-
Matrix arithmetic and vector-matrix computations.
-
1. When we build Pariser-Parr dictionary, we assume user provides connectivity matrix as ('atomx, 'atomy', 'Rxy'). Where $R_{xy}$ is a distance between atoms x and y.
2. To compute hopping term $\…
-
Here are some goals we set at the 4/15 meeting:
- "Nearby" function to get a close Toeplitz, Hankel, Cauchy, etc. approximation to a given matrix
- "Norms" fast norm computations for our structure…
-
Compute the multiplication of the interpolation matrix and the basis vector only once to accelerate the online stage :
$\alpha_i^{N,H} = (u_H^\mathcal{N}, \xi_i^h)_{L^2}=(u_H^\mathcal{N})^T (M^{…
-
### Is your feature request related to a problem? Please describe.
it is not a problem by itself, but will allow improved efficiency in scipy.optimize.least_squares. Such function, asks for separate …
-
It should be possible to delegate matrix multiplication to several threads to improve the performance. This is what NumPy seems to do for floating-point computations. We are still faster for integer/f…
-
It would nice to have basic density matrix related functionalities that are part of basic quantum information theory computations (partial trace, entropy, fidelity, purity..)
-
I've been trying to use pykeops to generate rows of a kernel matrix. However, when I change the kernel hyperparameter sigma from an int to a float (5 to 5.), the outputs become nan. Do the developers …
-
The main `provoc()` function would still work as intended, but would accept list arguments to many of the parameters. The computations would be moved to another function that is easy to call from `pro…
-
Hi there,
I'm new to quantization. From my understanding, "8da4w" means that the weights are pre-quantized to 4 bits, and the activations are quantized to 8 bits at runtime. Following this, the GEM…