Numerical verification script for a universal baseline constant — reproducible SageMath code
[supprimé]
[supprimé]
Woxi is a free open source interpreter for the Wolfram Language,
written in Rust. The guiding rule of the project: whatever it implements
must produce exactly the same output as wolframscript / Mathematica — the test suite
(26'000+ unit tests) compares against it directly.
Version 0.3.0 is out now. Highlights since 0.2.0:
.nb files — still far from all
of them, but dozens (Kepler's Second Law, Parabolic Mirror, Doyle
Spirals, …) serve as end-to-end tests. Manipulate covers a good part
of the Demonstrations control vocabulary (interactive Locators,
Trigger, Button, SetterBar, PopupMenu, controls nested in
TabView/Grid, dependent controls, TrackedSymbols, Animate
bodies, …), and typeset notebook input is read back as code.Around error bars, Callout, Filling, PlotMarkers
and friends, plus new ComplexListPlot and ListLinePlot3D.Integrate/D/Series/Limit coverage,
LinearProgramming with an exact simplex solver, NonlinearModelFit,
Reduce over higher-degree polynomial inequalities, RootReduce, and
many canonical-ordering fixes so results print exactly as
wolframscript prints them.Try it without installing anything in our playground and JupyterLite instance in the browser: https://woxi.ad-si.com
Install: cargo install woxi, pip install woxi (Python bindings), or
npm install woxi-wasm. Full changelog:
https://github.com/ad-si/Woxi/blob/main/changelog.md
If Woxi is useful to you, I'm now accepting donations on GitHub: https://github.com/sponsors/ad-si I hope to receive enough support to keep the project going for a long time, and to eventually reach full parity with Wolfram's implementation.
Square root series
Hi! , i just derived a sequence that calculates square roots, but I don't know if it is already proven, found, named or anything as I can't find anything about it online,
The max i found was some babylonian approximation that was wildly inaccurate and just some approximations.
Can anyone help me verify if it is already found?
√n = a + sum_i=0_∞ (e_i)
a is the nearest smaller perfect square
e or the error terms help with accuracy
e_v = (n-(a+e_v-1)^2)/(2a+1)
Where i have derived that
e_0 =( n-a^2)/(2a+1)
And
From e_0 , i found that e_-1=0 ,
I have already tried it on many numbers and it has great approximations like 1-2 decimal places per error term
You can use it yourself and tell me if you found it useful and also tell me if it is already found or derived, so if not, i would publish the proofs
Streaming audio is quite cumbersome in WL. It would be nice to have speakers and a mic as a part of Wolfram Device Framework. Unfortunately, there are only “Camera”, “SerialPort” available.
Jushua and me developed a small library on top of Device Framework and miniaudio.c
https://github.com/JerryI/AudioDriverWL
Now you can simply DeviceOpen[“Microphone”] and have an access to raw numeric array from OS ring buffer or Audio object if you wish. It supports all Wolfram platforms (Mathematica, wolframscript, WLJS). Binaries were prebuilt for Win and Mac only for the moment
Both: sync (poll device in a loop) and async buffer operations (use callback when data arrives) are supported
I tested it with built-in speakers, headphones mic as well as with iRig 2 HD. The latency for streaming (Mic-WL Handler-Speakers) on my Mac is around 60ms or less if you set lower sample rate. Which can still work as a horrible guitar pedal :D