Видео

I started to learn D about 1 month ago. Until now I like it.

YOOO interfaces and traits!! this will be huge holy shit

Possibly the best final slide of all time.

I was reading your paper. This is very hard to understand without your video. Danke Schoen!

I had issues with the fakeroot since it was not configured on my local hpc cluster you might have to this manually on the login node with "sudo singularity config fakeroot --add root"

Great Points! Thanks for help! 😊 😊

Huge D fan here! Thanks for the video.

I really am interested in this working to remove unnecessary language gaps that lays out bear traps for the unwary. The binary protocol handling reminds me of the work I did with VAX PASCAL handling the compressed binary exchange feeds and market data. One defines the structures and permutations thereof and one could then just access the fields. No pointer arithmetic or bit shifting.

Thank you! Very simple and straightforward compared to building from source

After a huge search on RASER simulation, I got this video. The installation process is so complex because of its dependencies. I am stuck for Fenics module. Can you add me to your technical team?

I was able to follow this on 11/15/2023, logged in (KeyCloak -> CI Logon), got the BinderHub landing, pasted your matthewfeickert/pyhep-notebook-talk-example repository, the image was still in Harbor, pulled the image, and the notebook launched within 2 minutes, talk.ipynb worked perfectly. Thanks Matthew!

D is actually kinda cool

Is this the same D language that Oracle used for DTrace?

D best language !!

Where can i get the notebook that is being used in this video?

Appreciation is Due.

in iminuit, how can I limit the fit range , e.g., I only want to fit data between x= 100 to x=500? Is it possible to use the m.limits for the data range?

Very good presentation ! I'll do the proposed method in my PHd work. I'm encountering problems to train my models with negative weights. I 'm only finding difficulties to understand in which sense the "likelihood ratio", the so called observable, is unbinned. We can't cluster events expecting them to have exact the same observable value. The proposed observable will be represented as a histogram and we'll cluster the events which reside in chosen real number intervals of the produced histogram. How can it be unbinned ?

I got the gaussian one to run with cuda. It is faster on my PC than the parallel loop version without copying back to the host and with warming up the GPU a bit (i.e. run it a few times). @cuda.jit def cuda_guass_2d(output, height, width, scale): i, j = cuda.grid(2) i_in_bounds = 0 < i < width j_in_bounds = 0 < j < height if i_in_bounds and j_in_bounds: x = (i - width / 2) * (10 / width) y = (j - height / 2) * (10 / height) norm = x**2 / 2 + y**2 / 2 taylor_series = 1 + norm + norm**2 / 2 # Numba cuda supports math.exp output[i, j] = scale / taylor_series def execute_gauss_on_cuda(height, width): grid_cuda = cuda.device_array((height, width)) block_dim = (16, 16) grid_dim = ((height // block_dim[0]) + 1, (width // block_dim[1]) + 1) scale = 1. / np.sqrt(2 * np.pi) cuda_guass_2d[grid_dim, block_dim](grid_cuda, height, width, scale) cuda.synchronize() return grid_cuda

awesome explanation, never thought this concept can be this easy

Can you share the link for the slides presented here?

Hello have to 2x for loop through a matrix by (1360 x 1024 ), and (slice aout a vector of 128 in length )it takes hell of time to do, how can I speed it up with Numba an CUDA, very much thanks.😃

the github repo with the notebook: github.com/aoeftiger/pyhep2020 the slides: aoeftiger.github.io/pyhep2020/

go to 1:18 to skip the initial fixing the screen sharing

Зачёт, не фига не понятно, но очень интересно :-)

That's really amazing. I also work on di higgs analysis by ggF at integrated luminosity of 137fb-1. For the hh->bbtautau. It helpful please share the link about these descriptions.