Monday, August 20, 2018

For posterity...

Ok, so this blog entry is more for me than you, so feel free to skip over the boring bits.  To make it easier, I'll make the non-boring bits red.

So the purpose of today's entry is to tell the story of my PhD.

It all started long ago (2009), in a room much like this one.  Actually, it might have been this one, or that one (you can't see me pointing).

A friend of mine, I think I'll call her Shally Smolderazzo to protect her identity, contacted me to let me know about an online astronomy course that was at Masters level and run out of Swinburne.  I'd been interested in astronomy for some time, and had even purchased my own 6" Newtonian telescope.  So I decided to look up Swinburne Astronomy Online.

Sure enough, it was possible to get a Masters degree (by coursework) entirely online.  I had two Bachelors degrees through the University of Melbourne, and I now worked at the University, but it is always good to have a higher degree if you want to work at a university.  So after discussing it with DW, I signed up for the first of 12 units required for the Masters (2010).

I was not a good student when I was at Uni previously.  I was easily distracted and lacked commitment.  I scraped though my degrees, but I did the bare minimum.  I had decided being a student was not for me.  Fast forward some 15 years, and I was ready to get back to study.  I was eager, I was focussed, I was older and wiser.

My first unit was fantastic.  Not only did I learn a lot of basic astronomy, I started learning how to actually do research.  I dived into the work and engaged with the online community, a bunch of like-minded people with the same fascination with astronomy that I had.  We were all paying for this course, so everyone was invested, and that makes for a lively class.  While I never met any of my classmates in person, it was great to connect, albeit briefly, with people like me.

I received a high distinction for that first unit, and the second unit as well.  For my third and fourth unit, I just missed out on high distinctions, but I wasn't too worried about the marks anyway.  I was enjoying the study and I was happy to plod along doing the work.

And then I met Chris, who would eventually become my principal research supervisor.  He came to Melbourne University to discuss our Tiled Display Wall, called an Optiportal.  We named ours OziPortal.  Chis was a researcher at Swinburne and was interested in setting something similar up at Swinburne.  He mentioned that he an I had met some years earlier, but I still don't remember it.

Chris recognised my name as one of the Swinburne Astronomy Online students, and asked me how I was enjoying the course.  We chatted about it for a while, and he mentioned that the work I was doing with the Tiled Display Wall at Melbourne Uni could be the basis for a research project in the Masters course.  To complete the Masters program, each student has to complete a research unit.  Chris offered to act as supervisor for my research unit, if I wanted him to do so.  After some discussion, Chris suggested that it was probably worth more than just a unit in the coursework program, and that maybe I should consider converting to a research degree.

Meanwhile, a former colleague from Melbourne Uni, Leon, was now IT Director at Swinburne.  I ran into him at a conference, and discussed this with him.  He agreed with Chris, that a research degree was probably a better option, and he agreed to support my application.

At that time, I had completed the first four units of the coursework program, and was able to withdraw from the program with a Graduate Certificate in Astronomy, and at the same time enrol in the Masters by Research.  We had discussed whether a Masters or PhD was the better option, and Chris advised that we should start with the Masters, and convert to PhD if we decided that there was enough in it.

It actually took nearly 8 months to get my enrolment through, due in part to my poor results from my undergraduate days, and also in part to the somewhat flawed processes at Swinburne.  I also had two supervisors from Melbourne Uni, Richard and Steven (2012).

However, finally I was a student again, or to be politically correct, a graduate researcher.  I started work on my first paper with Chris, which was for a conference called PacificVis.  It was soundly rejected.  It wasn't because it was bad, it was just poorly targeted.  So we rewrote it for a local conference and it was accepted.  It was a collaboration with all my supervisors and some colleagues.  It was a position paper, meaning it described the state of the art, and didn't contain any original research, but it was still pretty exciting to present my first peer-reviewed work at a conference (THETA 2013).

But it was time to get my hands dirty with some real research.  Chris and I designed an experiment to test the notion that Tiled Display Walls were actually needed, that people did actually perform better using these expensive displays when compared with standard displays.  This was the reason many universities were building them at that time, so we wanted to find a way to confirm or deny that hypothesis.  We tested many people, astronomers and non-astronomers and published the results in PASA (Publications of the Astronomical Society of Australia), a well respected journal in astronomy (2014).

This was the first real feather in my cap.  I had been through a thorough peer-review process for an impact-rated journal, and this would be the centre-piece of my thesis.  During the writing of the paper, I had my confirmation of candidature and presented my early results to the astro community at Swinburne.  The review panel were very enthusiastic, and suggested I consider converting the research to a PhD.  I thanked them for the advice, but had decided that while I did want to do a PhD eventually, I wanted it to be more astronomy related than technology driven.  I remember them clearly saying it could have more astronomy in it, but I still declined politely.

So as soon as the paper was published, I started work on the Masters thesis, with the literature review the next major piece of work.  I read many papers, summarised them and put together what I thought was a pretty decent lit review.  Chris thought it was excellent, and he said it was better than most PhD lit reviews he had read.  He also asked me to advise on an astronomy observing campaign called Deeper, Wider, Faster (DWF) that was underway and needed some visualisation guidance.  After meeting with the principal investigators of that project, Jeff and Igor, and on the back of Chris' comments about my lit review.  I started thinking that maybe I should convert to a PhD.

Apparently I had left it two late to convert, as I was actually nearing the submission stage for the Masters, but Chris decided to push for the conversion, and secured the support of the review panel after I wrote a substantial conversion plan.  Swinburne HDR umm'd and ahh'd for a bit and then agreed.  I was now a PhD candidate (2015).

Three opportunities for papers arose almost simultaneously.  The first was the DWF campaign, which I started work on with Jeff and Igor.  The second was another conference, where I presented a summarised and slightly extended version of my journal paper (THETA 2015).  And the third was a program set up by Amazon Web Services (AWS) to investigate the use of AWS to support SKA research.  Chris and I submitted a proposal to look at the potential of GPU-enabled remote desktops, which would allow the SKA data to be computed on in situ in the data centres, rather than downloaded to local desktops.  We were successful with the grant, but it was very slow to get the prototype desktop working.

Meanwhile, at the end of the 2015, the first DWF campaign ran, just before Christmas.  It was hot in the room despite the air conditioner, because we had around 15 astronomers, and 20 or so computers, with 24 high-res displays and two HD projectors all working at the same time.  I remember when the first on-sky session was just about to start and the TDW I was responsible for operating froze.  Jeff had organised for a documentary crew to record the event, and I had put up an amazing image on the display to impress them.  They were taking photos just as the first telescopes where about to come on-sky, when I noticed that my head node had stopped responding.  My heart stopped.

But, as luck would have it, something went wrong with the processing pipeline, which meant that the images capture by the primary remote observatory could not be processed by the local computers, which meant I had no new content to display.  So while everyone crowded around the processing team, and ignoring me, I desperately tried to fix my TDW without anyone knowing.

I rebooted the head node, which I knew would not disrupt the display on the TDW, so it would still be showing the great big nebula I'd put up earlier.  People were still taking selfies in front of it.  But the head node didn't come back up properly.  It wouldn't reboot.  Now I was stuck.  But a stroke of inspiration hit me.  Maybe I could use a cloud node.

I had cloud resources set up for the AWS project, so I created a new virtual machine (VM) in about 2 minutes and started installing the software I needed.  This took maybe 10 minutes to complete, and then I configured the new VM head node to act as the head node of the TDW.  Now this new machine had a different address to previous head node, so the TDW nodes needed to be told about it.  So I had to reboot them and let them connect to the new head node.  I remote logged in to each one and changed the details so they would look at the new head node and then rebooted them all with a remote boot command.  The TDW shutdown and restarted as I held my breath.  Everyone looked around when that happened because the room suddenly go darker when the displays turned off briefly.  Jeff asked me what was happening and I said it just needed a reboot.  He nodded and went back to focussing on the pipeline problems.

One by one, the TDW nodes came up and connected to the new head node.  Slightly less than 20 minutes had passed when the pipeline processing team fixed their problem and everyone returned to their positions in front of the TDW, ready for the images to inspect.  Each run was 20 minutes, and we had lost the first one, but the processing pipeline team were ready for the second one.  As soon as the data hit, the pipeline kicked into action and images came pouring in.  And the were directed to the new VM, not that anyone new it was new, or a VM.  I simply gave the team a new IP address to send to.

And bang.

The first image popped up on the TDW, right were it was supposed to.  Then the second, then the third, and so on.  Astronomers were rushing in to look at them, eager to find some speck of significance.  The room was abuzz, people calling out, and rushing from image to image.  I finally exhaled, eight minutes later.  People assumed I was seating so much because it was hot.  No one knew, until they were reviewing the paper about to be submitted, what had happened on day one of the DWF campaign.

The start of 2016 saw the development of the second journal paper of my research.  However, there were 15 authors on that paper, with 5 principal authors.  It took months to get it ready, and in fact, the second DWF run which took place in the middle of 2016 had occurred before I was ready to submit.  I didn't include that run in my paper and submitted it as it was.  At the same time, I submitted a short paper (5 pages) to the Astronomical Data and Simulation Software Conference in Italy, which was accepted.  Not only that, I was lucky enough for the conference to cover the costs of the conference and accommodation, and work covered the flights.  So I was able to present my work at an international conference.  As it happened, my journal paper was rejected two days after I delivered the conference paper in Italy.  I also delivered my mid-candidature review around this time.

The rejection wasn't too bad, it was simply calling for major revisions.  In fact, the simple revision was to include the second run in the results.  These revisions, which took until early 2017 to complete, were accepted in the second attempt, again to PASA.

So now my attention returned to the AWS project.  AWS weren't able to help me get the prototype desktop working, but some very clever people at my work were able, so I now had a prototype.  Chris and I again designed an experiment to test how well a remote desktop with GPU acceleration could work for astronomers.  This paper produced some excellent results, and Chris was eager to publish it in a Tier 1 journal, namely the Monthly Notices of the Royal Astronomical Society.  He acknowledged that the likelihood of acceptance was low, but tried anyway.  And the reviewers at the MNRAS suggested that we try another journal.  So we aimed at the journal Astronomy and Computing, which was the most suitable journal anyway.  And they accepted, though with a strange review result.  One reviewer loved the paper, but the other reviewer accused us of wilful bias in experiment design, poor writing and deliberately misconstruing the results.  Fortunately the journal editor did not agree with this review and completely rejected it.  We did have some minor amendments to complete the paper, which were completed in early 2018.

So now I had three journal papers and three conference papers under my belt, which is pretty respectable for a PhD thesis.  I decided to work part-time in March of 2018 and spend the rest of the time working on my thesis.  For six weeks of hard writing, and corrections, I hammered out a 260+ page, 101 thousand word thesis.  I had a good process, with me writing a chapter, sending to Chris for review, then updating and sending to Rich and Steve for review.  Steve didn't send any comments back, but Rich was prompt and thorough.  We ploughed through it and on the 13th of April (a Friday no less), I went up to Swinburne to submit.

I had printed out the thesis, but apparently that requirement had been dropped, though not removed from the Swinburne website.  But they accepted my printed copy anyway.  I headed home, giddy with relief, to a welcoming family.  We all went to lunch.

Eight weeks passed and I had heard nothing.  So I contacted Chris, who contact Swinburne, who contacted the reviews, who contacted Swinburne, who contacted Chris who contacted me.  That took another week.  So exactly nine weeks after submission, almost to the minute, I received the news that my thesis had been accepted, pending minor revisions.  I took the rest of the day off work.

My elation was short lived, because despite the positive result, I still had work to do.  So I buckled down again and started working on the revisions.  A week later I met again with Chris to finalise the changes required.  I thought we would knock them over that day, but alas, it was not to be.  So I took another couple of days, bouncing changes back and forth with Chris, until the following Wednesday, when he decided he was satisfied with the changes, and submitted the final thesis on my behalf. 

So I waited to hear from Swinburne HDR.  I was pretty impatient and on the Friday I rang them to find out what the process was.  I was informed that the thesis amendments would be reviewed by the HDR sub-committee the following Wednesday, and I would probably find out the results before the end of the following week.  At least I now had a timeline.

On Wednesay, the 4th of July, I received my Letter of Award, stating that my PhD thesis had been accepted by Swinburne HDR, and I was now able to call myself Doctor.  I was pretty bloody jubilent, let me tell you. 

However, it was a conditional acceptance, meaning there were several additional forms that needed to be submitted to actually complete.  I had a look at the list, and was sure I had submitted them.  In fact, my supervisor had submitted them, but apparently, they were expecting them from me, so they ignored the ones from him.  Only they didn't tell me that.  I emailed them asking if there was anything missing, and included the documents again.  They finally agreed that they had everything, and sent me a Statement of Completion.

Now I was done.  Or so I thought.  The final step is the actual graduation, which I would be invited to via an email.  I thought I had missed the cut off for August and would have to wait until December.  However, DW asked me to check the requirements, and I spotted a link on the page called "My graduation".  I clicked on it, and it prompted me for my username and password.  I logged in and was greeted with the message, "Congratulations!  You are enrolled to graduate on August 22nd!"

Don't get me wrong, this was a great outcome, but it took me by surprise.  Anyway, as I write this, it is the 20th of August.  In two days I will graduate and finally my PhD will be behind me.

And that is the story of how I got my PhD.

Ciao!

Thought for the Day: I'm pretty satisfied with level of academic achievement.  For now.

Wednesday, July 25, 2018

The doctor is in

It has been many moons since I last posted to this blog.  In fact, it has been many moons since I wrote anything significant other than my thesis or for work.

That's right, I have written my thesis.  Not only written it, but submitted it, had it accepted and have finally completed my PhD.  I am now officially not that sort of doctor.  What sort of doctor am I, I hear you ask?  I am the sort that you would call if your ultra high resolution tiled display wall powered by a GPU-enabled virtual hosted cloud-based desktop computer wasn't working.  I wouldn't be able to fix it, because I'm not that sort of doctor either.  But I could tell you what you should have considered before you bought it, and I'd probably be right.

So now I lurk around conversations at parties waiting for my area of expertise to come up, so I can jump in and show off.  It hasn't happened yet, but soon...

Having focussed myself on the big research questions for the last few years, I now find that I have to find something new to keep myself from annoying DW.  Yes, DW still hasn't left me for an actual that sort of doctor, though now when she jokingly tells people she is married to a doctor, there is a tiny little tear in her eye.

So I am turning back to my old tunes.  Literally, in the case of my guitar playing, and figuratively, in the case of my writing.  Over the last few years I have been jotting down all my ideas.  I've stopped telling them to DW because I've been able to bore regale her with juicy details about my research.  But all those ideas have still been happening, and now I have to dive back in and pick one.  That is the trouble of course, choosing just one.

And then there are all the jobs around the house, but they've been around for ages, so I'm sure they can wait a few more years...

...apparently they can't wait a few more years, or even a few more minutes.  So I'm off to do some jobs around the house.

Ciao!

Thought of the day: Happy wife, whatever it takes.