In Advanced view when Activity on any of the items is set to Suspend, it will suspend that item for eternity until 'based on preferences' or 'always' is chosen.

In Simple view there is a Suspend button at the bottom, which suspends calculations for just one hour.
It does the same thing as the Snooze option in the (Windows) right-click BOINC Manager icon in the system tray.

For consistenc

Heston model has accurate density approximations for European option prices, which are of interest.

The module implementing this method should live under tf_quant_finance/volatility/heston_approximation.py. It should support both European option puts and calls approximations. Tests should be in heston_approximation_test.py in the same folder.

In,
https://github.com/flame/blis/blob/master/docs/Performance.md
I noticed that MKL on EPYC was abysmal, but that's just because it uses SSE2 on AMD by default.
You can set the following (undocumented!) environment variable to tell MKL to use AVX2 CPU instructions:
export MKL_DEBUG_CPU_TYPE=5

The docs generated for #431 at https://mratsim.github.io/Arraymancer/pca.html have broken formatting:

• Upstream bug on the type:
  
• Random bolding:
  and
  ![image](https:/

Almost all of Kokkos should be noexcept, but let's break it up into pieces that can be tackled in reasonable-sized issues. This one is the most urgent.

The nonlinear solver from the Sundials library KINSol has the following default behavior:

KINSol limits the maximum Newton update if

|| scale * update ||_2 > max(1000 * || scale * u0 ||_2, 1).

This forces KINSol to take more steps than necessary in the linear case.

Providing the option KINSetMaxNewtonStep(kin_mem, max_step) changes the default behavior to

|| scale * upda

Implement proposed spec maintenance change: KhronosGroup/SYCL-Docs#42

Please describe the problem you are trying to solve?
I would like to export and visualize the interface meshes of a participant using paraview glance.
This is currently not possible as the legacy vtk files are not supported.

Describe the solution you propose.
Add an additional export format which can be imported into glance.

**Describe

@lberti could you commit the falling sphere case (geo+cfg+json) ?
please add a README.adoc to explain the expected results

@vincentchabannes needs it to test new developments

should something like:
--prefix=/local.CPUnative --includedir=/local.CPUnative/include/libflame --libdir=/local.CPUnative/lib/libflame
be expected to make install to respective dirs? Because it has not effect.

make install also fails symlinking if PREFIX/inlude dir already exists.

System: centos 7.2 + devtools 4

The proxy does not show any usefull info when run without any parameter and seems to have not help message. This would really be useful for new users / forgetful people.

$ proxy
fopen: No such file or directory
fopen: No such file or directory

$ proxy --help
fopen: No such file or directory
fopen: No such file or directory

$ proxy -h
fopen: No such file or director

The fp_reduction_latency benchmarks were the very first benchmark, optimization and primitive code tested in Laser.

Unfortunately it is currently very confusing.

It should be reorganized:

• 1. Multiple accumulators: https://github.com/numforge/laser/blob/af191c086b4a98c49049ecf18f5519dc6856cc77/benchmarks/fp_reduction_latency/reduction_bench.nim and https://github.com/numforge/laser/blo

This is currently Cython only and should be relatively easy to write with cpy allowing us to use this on all backends.

As I have mentioned in #36, I fixed the runtime error in level set examples.
However, All three relevant examples are giving strange and messed results than.
See the below pictures.

![image](https://user-images.githubusercontent.com/12082680/31044801-1d8c316e-a609-11e7-9f39-ccb35529

At the very least opencoarrays.F90 should be documented. Test should probably be documented as well as some build system stuff etc.

I recently stumbled upon another test framework, called Catch2. It looks very interesting for us and is already used in couple of big open source projects. The biggest advantage is that it is a single header include. So we do not have to build GoogleTest anymore and we do not need t

For large parameter sets the optimizers can take a non-trivial amount of time. They lack timers to indicate how long the optimization took vs the configuration generation.