[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [science-iwg] Eclipse Proposal - Matrix Algorithms
|
HI all,
As Erwin says, to get good performance most matrix algorithms need to know or dictate data layout so this will mean implementations are specialized to data structures.
It may be possible with some effort to work out a common interface that performs well for the algorithms. This will allow all data class packages to supply specific adapters to use the proposed package. Is this feasible?
Regards,
Peter
-----Original Message-----
From: science-iwg-bounces@xxxxxxxxxxx <science-iwg-bounces@xxxxxxxxxxx> On Behalf Of erwindl0@xxxxxxxxx
Sent: 20 June 2019 11:38
To: 'Science Industry Working Group' <science-iwg@xxxxxxxxxxx>
Subject: Re: [science-iwg] Eclipse Proposal - Matrix Algorithms
Hi Philip,
Interesting proposition.
I'm not that experienced in this domain, but wouldn't this always be linked to a concrete a-priori data structure / matrix representation?
E.g. January or ND4J or ... have their own multi-dimensional array representations.
Or would this project provide its own matrix structure API and implementations (and bridges to some other well-known projects' data structures)?
cheers
erwin
-----Original Message-----
From: science-iwg-bounces@xxxxxxxxxxx <science-iwg-bounces@xxxxxxxxxxx> On Behalf Of Philip Wenig
Sent: Thursday, 20 June 2019 09:48
To: science-iwg@xxxxxxxxxxx
Subject: [science-iwg] Eclipse Proposal - Matrix Algorithms
Hi folks,
does it make sense to file a new Eclipse Proposals "Matrix Algorithms"
under the umbrella of the Science working group?
We have dozens of algorithms in ChemClipse/OpenChrom which we would like to migrate to a separate project. The goal is, that they can be reused more easily in other projects. I assume
Diamond Light Source
Oak Ridge
deeplearning4j
...
are facing the same problem. At the moment, these algorithms are part of ChemClipse, DAWNSci, deeplearning4j ... and to use them, you have to clone the complete project even if you only need a certain algorithm. We could provide the following algorithms (implementations) in Java:
SVD (Singular Value Decomposition)
NIPALS (Non-linear Iterative Partial Least Squares) OPLS (Orthogonal Projections to Latent Structures)
Currently, we are working on:
MCR-AR (Multivariate Curve Resolution - Alternating Regression)
In the future, we are interested to implement:
PARAFAC2 (Parallel Factor Analysis 2)
TCC (Tucker's Congruence Coefficient)
Java isn't that strong in matrix operations, though we could use matrix libraries and/or create the algorithms in a way that different matrix libraries (Java, C, ...) can be used, e.g.:
ND4J (http://nd4j.org)
Eclipse January (https://www.eclipse.org/january) EJML (http://ejml.org) OpenBLAS (https://www.openblas.net)
In my opinion, we should have a dependency to Apache Commons Math and setup the project in an encapsulated way, similar as SWTChart for plotting charts (it requires only double[] arrays to plot the data).
https://commons.apache.org/proper/commons-math
https://github.com/eclipse/swtchart
What's your opinion?
Feedback is highly appreciated.
Best,
Philip
--
~~~~~~~~~~~~~~~~~~~~~~~~
OpenChrom - the open source alternative for chromatography / mass spectrometry Dr. Philip Wenig » Founder » philip.wenig@xxxxxxxxxxxxx » http://www.openchrom.net ~~~~~~~~~~~~~~~~~~~~~~~~
_______________________________________________
science-iwg mailing list
science-iwg@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/science-iwg
_______________________________________________
science-iwg mailing list
science-iwg@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/science-iwg
--
This e-mail and any attachments may contain confidential, copyright and or privileged material, and are for the use of the intended addressee only. If you are not the intended addressee or an authorised recipient of the addressee please notify us of receipt by returning the e-mail and do not use, copy, retain, distribute or disclose the information in or attached to the e-mail.
Any opinions expressed within this e-mail are those of the individual and not necessarily of Diamond Light Source Ltd.
Diamond Light Source Ltd. cannot guarantee that this e-mail or any attachments are free from viruses and we cannot accept liability for any damage which you may sustain as a result of software viruses which may be transmitted in or with the message.
Diamond Light Source Limited (company no. 4375679). Registered in England and Wales with its registered office at Diamond House, Harwell Science and Innovation Campus, Didcot, Oxfordshire, OX11 0DE, United Kingdom
