We're sorry but this page doesn't work properly without JavaScript enabled. Please enable it to continue.
Feedback

Radio Astronomy with Python

00:00

Formal Metadata

Title
Radio Astronomy with Python
Subtitle
Gaussian Processes and Neural Networks applied to photometric redshift reconstruction
Title of Series
Number of Parts
130
Author
License
CC Attribution - NonCommercial - ShareAlike 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this
Identifiers
Publisher
Release Date
Language

Content Metadata

Subject Area
Genre
Abstract
Looking at higher redshifts is equivalent to looking back in time: they improve the studies of cosmology, expanding our knowledge of the universe. It allows us to study various physical phenomena like the power spectrum of galaxies which describes the distribution of galaxies on a range of scales, galaxy clustering, and large scales, the detection of the Baryon Acoustic Oscillation feature. As a result, a significant amount of work has been done to increase the efficiency and accuracy of the process via new algorithms and optimization of existing ones. Astronomical datasets are undergoing a rapid growth in size and complexity as past, ongoing and future surveys produce massive multi-temporal and multi-wavelength datasets, with huge information to be extracted and analyzed. The alternative to a full spectroscopic survey is to obtain multi-color images of the sky and perform photometric redshift estimates for the galaxies we have available. When dealing with this problem, there are two main approaches: model-driven data analysis (template fitting methods) and data-driven analysis, which can use machine learning methods. To solve this problem, we use data-driven analysis, more specifically GPz (which uses Gaussian processes) and ANNz2 (which mainly uses neural networks), both python software. Prerequisites: machine learning and basic math knowledge
61
Thumbnail
26:38
95
106
MathematicsDoppler-EffektSpectrum (functional analysis)Shift operatorFrequencySource codeNumberWaveRevision controlUniverse (mathematics)State observerTheory of relativityoutputObject (grammar)Task (computing)Projective planeMultiplication signStudent's t-testBus (computing)Graph coloringCurvatureMeeting/Interview
Wave packetDigital photographyAlgorithmArchaeological field surveyMathematical analysisThermal radiationSet (mathematics)MeasurementGraph coloringInformationSpectroscopyDifferent (Kate Ryan album)Shift operatorPhysical systemServer (computing)Order of magnitudeSpectrum (functional analysis)Level (video gaming)Virtual machineMathematicianoutputArtificial neural networkGaussian processTheoryDecision theoryObject (grammar)Source codePhysicalismStatisticsElement (mathematics)Binary fileRevision controlSpektrum <Mathematik>VarianceDecision tree learningMatter waveSoftwareTemplate (C++)Functional (mathematics)PixelDisk read-and-write headKey (cryptography)WaveWaveletService (economics)Meeting/Interview
Core dumpComputer animation
Meeting/InterviewComputer animation
Computational fluid dynamicsField (computer science)Meeting/InterviewComputer animation
Doppler-EffektMathematicianNear-ringSoftware testingGroup actionHypothesisMeeting/Interview
Bus (computing)MathematicsRevision controlMusical ensembleWaveSoftwareFrequencyDifferent (Kate Ryan album)Variety (linguistics)Source codeMathematical analysisFitness functionoutputTemplate (C++)VarianceShift operatorWave packetArtificial neural networkRight angleGraph coloringData analysisVirtual machineUniverse (mathematics)Water vaporSpectrum (functional analysis)PhysicalismAlgorithmMeasurementGaussian processSet (mathematics)Digital photographyNetwork topologyDecision theoryMultiplication signTheoryEvent horizonCodeResultantLogical constantComputer animationMeeting/Interview
SoftwareMedical imagingStochastic processQuicksortComputer animation
NumberMedical imagingGraph coloringMeasurementService (economics)Meeting/Interview
Server (computing)InternetworkingComputer animationMeeting/Interview
Transcript: English(auto-generated)
restart I am I am a master student I working on radio telescope project here in Brazil and my task is to reconstruct a red shift what off far
away objects and I will tell you how I do this first you had to know what is a
redshift to understand what is redshift you have to record what is the Doppler effect the Doppler effect is the changing frequency of a wave in relation to an observer who is moving relative to the wave source one example of Doppler effect shift is the change of pitch heard when a bus
sounding a horn approach and recedes from the observer compared to the input frequency they receive the frequency is higher during the approach and we it's identical when they sent it pass it's passing by us and it's
lower during the recession redshift is the light version of this phenomenon you whisper and see it all the time and we we can we can see if
you look at the sky we we know that that that there is blue the blue stars and red stars and the the red stars and are far away from us then the
bluest ones and because we have we have this in the color spectrum spectrum this frequency is this frequency is higher than this and when
we see a celestial object that there is red it means that is moving far away from us in the expanding universe this is this is this is why we this
one main reasons we stood this thing because we want to know that if the universe is expanding or if it is contracting or if it is stationary when we the rebel constant is is a number that measures how how the universe is
expanding and the rebel number is calculated measuring the the red shift
there are two main approach to to know where the red shift of a star we can do
data driving analysis is that what I do and we can do what the plate feature analysis using physics theory I don't know physics theory because I am
research I do I use that data and use machine to machine learn methods to study this phenomenon well if you use a machine learning methods we have to have a training set and wish we we privilege know where what is head what
is the right redshift and to to and we have two kinds of data a you we use our training set we use spectroscopy redshift and photos the red shift a
spectroscopy redshift you refer to the measurement of radiation intensity as a function of a wavelet and we just use it to describe the all
the wave of the spectra of the intensity of all electromagnet key and of elements of our atom of our stars we have infrared information we have all
the other of the colors information and a bunch gave gay gamma ray information and do it it's it's a complete it's a more complete analysis of often object it could be it can be an atom or it can be a star but
photos a red shift it's and so just some colors so some we can just how is the intensity of red yellow blue and infrared in the stars there are
many system there are many system of photos a red shift it's it depends on the survey and the data you you analyze analyze and for example I use
pen stars data which I have five colors and we call this colors and make magnitudes and we we compare the this photos a red shift you have with five colors without all the beans of the spectroscopy data we have in our
training set but and frequently we have many sources ministers of uncertainty and some of they are there we have a lack of input some train to
training where because we will when we do this analysis we have the map of skies and we and it is divided in pixels but it we can okay we can have
but we are comparing and spectroscopy data to and red shift to one and there there are some missing galaxies in one or another and we we have because
we had what we compare to serve it to service to to match our training set but we have some being the uncertainty of the difference
algorithms algorithms we use and we have a completed training sets because we are dealing with a very a very sensitive data that there is a there
that could be in completely and do depends on the server you are working on and here in my work I use two of the most popular astronomical a
software's well for doing I use GPZ and a NZ to and the first and is approach that use Gaussian processes to estimate a photometric redshift and we
are foxy on the on the variance of the set because our data has met has a bunch of your a horse and we have to use statistical analysis because the
because the source of the of your research is far far away we have to we don't have a good precision well wait and the second we use is a in easy to
that is the updated version of the original algorithm which use neural networks and in boosted decision and regression trees does it does anyone have a question hello hi I'm Francesca Baskin and I am just
come so I was wondering if you could resume briefly the core of your worker and I work with reconstruction of a photometric redshift do you know what
is redshift yes in general but okay if you can explain something else do you are you are you astronomer or something um no I work in the field of
computational fluid dynamics so totally another field and I'm a beginner in Python so very basic notion I will try to explain what I do because I I'm
still learning about it I am a mathematician and I were I mean working in a cosmology group and and I'm doing this in my master thesis and to
understand what a redshift is and you have to record what is the Doppler effect and the you know and when you are there is a bus and near to you and there is a there is a change of pitch pitch heard when when this bus
sounding a hard approach and received from you and compared to the empty frequency and the empty frequency of the way the wave is higher during the
approach and when and it's lower during the recession redshift is the light version of this phenomenon and we experience it and it all the time with sound you have you can see this with this where we had here blue and we
have here red the frequency is is high of the light when we are here in the blue and we as we go and the frequency of the light is lower and reaching red
so when you see and I start that is red it's bluest then water it's more near it's more near then why why we why we start this thing because we want to
know if the universe is expanding of it is contracting off if it is doing nothing we have it we have some events that a universe is expanding
and the measure of redshift allows us to measure this using a thing called rubber constant and this is one main reason we start this this thing there
are two main approach we use when we measure a redshift we can do a data analysis that is why what I do because I don't know enough physics to do a template feature analysis and then they use how healthy a physics theory
and I I in my research I use machine learning techniques and then there are many there are many a machine learning software's that we can use in for for
it but I I used specifically to source I use a GPZ they use get the Gaussian process and I use a an easy to the there is M and use neural networks
and boosted decision trees but but to do our and but to do this we have to have our training set and because of this we have the spectroscopy redshift
and our photos a redshift what is the difference between the two the difference is that photos a redshift has have has many as I'm just a five
color six colors it's colors that we actually see accepted the infrared but spectroscopy redshift is all the it's all the bands of the spectroscopy all the bands of the spectrum and it include and codes x-rays a gamma rays
and a variety of different bands of the spectrum of them the air tone it's using at all analysis and you it's used in stellar analysis so and we have
a lot of difficult when we are doing this because we have a very uncertainty and source we have frequently we have incomplete training sets we have
the uncertainty in the algorithm we use and we have a lot of uncertainty in our imputes of training and because of this we do not use a package like
psychic learning takes or fall directly we use it use there but I we use they in some way we can get its variance because we have we we we
work with some uncertainty and of the redshift because we cannot because because we have this certain source and we cannot have the exactly result and this is why we have we work with specific softwares which are described
by academic papers in my work I use this this to a GPZL and in is to do
you have some questions yeah I have no experience about the about these two softwares but I was wondering because my work is about reconstruction in medical imaging and so do you do a sort of reconstruction process my name my
data is is it's not image it's it's a it's a number are you we it's the measure intensity of some some color it's not it's not the the image of
the stars do you do you know SSD we I work with two service they are
available on the internet and I cannot find the chat but if you if you if you search you can you can find them and it's placed ten stars with P a and n
stars and s d s s