Operator Discretization Library: Difference between revisions
From EverybodyWiki Bios & Wiki
m Remove Afc Messages and templates |
m Moved page from wikipedia:en:Draft:Operator Discretization Library (history) |
||
| Line 1: | Line 1: | ||
{{AFC submission|d|nn|u=A ringh|ns=118|decliner=Curb Safe Charmer|declinets=20200209202943|ts=20190930032409}} <!-- Do not remove this line! --> | |||
{{AFC submission|d|v|u=A ringh|ns=118|decliner=Missvain|declinets=20190828221334|small=yes|ts=20190523073825}} <!-- Do not remove this line! --> | |||
{{AFC comment|1=I am not convinced that any of the papers cited are ''about'' ODL. I suspect they are merely ''mentions'' of ODL. [[User:Curb Safe Charmer|Curb Safe Charmer]] ([[User talk:Curb Safe Charmer|talk]]) 20:29, 9 February 2020 (UTC)}} | {{AFC comment|1=I am not convinced that any of the papers cited are ''about'' ODL. I suspect they are merely ''mentions'' of ODL. [[User:Curb Safe Charmer|Curb Safe Charmer]] ([[User talk:Curb Safe Charmer|talk]]) 20:29, 9 February 2020 (UTC)}} | ||
| Line 19: | Line 22: | ||
'''Operator Discretization Library''' (ODL) is an [[Open-source software|open source]] toolbox for [[Inverse problem|inverse problems]] written in [[Python (programming language)|python]]. | '''Operator Discretization Library''' (ODL) is an [[Open-source software|open source]] toolbox for [[Inverse problem|inverse problems]] written in [[Python (programming language)|python]]. | ||
The main goal of ODL is to facilitate the use of advanced mathematical tools for real-world problems, without having to implement all necessary parts from the bottom up. The focus of the toolbox is fast prototyping in inverse problems. Its main audience is mathematicians and applied scientists in imaging. | The main goal of ODL is to facilitate the use of advanced mathematical tools for real-world problems, without having to implement all necessary parts from the bottom up. The focus of the toolbox is fast prototyping in inverse problems.<ref name="odl-git"/><ref name="zenodo"/><ref name="ringh2017"/> Its main audience is mathematicians and applied scientists in imaging. | ||
<ref name="matej2017"/><ref name="adler2017 | <ref name="matej2017"/><ref name="adler2017"/><ref name="chen2018"/><ref name="buurlage2018"/><ref name="zickert2018"/><ref name="marlevi2019"/>{{excessive citations inline|date=February 2020}} | ||
== History == | == History == | ||
ODL was initially developed at [[KTH Royal Institute of Technology]] as part of a research project funded by [[Swedish Foundation for Strategic Research]].<ref name=" | ODL was initially developed at [[KTH Royal Institute of Technology]] as part of a research project funded by [[Swedish Foundation for Strategic Research]].<ref name="odl-git"/><ref name="ssf-webpage"/> | ||
It is currently being developed at <ref name="zenodo"/> | It is currently being developed at <ref name="zenodo"/> | ||
| Line 32: | Line 35: | ||
* [[Elekta]] | * [[Elekta]] | ||
* [[Thermo Fisher Scientific]] | * [[Thermo Fisher Scientific]] | ||
==Solving inverse problems== | |||
A real world domain is converted into a software with a [[Internal_model_(motor_control)#Forward_models|forward model]]. Bringing a forward model into a goal state by finding the input values is called an [[inverse problem]].<ref name=adler2017/> The Operator Discretization Library was developed on top of the [[tensorflow]] library<ref>{{cite arxiv |title=Neural network augmented wave-equation simulation |author=Siahkoohi, Ali and Louboutin, Mathias and Herrmann, Felix J |year=2019 |class=physics.comp-ph |eprint=1910.00925 }}</ref> to solve these inverse problems with [[State space planning|state space sampling]]. Unfortunately, state space sampling<ref>{{cite journal |doi=10.1137/17m1134834 |year=2018 |publisher=Society for Industrial \& Applied Mathematics (SIAM) |volume=28 |number=4 |pages=2783–2808 |author=Antonin Chambolle and Matthias J. Ehrhardt and Peter Richtarik and Carola-Bibiane Schönlieb |title=Stochastic Primal-Dual Hybrid Gradient Algorithm with Arbitrary Sampling and Imaging Applications |journal=SIAM Journal on Optimization }}</ref> and other optimization techniques like gradient descent<ref>{{cite journal |title=Indirect Image Registration: Geodesic Shooting for LDDMM and Deep Learning Approaches |author=Kowalewski, Rosa |year=2018 }}</ref> take a large amount of execution time. An example problem will need up to 40 hours on a GTX 1080 TI graphics card until the Operator Discretization Library has detected a [[Computational human phantom]].<ref>{{cite journal |doi=10.1109/tmi.2018.2799231 |year=2018 |publisher=Institute of Electrical and Electronics Engineers (IEEE) |volume=37 |number=6 |pages=1322–1332 |author=Jonas Adler and Ozan Oktem |title=Learned Primal-Dual Reconstruction |journal=IEEE Transactions on Medical Imaging |pmid=29870362 |arxiv=1707.06474 }}</ref> | |||
==References== | ==References== | ||
{{Reflist|refs= | {{Reflist|refs= | ||
<ref name=matej2017>{{cite journal |last1=Matěj |first1=Zdeněk |last2=Mokso |first2=Rajmund |last3=Larsson |first3=Krister |last4=Hardion |first4=Vincent |last5=Spruce |first5=Darren |date=2017 |title=The MAX IV imaging concept | <ref name=matej2017>{{cite journal |last1=Matěj |first1=Zdeněk |last2=Mokso |first2=Rajmund |last3=Larsson |first3=Krister |last4=Hardion |first4=Vincent |last5=Spruce |first5=Darren |date=2017 |title=The MAX IV imaging concept |journal=Advanced Structural and Chemical Imaging |volume=2 |issue=1 |pages=16 |doi=10.1186/s40679-016-0029-7|pmid=28003953 |pmc=5133273 }}</ref> | ||
<ref name=adler2017>{{cite journal |last1=Adler |first1=Jonas |last2=Öktem |first2=Ozan |date=2017 |title=Solving ill-posed inverse problems using iterative deep neural networks | <ref name=adler2017>{{cite journal |last1=Adler |first1=Jonas |last2=Öktem |first2=Ozan |date=2017 |title=Solving ill-posed inverse problems using iterative deep neural networks |journal=Inverse Problems |volume=33 |issue=12 |pages=124007 |doi=10.1088/1361-6420/aa9581|arxiv=1704.04058 |bibcode=2017InvPr..33l4007A }}</ref> | ||
<ref name=ringh2017>{{cite book |last1=Ringh |first1=Axel |last2=Zhuge |first2=Xiaodong |last3=Palenstijn |first3=Willem Jan |last4=Batenburg |first4=Kees Joost |last5=Öktem |first5=Ozan |editor1-last=Kropatsch |editor1-first=Walter G. |editor2-last=Artner |editor2-first=Nicole M. |editor3-last=Janusch |editor3-first=Ines |title=Discrete Geometry for Computer Imagery. |publisher=Springer, Cham |date=2017 |pages= | <ref name=ringh2017>{{cite book |last1=Ringh |first1=Axel |last2=Zhuge |first2=Xiaodong |last3=Palenstijn |first3=Willem Jan |last4=Batenburg |first4=Kees Joost |last5=Öktem |first5=Ozan |editor1-last=Kropatsch |editor1-first=Walter G. |editor2-last=Artner |editor2-first=Nicole M. |editor3-last=Janusch |editor3-first=Ines |title=Discrete Geometry for Computer Imagery. |publisher=Springer, Cham |date=2017 |pages=109–121 |chapter=High-Level Algorithm Prototyping: An Example Extending the TVR-DART Algorithm |isbn=978-3-319-66272-5 |doi=10.1007/978-3-319-66272-5_10|url=https://ir.cwi.nl/pub/26852 }}</ref> | ||
<ref name=chen2018>{{cite journal |last1=Chen |first1=Chong |last2=Öktem |first2=Ozan |date=2018 |title=Indirect image registration with large diffeomorphic deformations | <ref name=chen2018>{{cite journal |last1=Chen |first1=Chong |last2=Öktem |first2=Ozan |date=2018 |title=Indirect image registration with large diffeomorphic deformations |journal=SIAM Journal on Imaging Sciences |volume=11 |issue=1 |pages=575–617 |doi=10.1137/17M1134627|arxiv=1706.04048 }}</ref> | ||
<ref name=buurlage2018>{{cite journal |last1=Buurlage |first1=Jan-Willem |last2=Kohr |first2=Holger |last3=Palenstijn |first3=Willem Jan |last4=Batenburg |first4=Kees Joost| date=2018 |title=Real-time quasi-3D tomographic reconstruction | <ref name=buurlage2018>{{cite journal |last1=Buurlage |first1=Jan-Willem |last2=Kohr |first2=Holger |last3=Palenstijn |first3=Willem Jan |last4=Batenburg |first4=Kees Joost| date=2018 |title=Real-time quasi-3D tomographic reconstruction |journal=Measurement Science and Technology |volume=29 |issue=6 |pages=064005 |doi=10.1088/1361-6501/aab754|bibcode=2018MeScT..29f4005B |url=https://ir.cwi.nl/pub/27711 }}</ref> | ||
<ref name=zickert2018>{{cite journal |last1=Zickert |first1=Gustav |last2=Maretzke |first2=Simon |date=2018 |title=Cryogenic electron tomography reconstructions from phaseless data | <ref name=zickert2018>{{cite journal |last1=Zickert |first1=Gustav |last2=Maretzke |first2=Simon |date=2018 |title=Cryogenic electron tomography reconstructions from phaseless data |journal=Inverse Problems |volume=34 |issue=12 |pages=124001 |doi=10.1088/1361-6420/aade22|bibcode=2018InvPr..34l4001Z }}</ref> | ||
<ref name=marlevi2019>{{cite thesis |last=Marlevi |first=David |date=2019 |title=Non-invasive imaging for improved cardiovascular diagnostics |type=PhD |publisher=KTH Royal Institute of Technology |url=http://www.diva-portal.org/smash/record.jsf?pid=diva2%3A1344648}}</ref> | <ref name=marlevi2019>{{cite thesis |last=Marlevi |first=David |date=2019 |title=Non-invasive imaging for improved cardiovascular diagnostics |type=PhD |publisher=KTH Royal Institute of Technology |url=http://www.diva-portal.org/smash/record.jsf?pid=diva2%3A1344648}}</ref> | ||
<ref name="zenodo">{{cite | <ref name="zenodo">{{cite journal |url=https://zenodo.org/record/1442734 |title=odlgroup/odl: ODL 0.7.0 |first1=Jonas |last1=Adler |first2=Holger |last2=Kohr |first3=Axel |last3=Ringh |first4=Julian |last4=Moosmann |first5=Sebastian |last5=Banert |first6=Matthias J. |last6=Ehrhardt |first7=Gregory R. |last7=Lee |first8=Kati |last8=Niinimäki |first9=Barbara |last9=Gris |first10=Olivier |last10=Verdier |first11=Johan |last11=Karlsson |first12=Gustav |last12=Zickert |first13=Willem Jan |last13=Palenstijn |first14=Ozan |last14=Öktem |first15=Chong |last15=Chen |first16=Hector |last16=Andrade Loarca |first17=Michael |last17=Lohmann |journal=Zenodo |date=9 September 2018 |doi=10.5281/zenodo.1442734 |bibcode=2018zndo...1442734A |access-date=20 September 2019 |display-authors=2}}</ref> | ||
<ref name=ssf-webpage>{{cite web |url=https://strategiska.se/forskning/pagaende-forskning/applied-mathematics-2013/projekt/5955/ |title=Lågkomplexitetsrekonstruktionsmetoder för medicin |author=<!--Not stated--> |access-date=20 September 2019}}</ref> | <ref name=ssf-webpage>{{cite web |url=https://strategiska.se/forskning/pagaende-forskning/applied-mathematics-2013/projekt/5955/ |title=Lågkomplexitetsrekonstruktionsmetoder för medicin |author=<!--Not stated--> |access-date=20 September 2019}}</ref> | ||
| Line 58: | Line 64: | ||
==External links== | ==External links== | ||
* [https://odlgroup.github.io/odl/ Documentation] | * [https://odlgroup.github.io/odl/ Documentation] | ||
* [https://doi.org/10.5281/zenodo.1442734 Code published on Zenodo] | * [https://doi.org/10.5281/zenodo.1442734 Code published on Zenodo] | ||
{{Source Wikipedia}} | {{Source Wikipedia}} | ||
Revision as of 02:42, 14 February 2021
| Original author(s) | Jonas Adler, Holger Kohr and Ozan Öktem |
|---|---|
| Developer(s) | The ODL Development Team |
| Initial release | 10 February 2015 |
| Written in | Python |
| Engine | |
| Operating system | Linux, macOS, Microsoft Windows |
| Type | Mathematical software |
| License | Mozilla Public License 2.0 |
| Website | github |
Search Operator Discretization Library on Amazon.
Operator Discretization Library (ODL) is an open source toolbox for inverse problems written in python.
The main goal of ODL is to facilitate the use of advanced mathematical tools for real-world problems, without having to implement all necessary parts from the bottom up. The focus of the toolbox is fast prototyping in inverse problems.[1][2][3] Its main audience is mathematicians and applied scientists in imaging. [4][5][6][7][8][9][excessive citations]
History
ODL was initially developed at KTH Royal Institute of Technology as part of a research project funded by Swedish Foundation for Strategic Research.[1][10] It is currently being developed at [2]
- KTH Royal Institute of Technology
- Technical University of Berlin
- Centrum Wiskunde & Informatica
- University of Cambridge
- Elekta
- Thermo Fisher Scientific
Solving inverse problems
A real world domain is converted into a software with a forward model. Bringing a forward model into a goal state by finding the input values is called an inverse problem.[5] The Operator Discretization Library was developed on top of the tensorflow library[11] to solve these inverse problems with state space sampling. Unfortunately, state space sampling[12] and other optimization techniques like gradient descent[13] take a large amount of execution time. An example problem will need up to 40 hours on a GTX 1080 TI graphics card until the Operator Discretization Library has detected a Computational human phantom.[14]
References
- ↑ 1.0 1.1 "ODL Github repository". Retrieved 23 May 2019.
- ↑ 2.0 2.1 Adler, Jonas; Kohr, Holger; et al. (9 September 2018). "odlgroup/odl: ODL 0.7.0". Zenodo. Bibcode:2018zndo...1442734A. doi:10.5281/zenodo.1442734. Retrieved 20 September 2019.
- ↑ Ringh, Axel; Zhuge, Xiaodong; Palenstijn, Willem Jan; Batenburg, Kees Joost; Öktem, Ozan (2017). "High-Level Algorithm Prototyping: An Example Extending the TVR-DART Algorithm". In Kropatsch, Walter G.; Artner, Nicole M.; Janusch, Ines. Discrete Geometry for Computer Imagery. Springer, Cham. pp. 109–121. doi:10.1007/978-3-319-66272-5_10. ISBN 978-3-319-66272-5. Search this book on
- ↑ Matěj, Zdeněk; Mokso, Rajmund; Larsson, Krister; Hardion, Vincent; Spruce, Darren (2017). "The MAX IV imaging concept". Advanced Structural and Chemical Imaging. 2 (1): 16. doi:10.1186/s40679-016-0029-7. PMC 5133273. PMID 28003953.
- ↑ 5.0 5.1 Adler, Jonas; Öktem, Ozan (2017). "Solving ill-posed inverse problems using iterative deep neural networks". Inverse Problems. 33 (12): 124007. arXiv:1704.04058. Bibcode:2017InvPr..33l4007A. doi:10.1088/1361-6420/aa9581.
- ↑ Chen, Chong; Öktem, Ozan (2018). "Indirect image registration with large diffeomorphic deformations". SIAM Journal on Imaging Sciences. 11 (1): 575–617. arXiv:1706.04048. doi:10.1137/17M1134627.
- ↑ Buurlage, Jan-Willem; Kohr, Holger; Palenstijn, Willem Jan; Batenburg, Kees Joost (2018). "Real-time quasi-3D tomographic reconstruction". Measurement Science and Technology. 29 (6): 064005. Bibcode:2018MeScT..29f4005B. doi:10.1088/1361-6501/aab754.
- ↑ Zickert, Gustav; Maretzke, Simon (2018). "Cryogenic electron tomography reconstructions from phaseless data". Inverse Problems. 34 (12): 124001. Bibcode:2018InvPr..34l4001Z. doi:10.1088/1361-6420/aade22.
- ↑ Marlevi, David (2019). Non-invasive imaging for improved cardiovascular diagnostics (PhD). KTH Royal Institute of Technology.
- ↑ "Lågkomplexitetsrekonstruktionsmetoder för medicin". Retrieved 20 September 2019.
- ↑ Siahkoohi, Ali and Louboutin, Mathias and Herrmann, Felix J (2019). "Neural network augmented wave-equation simulation". arXiv:1910.00925 [physics.comp-ph].CS1 maint: Multiple names: authors list (link)
- ↑ Antonin Chambolle and Matthias J. Ehrhardt and Peter Richtarik and Carola-Bibiane Schönlieb (2018). "Stochastic Primal-Dual Hybrid Gradient Algorithm with Arbitrary Sampling and Imaging Applications". SIAM Journal on Optimization. Society for Industrial \& Applied Mathematics (SIAM). 28 (4): 2783–2808. doi:10.1137/17m1134834.
- ↑ Kowalewski, Rosa (2018). "Indirect Image Registration: Geodesic Shooting for LDDMM and Deep Learning Approaches".
- ↑ Jonas Adler and Ozan Oktem (2018). "Learned Primal-Dual Reconstruction". IEEE Transactions on Medical Imaging. Institute of Electrical and Electronics Engineers (IEEE). 37 (6): 1322–1332. arXiv:1707.06474. doi:10.1109/tmi.2018.2799231. PMID 29870362.
External links
This article "Operator Discretization Library" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Operator Discretization Library. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
