Re: [Jprogramming] in practice, for numerical computing, how often are > 5 dim tensors used?

ssays/FFT . For argument > vectors of length 2^n, the algorithm creates rank n binary hypercubes, > shape n$2. (Subfunction "cube".) > > > > On Fri, Dec 29, 2017 at 8:10 PM, TongKe Xue wrote: > > > Hi, > > > > > > 1. I agree that the concep

[Jprogramming] in practice, for numerical computing, how often are > 5 dim tensors used?

Hi, 1. I agree that the concept of rank + cell + frames + implicit parallel8ism is very cool. 2. 5 dimension is commonly defined as N, C, D, H, W: N = number of images C = channels D = depth H = height W = width In practice, how often do we use algorithms with > 5 dim tensors?

Re: [Jprogramming] gpu-backed J vs Tensorflow

gt; overhead of conversion from/to double precision may or may not be > significant, it depends on applications, ymmv. > > On Dec 20, 2017 4:39 AM, "TongKe Xue" wrote: > > > Hi, > > > > In my experience, on the CPU, J beats Java. I suspect this is due to

[Jprogramming] gpu-backed J vs Tensorflow

Hi, In my experience, on the CPU, J beats Java. I suspect this is due to Java's GC and J's ability to via "higher representation of ranks/loops" to run highly optimized code. Is there any reason to believe that GPU-backed-J would beat Tensorflow on Tensor / Deep Learning work ? Given that

[Jprogramming] Learning J by Passing Unit Tests

Hi, I am writing a mini-J interpreter. Lexer appears to work fine. Dyad execution appears to be working (with rank, frame, cell, padding, implemented.) Is there any order to https://github.com/jsoftware/jsource/tree/master/test ? I'm looking for a resource where it is an ever increasin

[Jprogramming] J-like on CUDA / WebAssembly

Hi, 1. There are two questions in this email, but they're closely related. Feel free to reply to either or both. 2. "J-like" here means NumPy/MatLab, but with J syntax. It only needs support tensors of Floats -- no need for Complex, Chars, Boxes, file IO, GUI, ... 3. I know about https://

Re: [Jprogramming] i. (2 2 $ 1 2 3 4)

was > 0 1 2 3 here > > > > On Dec 13, 2017 7:59 AM, "TongKe Xue" wrote: > >> Both explanations are great. Unfortunately, I asked an XY problem. Let >> me re-ask as follows: >> >> I am trying to writ

Re: [Jprogramming] i. (2 2 $ 1 2 3 4)

_ > > Consider the following where the rows and column lengths of i. 1 2 are > made to match those of i. 3 4. > >(i. 1 2),:i. 3 4 > 0 1 0 0 > 0 0 0 0 > 0 0 0 0 > > 0 1 2 3 > 4 5 6 7 > 8 9 10 11 > > But that may not be explanation enou

[Jprogramming] i. (2 2 $ 1 2 3 4)

Hi, I understand what (2 2 $ 1 2 3 4) does. I understand what i. 1 2 does I understand what i. 3 4 does. I have read http://www.jsoftware.com/help/jforc/loopless_code_i_verbs_have_r.htm#_Toc191734331 I understand the concept of verb-rank, of frames + cells, of "promoting one frame to

[Jprogramming] ubuntu jsoftware install

Hi, I'm following the instructions at: http://code.jsoftware.com/wiki/System/Installation/Linux I run into the following issue: sudo dpkg -i j806_amd64.deb ; echo "===" ; /usr/bin/ijconsole ; echo "==="; uname -a (Reading database ... 88939 files and directories currently installed.) Pr