Incorrect results from hypergeom function in MATLAB R2018a.
7 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
yann
el 24 de Mzo. de 2024
Comentada: Dyuman Joshi
el 19 de Abr. de 2024
In view of the following exact result:
2F1(1-n,1+n;3/2;1/2) =sin(nπ/2)/n
one would expect the result:
hypergeom([1−n,1+n],1.5,0.5)=0 if n is even.
However, for n=300, Matlab R2018a yields:
hypergeom([1−n,1+n],1.5,0.5)=Inf.
Why does this happen?
3 comentarios
Manikanta Aditya
el 24 de Mzo. de 2024
Mostly the issue here is due to the limitations of numerical precision and the way 'hypergeom' function handles larger inputs.
In your case, when n=300, the parameters of the hypergeom function become very large ([1−300,1+300]), which can lead to numerical instability and might result in MATLAB returning Inf.
If you need to compute the hypergeometric function for large parameters, you might need to use more specialized numerical methods or software that can handle such cases.
Thanks!
Respuesta aceptada
Amal Raj
el 16 de Abr. de 2024
The difference is likely due to improvements or changes in MATLAB's numerical algorithms between versions R2016b and R2018b. While such discrepancies can be confusing, they often reflect efforts to improve the accuracy and reliability of numerical computations. For critical applications, it's always a good practice to check the release notes of MATLAB for any changes in numerical functions you rely on and to consider the implications of numerical precision and stability in your computations.
1 comentario
Dyuman Joshi
el 19 de Abr. de 2024
FYI - Release Notes for R2016b and R2018b are not available on the online MATLAB Release Notes documentation page. (The earliest one available right now is of R2019a, same as function documentations - only the information for last 5 years is available online)
I suppose one would have to contact the support team for that. And they would provide a better answer and clarification regarding the same as well, with appropriate information/explaination.
Más respuestas (0)
Ver también
Categorías
Más información sobre Special Functions en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!