YaccConstructor

[YaccConstructor @YaccConstructor]

@VictorTaelin Cool! A few technical questions.
Why 3 dimensions is the best choice? Why exactly 48 bits for pointers and other parts? What about huge data processing? Yes, all can be encoded as lambda-tetm, but this term should be stored. So, data transfer bottleneck still here.

[YaccConstructor @YaccConstructor]

Try fast linear algebra based static code analysis out: https://t.co/OqxF7mM1DQ

#cfpq #cfl #graphblas

[YaccConstructor @YaccConstructor]

Blazingly fast static code analysis with @GraphBLAS https://t.co/DXploWlahf

[YaccConstructor @YaccConstructor]

Gll-based #cfpq for #Neo4j: https://t.co/UYYiasMwwo

[YaccConstructor @YaccConstructor]

#gpu-accelerated linear algebra based graph analysis in #Python: https://t.co/FUtD9kM0Mb
Thanks to #opencl , spla is available on #Intel #AMD #Nvidia gpus

[YaccConstructor @YaccConstructor]

@breandan Yes, unfortunately implementation missed. Careful implementation and evaluation of the proposed algorithm may be an interesting challange.

Regarding datalog. This way to partial evaluation looks interesting: https://t.co/NPu9HGfbad

[YaccConstructor @YaccConstructor]

Let's make #CFPQ faster! Again. GLL + RSM in #kotlin: https://t.co/JBt4H2zr4q

Thanks to @vadyushkins

[YaccConstructor @YaccConstructor]

Linear algebra based @GraphBLAS-inspired graph analysis can works fine on #amd #intel #nvidia #gpu : https://t.co/YfuxJavodo
Thanks to #opencl

[HAFLANG - Functional Languages in Hardware @HaflangProject]

"Relaxing Reduceron's Template Constraints", a HAFLANG project memo by Craig Ramsay.

Relaxing the single-cycle property for beta reductions and with a spine splitting optimisation, results in 11.6% fewer cycles and 8% smaller generated template programs.

https://t.co/6rpFpowJ5j

[YaccConstructor @YaccConstructor]

@breandan Thus you can express arbitrary prefixes and suffixes added to \alpha. And when you parse such a graph, you should to check about 2|\alpha| cells for emptiness.

[YaccConstructor @YaccConstructor]

@breandan I'm not sure that your figure is about your case. I assume that you want to check the property for the given grammar, given \alpha, and arbitrary \beta and \gamma. Your figure is for fixed \beta and \gamma. What I propose is to surround alpha with two loops to express \Sigma^*

[YaccConstructor @YaccConstructor]

@breandan Looking forward for new details!

[YaccConstructor @YaccConstructor]

Brahma.FSharp 2.0 published: https://t.co/1DgJqfdE9E
- Code quotations for kernels specification: you can write flexible generic kernels in F#
- OpenCL as target: you are not limited by Nvidia devices
- Mailbox processor for host-device communication
#fsharp #gpu #opencl #dotnet

[YaccConstructor @YaccConstructor]

GLL-based CFPQ for @neo4j https://t.co/bbC8PmEmbT For multiple sources it is faster than @GraphBLAS-based CFPQ. Even in provenance-aware mode. Sometimes it is faster even for all-pairs reachability. GraphBLAS-based solution is highly parallel, while GLL-based is single threaded.

[YaccConstructor @YaccConstructor]

@breandan Moreover it is possible to use MCFGs as constraints for path querying. Graph query languages can be even more expressive. Respective algorithm under development.

[Aydın Buluç @aydoz]

This Thursday March 24th (10 AM PDT), I am going to give a virtual seminar for SPCL_BCast on "Parallel Sparse Matrix Algorithms for Data Analysis and Machine Learning". Connection info, abstract, etc can be found https://t.co/C0Pt2n9fgp. Thanks @spcl_eth for inviting me.

[YaccConstructor @YaccConstructor]

Our paper "Rational index of languages with bounded dimension of parse trees" is accepted to be presented at DLT-2022 https://t.co/23176nQRr9

[JetBrains @jetbrains]

JetBrains Research is an association of 20 scientific labs and groups that do research in various fields, including mathematics, physics, computational biology, robotics, and machine learning.
In this report, the lab leaders discuss their teams' work:
https://t.co/2o8PGuVbtW