Sunday, March 25, 2018
Tuesday, March 13, 2018
OS hunt for exo-planets
https://research.googleblog.com/2018/03/open-sourcing-hunt-for-exoplanets.html
Thursday, March 08, 2018
Posted by Chris Shallue, Senior Software Engineer, Google Brain Team
(Crossposted on the Google Open Source Blog)
Recently, we discovered two exoplanets by training a neural network to analyze data from NASA’s Kepler space telescope and accurately identify the most promising planet signals. And while this was only an initial analysis of ~700 stars, we consider this a successful proof-of-concept for using machine learning to discover exoplanets, and more generally another example of using machine learning to make meaningful gains in a variety of scientific disciplines (e.g. healthcare, quantum chemistry, and fusion research).
Today, we’re excited to release our code for processing the Kepler data, training our neural network model, and making predictions about new candidate signals. We hope this release will prove a useful starting point for developing similar models for other NASA missions, like K2 (Kepler’s second mission) and the upcoming Transiting Exoplanet Survey Satellite mission. As well as announcing the release of our code, we’d also like take this opportunity to dig a bit deeper into how our model works.
A Planet Hunting Primer
First, let’s consider how data collected by the Kepler telescope is used to detect the presence of a planet. The plot below is called a light curve, and it shows the brightness of the star (as measured by Kepler’s photometer) over time. When a planet passes in front of the star, it temporarily blocks some of the light, which causes the measured brightness to decrease and then increase again shortly thereafter, causing a “U-shaped” dip in the light curve.
A light curve from the Kepler space telescope with a “U-shaped” dip that indicates a transiting exoplanet.
However, other astronomical and instrumental phenomena can also cause the measured brightness of a star to decrease, including binary star systems, starspots, cosmic ray hits on Kepler’s photometer, and instrumental noise.
The first light curve has a “V-shaped” pattern that tells us that a very large object (i.e. another star) passed in front of the star that Kepler was observing. The second light curve contains two places where the brightness decreases, which indicates a binary system with one bright and one dim star: the larger dip is caused by the dimmer star passing in front of the brighter star, and vice versa. The third light curve is one example of the many other non-planet signals where the measured brightness of a star appears to decrease.
To search for planets in Kepler data, scientists use automated software (e.g. the Kepler data processing pipeline) to detect signals that might be caused by planets, and then manually follow up to decide whether each signal is a planet or a false positive. To avoid being overwhelmed with more signals than they can manage, the scientists apply a cutoff to the automated detections: those with signal-to-noise ratios above a fixed threshold are deemed worthy of follow-up analysis, while all detections below the threshold are discarded. Even with this cutoff, the number of detections is still formidable: to date, over 30,000 detected Kepler signals have been manually examined, and about 2,500 of those have been validated as actual planets!
Perhaps you’re wondering: does the signal-to-noise cutoff cause some real planet signals to be missed? The answer is, yes! However, if astronomers need to manually follow up on every detection, it’s not really worthwhile to lower the threshold, because as the threshold decreases the rate of false positive detections increases rapidly and actual planet detections become increasingly rare. However, there’s a tantalizing incentive: it’s possible that some potentially habitable planets like Earth, which are relatively small and orbit around relatively dim stars, might be hiding just below the traditional detection threshold — there might be hidden gems still undiscovered in the Kepler data!
A Machine Learning Approach
The Google Brain team applies machine learning to a diverse variety of data, from human genomes to sketches to formal mathematical logic. Considering the massive amount of data collected by the Kepler telescope, we wondered what we might find if we used machine learning to analyze some of the previously unexplored Kepler data. To find out, we teamed up with Andrew Vanderburg at UT Austin and developed a neural network to help search the low signal-to-noise detections for planets.
We trained a convolutional neural network (CNN) to predict the probability that a given Kepler signal is caused by a planet. We chose a CNN because they have been very successful in other problems with spatial and/or temporal structure, like audio generation and image classification.
Luckily, we had 30,000 Kepler signals that had already been manually examined and classified by humans. We used a subset of around 15,000 of these signals, of which around 3,500 were verified planets or strong planet candidates, to train our neural network to distinguish planets from false positives. The inputs to our network are two separate views of the same light curve: a wide view that allows the model to examine signals elsewhere on the light curve (e.g., a secondary signal caused by a binary star), and a zoomed-in view that enables the model to closely examine the shape of the detected signal (e.g., to distinguish “U-shaped” signals from “V-shaped” signals).
Once we had trained our model, we investigated the features it learned about light curves to see if they matched with our expectations. One technique we used (originally suggested in this paper) was to systematically occlude small regions of the input light curves to see whether the model’s output changed. Regions that are particularly important to the model’s decision will change the output prediction if they are occluded, but occluding unimportant regions will not have a significant effect. Below is a light curve from a binary star that our model correctly predicts is not a planet. The points highlighted in green are the points that most change the model’s output prediction when occluded, and they correspond exactly to the secondary “dip” indicative of a binary system. When those points are occluded, the model’s output prediction changes from ~0% probability of being a planet to ~40% probability of being a planet. So, those points are part of the reason the model rejects this light curve, but the model uses other evidence as well - for example, zooming in on the centred primary dip shows that it's actually “V-shaped”, which is also indicative of a binary system.
Searching for New Planets
Once we were confident with our model’s predictions, we tested its effectiveness by searching for new planets in a small set 670 stars. We chose these stars because they were already known to have multiple orbiting planets, and we believed that some of these stars might host additional planets that had not yet been detected. Importantly, we allowed our search to include signals that were below the signal-to-noise threshold that astronomers had previously considered. As expected, our neural network rejected most of these signals as spurious detections, but a handful of promising candidates rose to the top, including our two newly discovered planets: Kepler-90 i and Kepler-80 g.
Find your own Planet(s)!
Let’s take a look at how the code released today can help (re-)discover the planet Kepler-90 i. The first step is to train a model by following the instructions on the code’s home page. It takes a while to download and process the data from the Kepler telescope, but once that’s done, it’s relatively fast to train a model and make predictions about new signals. One way to find new signals to show the model is to use an algorithm called Box Least Squares (BLS), which searches for periodic “box shaped” dips in brightness (see below). The BLS algorithm will detect “U-shaped” planet signals, “V-shaped” binary star signals and many other types of false positive signals to show the model. There are various freely available software implementations of the BLS algorithm, including VARTOOLS and LcTools. Alternatively, you can even look for candidate planet transits by eye, like the Planet Hunters.
A low signal-to-noise detection in the light curve of the Kepler 90 star detected by the BLS algorithm. The detection has period 14.44912 days, duration 2.70408 hours (0.11267 days) beginning 2.2 days after 12:00 on 1/1/2009 (the year the Kepler telescope launched).
To run this detected signal though our trained model, we simply execute the following command:
python predict.py --kepler_id=11442793 --period=14.44912 --t0=2.2
--duration=0.11267 --kepler_data_dir=$HOME/astronet/kepler
--output_image_file=$HOME/astronet/kepler-90i.png
--model_dir=$HOME/astronet/model The output of the command is prediction = 0.94, which means the model is 94% certain that this signal is a real planet. Of course, this is only a small step in the overall process of discovering and validating an exoplanet: the model’s prediction is not proof one way or the other. The process of validating this signal as a real exoplanet requires significant follow-up work by an expert astronomer — see Sections 6.3 and 6.4 of our paper for the full details. In this particular case, our follow-up analysis validated this signal as a bona fide exoplanet, and it’s now called Kepler-90 i!
Our work here is far from done. We’ve only searched 670 stars out of 200,000 observed by Kepler — who knows what we might find when we turn our technique to the entire dataset. Before we do that, though, we have a few improvements we want to make to our model. As we discussed in our paper, our model is not yet as good at rejecting binary stars and instrumental false positives as some more mature computer heuristics. We’re hard at work improving our model, and now that it’s open sourced, we hope others will do the same!
If you’d like to learn more, Chris is featured on the latest episode of This Week In Machine Learning & AI discussing his work.
Thursday, March 08, 2018
Posted by Chris Shallue, Senior Software Engineer, Google Brain Team
(Crossposted on the Google Open Source Blog)
Recently, we discovered two exoplanets by training a neural network to analyze data from NASA’s Kepler space telescope and accurately identify the most promising planet signals. And while this was only an initial analysis of ~700 stars, we consider this a successful proof-of-concept for using machine learning to discover exoplanets, and more generally another example of using machine learning to make meaningful gains in a variety of scientific disciplines (e.g. healthcare, quantum chemistry, and fusion research).
Today, we’re excited to release our code for processing the Kepler data, training our neural network model, and making predictions about new candidate signals. We hope this release will prove a useful starting point for developing similar models for other NASA missions, like K2 (Kepler’s second mission) and the upcoming Transiting Exoplanet Survey Satellite mission. As well as announcing the release of our code, we’d also like take this opportunity to dig a bit deeper into how our model works.
A Planet Hunting Primer
First, let’s consider how data collected by the Kepler telescope is used to detect the presence of a planet. The plot below is called a light curve, and it shows the brightness of the star (as measured by Kepler’s photometer) over time. When a planet passes in front of the star, it temporarily blocks some of the light, which causes the measured brightness to decrease and then increase again shortly thereafter, causing a “U-shaped” dip in the light curve.
A light curve from the Kepler space telescope with a “U-shaped” dip that indicates a transiting exoplanet.
However, other astronomical and instrumental phenomena can also cause the measured brightness of a star to decrease, including binary star systems, starspots, cosmic ray hits on Kepler’s photometer, and instrumental noise.
The first light curve has a “V-shaped” pattern that tells us that a very large object (i.e. another star) passed in front of the star that Kepler was observing. The second light curve contains two places where the brightness decreases, which indicates a binary system with one bright and one dim star: the larger dip is caused by the dimmer star passing in front of the brighter star, and vice versa. The third light curve is one example of the many other non-planet signals where the measured brightness of a star appears to decrease.
To search for planets in Kepler data, scientists use automated software (e.g. the Kepler data processing pipeline) to detect signals that might be caused by planets, and then manually follow up to decide whether each signal is a planet or a false positive. To avoid being overwhelmed with more signals than they can manage, the scientists apply a cutoff to the automated detections: those with signal-to-noise ratios above a fixed threshold are deemed worthy of follow-up analysis, while all detections below the threshold are discarded. Even with this cutoff, the number of detections is still formidable: to date, over 30,000 detected Kepler signals have been manually examined, and about 2,500 of those have been validated as actual planets!
Perhaps you’re wondering: does the signal-to-noise cutoff cause some real planet signals to be missed? The answer is, yes! However, if astronomers need to manually follow up on every detection, it’s not really worthwhile to lower the threshold, because as the threshold decreases the rate of false positive detections increases rapidly and actual planet detections become increasingly rare. However, there’s a tantalizing incentive: it’s possible that some potentially habitable planets like Earth, which are relatively small and orbit around relatively dim stars, might be hiding just below the traditional detection threshold — there might be hidden gems still undiscovered in the Kepler data!
A Machine Learning Approach
The Google Brain team applies machine learning to a diverse variety of data, from human genomes to sketches to formal mathematical logic. Considering the massive amount of data collected by the Kepler telescope, we wondered what we might find if we used machine learning to analyze some of the previously unexplored Kepler data. To find out, we teamed up with Andrew Vanderburg at UT Austin and developed a neural network to help search the low signal-to-noise detections for planets.
We trained a convolutional neural network (CNN) to predict the probability that a given Kepler signal is caused by a planet. We chose a CNN because they have been very successful in other problems with spatial and/or temporal structure, like audio generation and image classification.
Luckily, we had 30,000 Kepler signals that had already been manually examined and classified by humans. We used a subset of around 15,000 of these signals, of which around 3,500 were verified planets or strong planet candidates, to train our neural network to distinguish planets from false positives. The inputs to our network are two separate views of the same light curve: a wide view that allows the model to examine signals elsewhere on the light curve (e.g., a secondary signal caused by a binary star), and a zoomed-in view that enables the model to closely examine the shape of the detected signal (e.g., to distinguish “U-shaped” signals from “V-shaped” signals).
Once we had trained our model, we investigated the features it learned about light curves to see if they matched with our expectations. One technique we used (originally suggested in this paper) was to systematically occlude small regions of the input light curves to see whether the model’s output changed. Regions that are particularly important to the model’s decision will change the output prediction if they are occluded, but occluding unimportant regions will not have a significant effect. Below is a light curve from a binary star that our model correctly predicts is not a planet. The points highlighted in green are the points that most change the model’s output prediction when occluded, and they correspond exactly to the secondary “dip” indicative of a binary system. When those points are occluded, the model’s output prediction changes from ~0% probability of being a planet to ~40% probability of being a planet. So, those points are part of the reason the model rejects this light curve, but the model uses other evidence as well - for example, zooming in on the centred primary dip shows that it's actually “V-shaped”, which is also indicative of a binary system.
Searching for New Planets
Once we were confident with our model’s predictions, we tested its effectiveness by searching for new planets in a small set 670 stars. We chose these stars because they were already known to have multiple orbiting planets, and we believed that some of these stars might host additional planets that had not yet been detected. Importantly, we allowed our search to include signals that were below the signal-to-noise threshold that astronomers had previously considered. As expected, our neural network rejected most of these signals as spurious detections, but a handful of promising candidates rose to the top, including our two newly discovered planets: Kepler-90 i and Kepler-80 g.
Find your own Planet(s)!
Let’s take a look at how the code released today can help (re-)discover the planet Kepler-90 i. The first step is to train a model by following the instructions on the code’s home page. It takes a while to download and process the data from the Kepler telescope, but once that’s done, it’s relatively fast to train a model and make predictions about new signals. One way to find new signals to show the model is to use an algorithm called Box Least Squares (BLS), which searches for periodic “box shaped” dips in brightness (see below). The BLS algorithm will detect “U-shaped” planet signals, “V-shaped” binary star signals and many other types of false positive signals to show the model. There are various freely available software implementations of the BLS algorithm, including VARTOOLS and LcTools. Alternatively, you can even look for candidate planet transits by eye, like the Planet Hunters.
A low signal-to-noise detection in the light curve of the Kepler 90 star detected by the BLS algorithm. The detection has period 14.44912 days, duration 2.70408 hours (0.11267 days) beginning 2.2 days after 12:00 on 1/1/2009 (the year the Kepler telescope launched).
To run this detected signal though our trained model, we simply execute the following command:
python predict.py --kepler_id=11442793 --period=14.44912 --t0=2.2
--duration=0.11267 --kepler_data_dir=$HOME/astronet/kepler
--output_image_file=$HOME/astronet/kepler-90i.png
--model_dir=$HOME/astronet/model The output of the command is prediction = 0.94, which means the model is 94% certain that this signal is a real planet. Of course, this is only a small step in the overall process of discovering and validating an exoplanet: the model’s prediction is not proof one way or the other. The process of validating this signal as a real exoplanet requires significant follow-up work by an expert astronomer — see Sections 6.3 and 6.4 of our paper for the full details. In this particular case, our follow-up analysis validated this signal as a bona fide exoplanet, and it’s now called Kepler-90 i!
Our work here is far from done. We’ve only searched 670 stars out of 200,000 observed by Kepler — who knows what we might find when we turn our technique to the entire dataset. Before we do that, though, we have a few improvements we want to make to our model. As we discussed in our paper, our model is not yet as good at rejecting binary stars and instrumental false positives as some more mature computer heuristics. We’re hard at work improving our model, and now that it’s open sourced, we hope others will do the same!
If you’d like to learn more, Chris is featured on the latest episode of This Week In Machine Learning & AI discussing his work.
Tuesday, February 27, 2018
'It's irrational to deny people access to LSD'
https://www.radionz.co.nz/national/programmes/sunday/audio/2018633601/prof-david-nutt-it-s-irrational-to-deny-people-access-to-lsd
Prof David Nutt: 'It's irrational to deny people access to LSD'
From Sunday Morning, 10:04 am on 25 February 2018
Outspoken UK drug expert David Nutt argues for regulated access to any drug less harmful than alcohol, including cannabis and Ecstasy.
He is a professor of neuropsychopharmacology at Imperial College London and author of Drugs: Without The Hot Air (2012).
Professor David Nutt: 'Any drug which is less harmful than alcohol ... should be available as an alternative to alcohol.'Prof David Nutt Photo: Supplied
“My view is that any drug that is less harmful to the person that uses it than alcohol should be available as an alternative to alcohol - in a form where the harms are minimised,” he told Sunday Morning.
“That doesn’t mean a free market, that doesn’t mean opening up cannabis shops to compete with supermarkets selling beer.
“It means having access to cannabis, and ecstasy and other drugs perhaps, such as mephedrone and mushrooms - in pharmacies, possibly with electronic cards that allow you to have a certain amount per year.”
Listen to the whole conversation with David Nutt duration 35′ :29″ Add to playlist Download
Download as Ogg
Download as MP3
Play Ogg in browser
Play MP3 in browser
Listen to the whole conversation with David Nutt
Prof Nutt said in the UK the leading causes of drug deaths were tobacco, alcohol and heroin and opiates – all much higher than deaths from ecstasy (MDMA).
“Regulated access to drugs that are less harmful than alcohol will reduce the harms of alcohol, hopefully reduce the use of drugs more harmful than alcohol like crystal meth and P and heroin, and will actually overall reduce the societal damage from drugs.”
Research on certain drugs and their potential therapeutic value was being denied, he said.
He has worked on a study of what happens in the brains of people under the influence of LSD and psilocybin (magic mushroom) using brain imaging scans.
They found “to our surprise” that the part of the brain that tends to drive depression was switched off by psilocybin, in a similar way to anti-depressants, and a follow up study suggested the effect could last for several weeks or months.
He argues psilocybin and similar drugs should be moved from Schedule one to Schedule two - still a class of controlled drugs alongside the likes of morphine - “so that your doctors can research them very easily and determine their therapeutic value in the New Zealand population".
He says psilocybin could also be used in treatment of cancer patients and terminal illness.
"There have been two studies done ... and one of the most promising uses is in helping people come to terms with dying. Often under psychedelics people become 'at one' with the universe. I'm very interested in that, in helping people deal with that terminal, difficult part of their lives. These two American studies showed that a single psilocybin treatment helped people become less anxious about dying."
He says opiates are not the be-all and end-all in terminal pain treatment.
"Under opiates you're constipated, you're confused, you have a dry mouth - you might be free of pain but it's certainly not necessarily a good death.
"And using opiates to deal with chronic pain syndrome you can also get excessive use and people selling them on the streets so I think psychedelics offer an alternative, not just to the pain of dying but also to treat pain syndromes.
"If you allow people to take morphine for pain control, which kills in the UK nearly 2000 people a year, it's irrational to deny people access to LSD which kills nobody."
He says there are huge national variations in the types of misused and dangerous drugs, but worldwide, the big problem was going to be the synthetic opioid fentanyl, he says, because the black market has now worked out it’s much cheaper to make than heroin.
Some fentanyl is 10,000 times more potent than heroin so the tiniest drop can cause an overdose.
“It’s now become clear about a third of the opiate deaths in the United States are due to fentanyl. That is going to sweep the world”
Prof David NuttProf David Nutt Photo: Supplied
Prof Nutt was chair of the UK government’s advisory committee on the misuse of drugs until 2009, when he was sacked for suggesting alcohol was more dangerous than many illegal drugs.
“If we care about the harms of drugs to the point where people are dying then you’ve got to do something about alcohol.
“This is why I feel passionately about the position of science in this debate.”
Though alcohol was a leading cause of death under the age of 50 in the UK, politicians did not want to take on the alcohol industry, he says, so they make a show of being hard on drugs.
“I believe in Britain one of the reasons we have this massive problem of early death from alcohol is because we allow our kids to start drinking at the age of about 14.
“I frankly am not encouraging anyone under the age of 18 to use any drug – the longer you can wait before using any drug the less likely you are to interfere with your education or your long-term health.
“Certainly under-age cannabis use would not be good, and I encourage people not to do it. But that doesn’t mean we should ban it from people who are fit and over the age of 18."
Prof Nutt is giving a public lecture titled ‘How an enlightened approach to drug policy could revolutionise medicine as well as reducing drug harms’ at the University of Auckland on 26 February.
Where to get help:
Alcohol Drug Helpline 0800 787 797 or text 8681 – confidential advice, information or support about drinking or other drug use.
Quitline 0800 778 778 – advisors provide support for people wanting to quit smoking
Prof David Nutt: 'It's irrational to deny people access to LSD'
From Sunday Morning, 10:04 am on 25 February 2018
Outspoken UK drug expert David Nutt argues for regulated access to any drug less harmful than alcohol, including cannabis and Ecstasy.
He is a professor of neuropsychopharmacology at Imperial College London and author of Drugs: Without The Hot Air (2012).
Professor David Nutt: 'Any drug which is less harmful than alcohol ... should be available as an alternative to alcohol.'Prof David Nutt Photo: Supplied
“My view is that any drug that is less harmful to the person that uses it than alcohol should be available as an alternative to alcohol - in a form where the harms are minimised,” he told Sunday Morning.
“That doesn’t mean a free market, that doesn’t mean opening up cannabis shops to compete with supermarkets selling beer.
“It means having access to cannabis, and ecstasy and other drugs perhaps, such as mephedrone and mushrooms - in pharmacies, possibly with electronic cards that allow you to have a certain amount per year.”
Listen to the whole conversation with David Nutt duration 35′ :29″ Add to playlist Download
Download as Ogg
Download as MP3
Play Ogg in browser
Play MP3 in browser
Listen to the whole conversation with David Nutt
Prof Nutt said in the UK the leading causes of drug deaths were tobacco, alcohol and heroin and opiates – all much higher than deaths from ecstasy (MDMA).
“Regulated access to drugs that are less harmful than alcohol will reduce the harms of alcohol, hopefully reduce the use of drugs more harmful than alcohol like crystal meth and P and heroin, and will actually overall reduce the societal damage from drugs.”
Research on certain drugs and their potential therapeutic value was being denied, he said.
He has worked on a study of what happens in the brains of people under the influence of LSD and psilocybin (magic mushroom) using brain imaging scans.
They found “to our surprise” that the part of the brain that tends to drive depression was switched off by psilocybin, in a similar way to anti-depressants, and a follow up study suggested the effect could last for several weeks or months.
He argues psilocybin and similar drugs should be moved from Schedule one to Schedule two - still a class of controlled drugs alongside the likes of morphine - “so that your doctors can research them very easily and determine their therapeutic value in the New Zealand population".
He says psilocybin could also be used in treatment of cancer patients and terminal illness.
"There have been two studies done ... and one of the most promising uses is in helping people come to terms with dying. Often under psychedelics people become 'at one' with the universe. I'm very interested in that, in helping people deal with that terminal, difficult part of their lives. These two American studies showed that a single psilocybin treatment helped people become less anxious about dying."
He says opiates are not the be-all and end-all in terminal pain treatment.
"Under opiates you're constipated, you're confused, you have a dry mouth - you might be free of pain but it's certainly not necessarily a good death.
"And using opiates to deal with chronic pain syndrome you can also get excessive use and people selling them on the streets so I think psychedelics offer an alternative, not just to the pain of dying but also to treat pain syndromes.
"If you allow people to take morphine for pain control, which kills in the UK nearly 2000 people a year, it's irrational to deny people access to LSD which kills nobody."
He says there are huge national variations in the types of misused and dangerous drugs, but worldwide, the big problem was going to be the synthetic opioid fentanyl, he says, because the black market has now worked out it’s much cheaper to make than heroin.
Some fentanyl is 10,000 times more potent than heroin so the tiniest drop can cause an overdose.
“It’s now become clear about a third of the opiate deaths in the United States are due to fentanyl. That is going to sweep the world”
Prof David NuttProf David Nutt Photo: Supplied
Prof Nutt was chair of the UK government’s advisory committee on the misuse of drugs until 2009, when he was sacked for suggesting alcohol was more dangerous than many illegal drugs.
“If we care about the harms of drugs to the point where people are dying then you’ve got to do something about alcohol.
“This is why I feel passionately about the position of science in this debate.”
Though alcohol was a leading cause of death under the age of 50 in the UK, politicians did not want to take on the alcohol industry, he says, so they make a show of being hard on drugs.
“I believe in Britain one of the reasons we have this massive problem of early death from alcohol is because we allow our kids to start drinking at the age of about 14.
“I frankly am not encouraging anyone under the age of 18 to use any drug – the longer you can wait before using any drug the less likely you are to interfere with your education or your long-term health.
“Certainly under-age cannabis use would not be good, and I encourage people not to do it. But that doesn’t mean we should ban it from people who are fit and over the age of 18."
Prof Nutt is giving a public lecture titled ‘How an enlightened approach to drug policy could revolutionise medicine as well as reducing drug harms’ at the University of Auckland on 26 February.
Where to get help:
Alcohol Drug Helpline 0800 787 797 or text 8681 – confidential advice, information or support about drinking or other drug use.
Quitline 0800 778 778 – advisors provide support for people wanting to quit smoking
Saturday, February 24, 2018
Sunday, February 18, 2018
Sunday, February 11, 2018
Subscribe to:
Posts (Atom)