data_note.pdf has the following statement \"units for both flux and flux err are arbitrary\".\nDoes it mean that comparing flux and flux err or doing arithmetical operations on them make no physical sense?
", "messages": [ { "id": "404322", "postDate": "10/15/2018 15:04:10", "content": "data_note.pdf has the following statement \"units for both flux and flux err are arbitrary\".\nDoes it mean that comparing flux and flux err or doing arithmetical operations on them make no physical sense?
", "rawMarkdown": "data_note.pdf has the following statement \"units for both flux and flux err are arbitrary\".\nDoes it mean that comparing flux and flux err or doing arithmetical operations on them make no physical sense?", "votes": null }, { "id": "404337", "postDate": "10/15/2018 15:28:10", "content": "No - logical and arithmetic operations on flux and flux_err are perfectly well defined. That statement means that the units of the quantities are not easily convertible to something like a brightness in terms of quantities like Watt/m^2/Hz (or because our sources are so faint, in Janskys, which is 10^-26 W/m^2/Hz). This is because the flux is really the difference of flux with respect to a template - this actually helps makes objects comparable to each other. Supernovae, for instance, occur inside all sorts of galaxies at different redshifts. If every supernova in the sample had a different background level, that'd make comparing objects much harder.
Note that this only corrects for the different background we measure the source against, and we've also applied a correction for the differing amount of dust in the Milky Way along different lines of sight (also documented in the data note, and starter kit). We have NOT corrected for their different redshifts - doing that would require knowing what the class of each source is, which is the goal of this challenge!
\n\nBest,\n-Gautham for the PLAsTiCC Team
", "rawMarkdown": "No - logical and arithmetic operations on `flux` and `flux_err` are perfectly well defined. That statement means that the units of the quantities are not easily convertible to something like a brightness in terms of quantities like Watt/m^2/Hz (or because our sources are so faint, in Janskys, which is 10^-26 W/m^2/Hz). This is because the flux is really the difference of flux with respect to a template - this actually helps makes objects comparable to each other. Supernovae, for instance, occur inside all sorts of galaxies at different redshifts. If every supernova in the sample had a different background level, that'd make comparing objects much harder. \n\nNote that this only corrects for the different background we measure the source against, and we've also applied a correction for the differing amount of dust in the Milky Way along different lines of sight (also documented in the data note, and starter kit). We have NOT corrected for their different redshifts - doing that would require knowing what the class of each source is, which is the goal of this challenge! \n\nBest,\n-Gautham for the PLAsTiCC Team", "votes": null }, { "id": "404351", "postDate": "10/15/2018 15:55:50", "content": "Gautham, thanks for fast reply.
", "rawMarkdown": "Gautham, thanks for fast reply.", "votes": null }, { "id": "404525", "postDate": "10/15/2018 22:20:59", "content": "i mean since there is already a thread I figured I'd ask, and if this sounds like a dumb question i apologize. so is flux_err a measured error via some physical phenomena or is it a statistical error that has been calculated via deviations and such. Specifically are they sampling a bunch getting an average and that's variation or standard deviation. or is this a case of \"known tolerances given a particular environment\" . you handle the features differently based on that .. statistical averages vs actual measurements.
", "rawMarkdown": "i mean since there is already a thread I figured I'd ask, and if this sounds like a dumb question i apologize. so is flux_err a measured error via some physical phenomena or is it a statistical error that has been calculated via deviations and such. Specifically are they sampling a bunch getting an average and that's variation or standard deviation. or is this a case of \"known tolerances given a particular environment\" . you handle the features differently based on that .. statistical averages vs actual measurements.", "votes": null }, { "id": "405213", "postDate": "10/17/2018 04:17:52", "content": "never mind i found my answer in https://www.kaggle.com/michaelapers/the-plasticc-astronomy-starter-kit says its 68% confidence which means it just an normal standard deviation not a measure of mechanical error.
", "rawMarkdown": "never mind i found my answer in https://www.kaggle.com/michaelapers/the-plasticc-astronomy-starter-kit says its 68% confidence which means it just an normal standard deviation not a measure of mechanical error.", "votes": null } ], "comments": [ { "id": 404337, "author_name": "gsnarayan", "author_url": "", "post_date": "10/15/2018 15:28:10", "content": "No - logical and arithmetic operations on flux and flux_err are perfectly well defined. That statement means that the units of the quantities are not easily convertible to something like a brightness in terms of quantities like Watt/m^2/Hz (or because our sources are so faint, in Janskys, which is 10^-26 W/m^2/Hz). This is because the flux is really the difference of flux with respect to a template - this actually helps makes objects comparable to each other. Supernovae, for instance, occur inside all sorts of galaxies at different redshifts. If every supernova in the sample had a different background level, that'd make comparing objects much harder.
Note that this only corrects for the different background we measure the source against, and we've also applied a correction for the differing amount of dust in the Milky Way along different lines of sight (also documented in the data note, and starter kit). We have NOT corrected for their different redshifts - doing that would require knowing what the class of each source is, which is the goal of this challenge!
\n\nBest,\n-Gautham for the PLAsTiCC Team
", "votes": null, "replies": [] }, { "id": 404351, "author_name": "alexfir", "author_url": "", "post_date": "10/15/2018 15:55:50", "content": "Gautham, thanks for fast reply.
", "votes": null, "replies": [] }, { "id": 404525, "author_name": "jscheibel", "author_url": "", "post_date": "10/15/2018 22:20:59", "content": "i mean since there is already a thread I figured I'd ask, and if this sounds like a dumb question i apologize. so is flux_err a measured error via some physical phenomena or is it a statistical error that has been calculated via deviations and such. Specifically are they sampling a bunch getting an average and that's variation or standard deviation. or is this a case of \"known tolerances given a particular environment\" . you handle the features differently based on that .. statistical averages vs actual measurements.
", "votes": null, "replies": [ { "id": 405213, "author_name": "jscheibel", "author_url": "", "post_date": "10/17/2018 04:17:52", "content": "never mind i found my answer in https://www.kaggle.com/michaelapers/the-plasticc-astronomy-starter-kit says its 68% confidence which means it just an normal standard deviation not a measure of mechanical error.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "404322": "data_note.pdf has the following statement \"units for both flux and flux err are arbitrary\".\nDoes it mean that comparing flux and flux err or doing arithmetical operations on them make no physical sense?", "404337": "No - logical and arithmetic operations on `flux` and `flux_err` are perfectly well defined. That statement means that the units of the quantities are not easily convertible to something like a brightness in terms of quantities like Watt/m^2/Hz (or because our sources are so faint, in Janskys, which is 10^-26 W/m^2/Hz). This is because the flux is really the difference of flux with respect to a template - this actually helps makes objects comparable to each other. Supernovae, for instance, occur inside all sorts of galaxies at different redshifts. If every supernova in the sample had a different background level, that'd make comparing objects much harder. \n\nNote that this only corrects for the different background we measure the source against, and we've also applied a correction for the differing amount of dust in the Milky Way along different lines of sight (also documented in the data note, and starter kit). We have NOT corrected for their different redshifts - doing that would require knowing what the class of each source is, which is the goal of this challenge! \n\nBest,\n-Gautham for the PLAsTiCC Team", "404351": "Gautham, thanks for fast reply.", "404525": "i mean since there is already a thread I figured I'd ask, and if this sounds like a dumb question i apologize. so is flux_err a measured error via some physical phenomena or is it a statistical error that has been calculated via deviations and such. Specifically are they sampling a bunch getting an average and that's variation or standard deviation. or is this a case of \"known tolerances given a particular environment\" . you handle the features differently based on that .. statistical averages vs actual measurements.", "405213": "never mind i found my answer in https://www.kaggle.com/michaelapers/the-plasticc-astronomy-starter-kit says its 68% confidence which means it just an normal standard deviation not a measure of mechanical error." }, "source": "meta" }