use optimized matrix multiplication. 2015-07-08 10:36 GMT-04:00 Josh Langsfeld jdla...@gmail.com: Maybe I'm missing something obvious, but couldn't you easily write your own 'cross' function that uses a couple

Re: [julia-users] Re: Efficient way to compute X' diag(w) Y

8, 2015 at 10:39 AM, Andreas Noack andreasnoackjen...@gmail.com wrote: You could, but unless the matrices are small, it would be slower because it wouldn't use optimized matrix multiplication. 2015-07-08 10:36 GMT-04:00 Josh Langsfeld jdla...@gmail.com: Maybe I'm missing something obvious

[julia-users] Re: caveat: Givens does not transpose

Hi Ivan This is fixed on 0.4 but needs a backport to 0.3. I'll take a look. Den tirsdag den 7. juli 2015 kl. 08.41.37 UTC-4 skrev Ivan Slapnicar: In [1]: Z=givens(1.0,2.0,1,3,3) Z, Z', transpose(Z) Out[1]: ( 3x3 Givens{Float64}: 0.447214 0.0 0.894427 0.0 1.0 0.0

Re: [julia-users] Re: ldltfactor for sparse matrices

: This is very interesting ! So UMFPACK is more robust and this is why I am not having any issues with the same matrix. Thanks. On Wednesday, May 27, 2015 at 6:15:57 PM UTC-3, Andreas Noack wrote: It could happen if a pivot is zero. CHOLMOD's ldlt is only making permutations in the symbolic

Re: [julia-users] Type inference cannot deduce that convert(Float32,x) returns a Float32?

The convert methods for Date.Period, Complex and Rational are inferred to give Any. The problem in Period is because of the use of the value method in line 4 of periods.jl. It extracts a field from an abstract type so even though all subtypes in base have the specified field and have it defined

[julia-users] Re: Simple Conjugate Gradients coded in Julia is much faster than cholfact

I think the chosen matrix has very good convergence properties for iterative methods, but I agree that iterative methods are very useful to have in Julia. There is already quite a few implementations in https://github.com/JuliaLang/IterativeSolvers.jl I'm not sure if these methods cover the

[julia-users] Re: option to return success flag in chol() and constructor for Triangular

You can use try/catch, e.g. julia try cholfact(A) catch e e.info end 1 In 0.3, you can construct a Triangular matrix with Triangular(B, :L) and in 0.4 with LowerTriangular(B). Den onsdag den 27. maj 2015 kl. 18.12.51 UTC-4 skrev Roy Wang: Is there an easy way to

Re: [julia-users] Re: ldltfactor for sparse matrices

confused with this error. On Wednesday, May 27, 2015 at 2:22:30 PM UTC-3, Andreas Noack wrote: In 0.3 the sparse LDLt and Cholesky factorizations are both in the cholfact function. If the matrix is symmetric, but not positive definite the result of cholfact will be an LDLt factorization. In 0.4

[julia-users] Re: ldltfactor for sparse matrices

In 0.3 the sparse LDLt and Cholesky factorizations are both in the cholfact function. If the matrix is symmetric, but not positive definite the result of cholfact will be an LDLt factorization. In 0.4 the factorizations have been split into cholfact and ldltfact. Den onsdag den 27. maj 2015

Re: [julia-users] Re: ldltfactor for sparse matrices

27, 2015 at 3:25:46 PM UTC-3, Eduardo Lenz wrote: Funny... I dont have CHOLMOD installed...but I am using the official windows installer.. I will try to make a fresh install. Thanks Andreas ! On Wednesday, May 27, 2015 at 2:59:30 PM UTC-3, Andreas Noack wrote: What do you get when you type

Re: [julia-users] Re: ldltfactor for sparse matrices

On Wednesday, May 27, 2015 at 5:54:20 PM UTC-3, Andreas Noack wrote: As I wrote in the first reply: in 0.3 the cholfact function returns the LDLt when the matrix is symmetric but not positive definite, e.g. julia A = sprandn(5,5, 0.5); julia A = A + A'; julia b = A*ones(5); julia

Re: [julia-users] Re: ldltfactor for sparse matrices

avaliable in a regular windows install. Thanks for your help Andreas ! On Wednesday, May 27, 2015 at 5:37:53 PM UTC-3, Andreas Noack wrote: You are using 0.3.8 and not 0.4. Have you tried cholfact(A)? 2015-05-27 16:33 GMT-04:00 Eduardo Lenz eduardo...@gmail.com: Just to make it clear

Re: [julia-users] forming an FFT matrix in julia

1. In Julia fft(A) is the 2d DFT of A. You can get MATLAB's behavior with fft(A, 1) 2. I might not understand what you are trying to do, but it appears to me that you can just apply the DFT to the full vector and then sample the elements of the vector. 2015-05-10 22:32 GMT-04:00 Edward Chen

Re: [julia-users] asymmetry between treatment of mean and cov

and not the covariance function. 2015-05-08 17:05 GMT-04:00 JPi pin...@gmail.com: Yes, the variance. But that doesn't explain why you can't get the covariance matrix of an array of vectors. On Friday, May 8, 2015 at 4:51:13 PM UTC-4, Andreas Noack wrote: Calculating the covariance requires two

Re: [julia-users] asymmetry between treatment of mean and cov

Calculating the covariance requires two sequences of data points. Either from two vectors or between the columns of a matrix. The mean is different as it requires one sequence. What did you expect to get from the covariance function of a vector? The variance? 2015-05-08 16:01 GMT-04:00 JPi

Re: [julia-users] Is anyone, anywhere using Julia on a supercomputer/cluster

As I'm writing this, I'm running Julia on a pretty new 90 node cluster. I don't know if that counts as medium size cluster, but recently it was reported on the mailing list that Julia was running on http://www.top500.org/system/178451 which I think counts as a supercomputer. 2015-04-28 19:58

Re: [julia-users] Yet Another String Concatenation Thread (was: Re: Naming convention)

I like the idea of something like factorize(MyType,...), but it is not without problems for generic programming. Right now cholfact(Matrix) and cholfact(SparseMatrixCSC) return different types, i.e. LinAlg.Cholesky and SparseMatrix.CHOLMOD.Factor. The reason is that internally, they are very

Re: [julia-users] Sparse cholesky and triangular solve

, Andreas Noack wrote: If B and D are different then why is it not okay to calculate x = C\D and then B'x afterwards? 2015-04-27 15:24 GMT-04:00 matt katz...@gmail.com: I would like to compute multiple quadratic forms B'*C^(-1)*D, where B, C, and D are sparse, and C is always the same symmetric

Re: [julia-users] Solving a linear system with sparse, approximately low rank, and symmetric A

at a time and deflate A by removing the corresponding eigenvector? Thanks, Mladen On Tuesday, April 21, 2015 at 9:09:29 AM UTC-5, Andreas Noack wrote: I'm not sure what the best solution is here because I don't fully understand your objective. If A has low rank then the solution

Re: [julia-users] Sparse cholesky and triangular solve

If B and D are different then why is it not okay to calculate x = C\D and then B'x afterwards? 2015-04-27 15:24 GMT-04:00 matt katzf...@gmail.com: I would like to compute multiple quadratic forms B'*C^(-1)*D, where B, C, and D are sparse, and C is always the same symmetric positive matrix. For

Re: [julia-users] Finite field / Gaulois field implementation

Hej Valentin There is a couple of simple examples. At least this http://acooke.org/cute/FiniteFiel1.html and I did one in this http://andreasnoack.github.io/talks/2015AprilStanford_AndreasNoack.ipynb notebook. The arithmetic definitions are simpler for GF(2), but should be simple

Re: [julia-users] type-stable way to get primary type

This problem is quite common in the LinAlg code. We have two type of definitions to handle the conversion of the element types. First an idea due to Jeff as implemented in https://github.com/JuliaLang/julia/blob/237cdab7100b29a6769313391b1f8d2563ada06e/base/linalg/triangular.jl#L20 and

Re: [julia-users] CholmodFactor not defined?

The reason is that it is not exported so you can either use the full path bar(Base.LinAlg.CHOLMOD.CholmodFactor{Float64,Int64}) = 1 or use the type first using Base.LinAlg.CHOLMOD.CholmodFactor bar(CholmodFactor{Float64,Int64}) = 1 2015-04-21 10:27 GMT-04:00 andreas scheidegge...@gmail.com:

Re: [julia-users] Solving a linear system with sparse, approximately low rank, and symmetric A

I'm not sure what the best solution is here because I don't fully understand your objective. If A has low rank then the solution is not unique and if it has almost low rank, the solution is very ill conditioned. A solution could our new shift argument to our complex Cholesky factorization. This

Re: [julia-users] How to declare correctly this matrix?

Hi Michela It is easier to help if your example is complete such that it can just be pasted into the terminal. The variable gmax is not defined in your example, but I guess it is equal to length(SUC_C). It is also useful to provide the exact error message. That said, I think the root of the

Re: [julia-users] Re: Fastest way for constructing a hermitian matrix

We mainly use SymTridiagonal for eigenvalue problem and therefore it is not necessary to allow for complex matrices because the Hermitian eigenvalue problem can be reduced to a real symmetric problem. It might be easier to specify the problem in Hermitian form, so we might change this. What is

Re: [julia-users] pinv(0) != pinv([0])

This has been fixed on master. I've just backported the fix to the release branch so it should be okay in 0.3.8. 2015-04-16 4:05 GMT-04:00 Rasmus Brandt rasmus...@gmail.com: Hey everyone, I just stumbled over this behaviour in Julia-0.3.7, which seems a bit unintuitive to me: julia pinv(0)

Re: [julia-users] Re: Livestream of ICME LA/Opt Seminar today 4:15pm PDT (Andreas Noack, MIT)

The notebook is now available from http://andreasnoack.github.io/talks/2015AprilStanford_AndreasNoack.ipynb Note that it is based on master so some parts of the code might fail on Julia release. 2015-04-11 15:21 GMT-04:00 Andreas Noack andreasnoackjen...@gmail.com: I've been in transit back

Re: [julia-users] Re: Livestream of ICME LA/Opt Seminar today 4:15pm PDT (Andreas Noack, MIT)

at Stanford is pleased to have Andreas Noack and Jiahoa Chen speaking in our Linear Algebra and Optimization seminar this Thursday and next. Today's talk will be livestreamed via YouTube starting at 4:15pm PDT. Livestream link: https://www.youtube.com/watch?v=a_bFB1BZbvI (Videos will also

Re: [julia-users] does pca() center the data?

Which pca? 2015-04-06 6:53 GMT-07:00 Steven Sagaert steven.saga...@gmail.com: does pca() center the input output data or do you have to do that yourself?

Re: [julia-users] does pca() center the data?

There is no pca in Julia Base 2015-04-06 9:16 GMT-07:00 Steven Sagaert steven.saga...@gmail.com: the one from the standard lib On Monday, April 6, 2015 at 4:01:00 PM UTC+2, Andreas Noack wrote: Which pca? 2015-04-06 6:53 GMT-07:00 Steven Sagaert steven@gmail.com: does pca() center

Re: [julia-users] Calling gemm! on part of an array

I think that you could use sub(A,:,2:2:4) for BLAS, but not sub(A,:,[2,4]) because the indexing has to be with ranges for BLAS to be able to extract the right elements of the matrix. 2015-03-29 17:34 GMT-04:00 Dominique Orban dominique.or...@gmail.com: Sorry if this is another [:] kind of

Re: [julia-users] Calling gemm! on part of an array

supposed to do, it just doesn't update C (which is awkward for a ! function, though I realize there's something else going on here). On Sunday, March 29, 2015 at 5:56:27 PM UTC-4, Andreas Noack wrote: I think that you could use sub(A,:,2:2:4) for BLAS, but not sub(A,:,[2,4]) because the indexing

Re: [julia-users] Calling gemm! on part of an array

:00 Dominique Orban dominique.or...@gmail.com: Unfortunately, my idx will be computed on the fly and there's zero chance that it would be a range. Is there a plan to support more general indexing in subarrays and/or ArrayView? On Sunday, March 29, 2015 at 6:13:16 PM UTC-4, Andreas Noack wrote

Re: [julia-users] code generation with @eval

Distributed reduce is already implemented, so maybe these slightly simpler with e.g. sum(A::DArray) = reduce(Base.AddFun(), A) 2015-03-26 8:41 GMT-04:00 Jameson Nash vtjn...@gmail.com: `eval` (typically) isn't allowed to handle `import` and `export` statements. those must be written explicitly

Re: [julia-users] Is there a Julia equivalent to Mathematica's Tally?

I think that countmap in StatsBase does that Den torsdag den 26. marts 2015 skrev DumpsterDoofus peter.richter@gmail.com: In Mathematica, there is a function called Tally which takes a list and returns a list of the unique elements of the input list, along with their multiplicities. I.e,

Re: [julia-users] Parallel performance test question

It has caused a lot of frustration. See #9118. I think the easiest right now is for p in procs() @spawnat p blas_set_num_threads(k) end 2015-03-17 23:19 GMT-04:00 Sheehan Olver dlfivefi...@gmail.com: Hi, I've created the following to test the performance of parallel processing on our

Re: [julia-users] Change array of arrays to a one dimensional array

new_array = vcat(data...) 2015-03-17 20:59 GMT-04:00 Christopher Fisher fishe...@miamioh.edu: Hi all- pmap outputs the results as an array of arrays and I am trying to find a flexible way to change it into a one dimensional array. I can hardcode the results as new_array =

Re: [julia-users] Looking for incomplete SVD -- need only some of the singular values

I've tried to make the package that Jiahao mentioned usable. I think it works, but it probably still has some rough edges. You can find it here https://github.com/andreasnoack/TSVD.jl and there is a help entry for tsvd that explains the arguments. For a 2000x2000 dense non-symmetric complex

Re: [julia-users] Solving a lower triangular system of equations

On 0.3.x there is a very expensive error bounds calculation in the triangular solve which the reason for the surprisingly slow calculation. This is not acceptable and we have therefore removed the error bounds calculation in 0.4. On my machine I get julia @time L\B; elapsed time: 2.535437796

Re: [julia-users] Tolerance for the checks in cholfact too small?

Good to to hear that. I've filed an issue to figure out how to make this more consistent https://github.com/JuliaLang/julia/issues/10520 2015-03-14 21:06 GMT-04:00 Kristoffer Carlsson kcarlsso...@gmail.com: On Sunday, March 15, 2015 at 1:02:11 AM UTC+1, Andreas Noack wrote: You can get

Re: [julia-users] Tolerance for the checks in cholfact too small?

You can get around this by specifying that the matrix is symmetric. This can be done with cholfact(Symmetric(A, :L)) which then bypasses the test for symmetry and cholfact only looks at the lower triangle. However, the error you got is not consistent with the way cholfact works for dense

Re: [julia-users] Saving timing results from @time

@elapsed is what you are looking for 2015-03-11 7:43 GMT-04:00 Patrick Kofod Mogensen patrick.mogen...@gmail.com : I am testing the run times of two different algorithms, solving the same problem. I know there is the @time macro, but I cannot seem to wrap my head around how I should save the

Re: [julia-users] Dimension mismatch error in Julia!

It is more helpful if you can provide a self contained example that we can run. However, I think you've been bitten by our white space concatenation. When you define f, the second element is written *-w^2*sin(x)-u*w^2*cos(x) -2γ*y* *but I think that is getting parsed as*

Re: [julia-users] [ANN] MatrixDepot 0.2.0: include a Julia interface to the UF Sparse Matrix collection

Hi Weijian This is a great functionality. It seems that you are using MAT.jl to read in the sparse matrices. You could consider using the the MatrixMarket reader in Base e.g. A = sparse(Base.SparseMatrix.CHOLMOD.Sparse(matrix.mtx)) It will also have the benefit of using the Symmetric matrix

Re: [julia-users] [ANN] MatrixDepot 0.2.0: include a Julia interface to the UF Sparse Matrix collection

to this reader when Julia v0.4 is released. Best, Weijian On Friday, 6 March 2015 20:35:43 UTC, Andreas Noack wrote: Hi Weijian This is a great functionality. It seems that you are using MAT.jl to read in the sparse matrices. You could consider using the the MatrixMarket reader in Base

Re: [julia-users] Unevenly distributed arrays

UTC+1, Andreas Noack wrote: I don't think it is possible right now. We have been discussing more flexible solutions, but so far nothing has been done. 2015-03-04 9:22 GMT-05:00 Simone Ulzega simone...@gmail.com: Is it possible to construct a DArray with unevenly distributed chunks? For example

Re: [julia-users] Unevenly distributed arrays

I don't think it is possible right now. We have been discussing more flexible solutions, but so far nothing has been done. 2015-03-04 9:22 GMT-05:00 Simone Ulzega simoneulz...@gmail.com: Is it possible to construct a DArray with unevenly distributed chunks? For example, I want to create a

Re: [julia-users] Trouble with simple probit regression not converging with glm

I don't see an obvious reason for this so please try to post this as an issue on GLM.jl. 2015-02-27 10:47 GMT-05:00 Andrew Newman andrew.brudev...@gmail.com: Hi Julia-users, I am trying to run a few simple regressions on simulated data. I had no problem with a logit and was able to run it

Re: [julia-users] Re: pmap not respecting srand set seed settings

Steve, I don't think that method works. The mapping between the argument to srand and the internal state of the MT is quite complicated. We are calling a seed function in the library we are using that maps an integer to a state vector so srand(1) and srand(2) end up as two quite different streams.

Re: [julia-users] Is it OK (stylistically) to override missing Base functionality in a package?

vanaf mijn iPhone Op 27-feb.-2015 om 21:27 heeft Andreas Noack andreasnoackjen...@gmail.com het volgende geschreven: I think it is fine that the type of the argument determines the behavior here. Having type in the name would be a bit like having `fabs(x::Float64)`. 2015-02-27 15:21 GMT-05:00

Re: [julia-users] Is it OK (stylistically) to override missing Base functionality in a package?

I'd like to have something like this. 2015-02-27 15:02 GMT-05:00 Jutho juthohaege...@gmail.com: Or in this particular case, maybe their should be some functionality like that in Base, or at least in Base.LinAlg, where is often necessary to mix complex variables and real variables of the same

Re: [julia-users] Is it OK (stylistically) to override missing Base functionality in a package?

type. Maybe something like realtype , or typereal if we want to go with the other type... functions. Op vrijdag 27 februari 2015 21:18:34 UTC+1 schreef Andreas Noack: I'd like to have something like this. 2015-02-27 15:02 GMT-05:00 Jutho juthoh...@gmail.com: Or in this particular case

Re: [julia-users] pmap not respecting srand set seed settings

@everywhere srand(seed) would give reproducibility, but it would probably not be a good idea since the exact same random variates will be generated on each process. Maybe something like for p in workers() @spawnat p srand(seed + p) end However, out RNG gives no guarantees about independence of

  1   2   3   4   >