Application of HPC and big data in post-pandemic times

Hnry M.Tufo ,Dvi A.Yun ,Gbril Morr ,Mtthw G.Knply ,Bi Zhng ,Shi Chn

a Dept.of Computer Science,University of Colorado,Boulder,USA

b Dept.of Applied Physics and Applied Mathematics,Columbia University,USA

c College of Information Science and Engineering and College of Marine Geosciences,Ocean University of China,Qingdao,China

d Dept.of Physics and Earth Sciences,University of Louisiana,Lafayette,Louisiana,USA

e Dept.of Computer Science and Engineering,State University of New York,Buffalo,USA

f Institute of Geophysics,China Earthquake Administration,Beijing,China

g Beijing Baijiatuan,Earth Science National Observation and Research Station,Beijing,China

Keywords:Geophysics High-performance computing Higher education Post-pandemic era

ABSTRACT We lay out the ramifications of the 2020 pandemic for all people in geosciences,especially the young,and argue for significant changes on training and career development.We focus primarily on its devastating impact in USA and compare with that in other countries especially China.We review the potential effect for the next four years or so on the aspirations of an academic career versus more realistic career goals.We urge people in mid-career about the need to reassess previous goals.We stress the need for students or researchers to acquire technical skills in high-performance computing(HPC),data analytics,artificial intelligence,and/or visualization along with a broad set of technical skills in applied computer science and mathematics.We give advice about hot prospects in several areas that have great potential for advancement in the coming decade,such as visualization,deep learning,quantum computing and information,and cloud computing,all of which lie within the aegis of HPC.Our forecast is that the pandemic will significantly reshape the job landscape and career paths for both young and established researchers and we discuss bluntly the dire situation facing junior people in geosciences in the aftermath of the pandemic around the world until 2024.

1.Introduction

It has been tough for everyone since January 2020.People from every country have suffered both economically and emotionally from the lockdowns and coronavirus caused deaths and sufferings that has raged across the world.What happens next?When will our society come back to“normal”? What lessons can we extract out of previous crises and periods of instability?

The present precarious situation in America is similar to other historical scenarios.The first we call attention to is the period around the summer of 1968,which was punctuated by the violent and destructive race riots that immediately precedes the Humphrey versus Nixon election.This crisis also happened to be contemporaneous with the blossoming of applied mathematics in the late 1960s due to significant military funding(e.g.,for perturbation theory experts active in boundarylayers (Lin,1976).Moreover,we saw the birth of Computer science departments across America(e.g.,https://www.cs.umn.edu/50th-schedule and https://www.colorado.edu/cs/50-years),which sprung from traditional departments in Mathematics and Statistics and Electrical Engineering.We note that computing,which has significantly impacted the course of history since World War II (Dongarra et al.,2018),increased exponentially during this period and gave birth to the first supercomputers(e.g.,the CDC 6600 and Cray I)and the field of HPC.

A second critical period came 12 years later after the energy crisis caused by the 1973 oil embargo produced a period of high inflation and malaise in America that stubbornly persisted until Paul Volker,the Chairman of the Federal Reserve from 1979 to 1987,shocked the system by increasing interest rates to 20%and kept them elevated until inflation was vanquished.At the same time,the long cold war with Russia was coming to a head with more threats issued by L.Brezhnev.Then appeared Ronald Reagan,replacing Jimmy Carter,who was saddled with the Iran hostage crisis and other unpopular actions such as not allowing Americans to participate in the 1980 summer Olympics in Moscow.Reagan's Strategic Defense Initiative(SDI)brought supercomputing to a new level of importance for national laboratories and universities.SDI gave great impetus to the Peter Lax and James Glimm white paper in 1982 (Lax,1982) that led to the establishment in the summer of 1985 of four National Supercomputer Centers at Princeton,Cornell,the University of Illinois,and the University of California at San Diego.These Center injected computation into the universities and laid the foundation for the emergence of computational science and the use of simulation as means of scientific discovery alongside theory and experimentation.We note that a mere decade later the USA committed to nuclear stockpile verification via simulation when they joined the nuclear test ban treaty.

We take to heart the original intent of Rahm Emanuel's infamous quote during the 2008 financial crisis:“"[y]ou never want a serious crisis to go to waste.”From the video where the quote originated (Rahm Emanuel on the Opportunities of Crisis,2008) it becomes quite clear what he meant:in a crisis you have an unprecedented opportunity to deal with long-term problems that historically have been so difficult to resolve that they have been kicked down the road time and time again.He goes on to tick off several areas where significant reform has been delayed too long.Included in this list are education and training,stating quite clearly that these are broken,to which we offer a hearty hear!

2.Situation in spring of 2021

Today we are facing the pandemic along with a new Cold War between China and the USA with Russia chiming in with China (Albert,2021) Why do we reference the past? The reason is to provide young researchers the historical perspective to judge for themselves what history may have in store for them after the impact of pandemic.This may takes several years(Christakis,2020).

Without doubt young people are facing the worst situation since the World War II.Moreover,they are looking at a shaky economic situation in the USA and all around the world.In summer of 2020 we had rampant unemployment,up to 22 % in Spain,and significant economic contraction in many of the leading industrialized nations(e.g.,8%USA,7.6%Japan,12 % Canada,10 % Germany,and 6.8 % China,which has not seen a contraction since at least 1992).In spring of 2021 the situation in both the USA and China has improved but that in Western Europe is not.

People everywhere are resigned to the fate of a long-term recovery that is potentially slower and more disruptive than the Great Recession of 2008.This pandemic gives us an excellent opportunity to reflect and alter course.Here are some historical highlights to think about for who were born after 1980,when major changes were taking place in both the USA with the arrival of Ronald Reagan to revive American spirits and China with Deng Xiaoping's pushing for new industrial policy to kick off the revived Chinese economy.

From Fig.1,which summarizes the different world crises over the past 75 years and the role played by high-performance computing in each time frame.We stress here computing began its ascent in the 1940's and 1950's from the end of the World War II to the Cold War(Westad,2017)marked by increasing tension between the USA and the Soviet Union.This demanded extensive resources and focus of manpower,which led to the establishment of the most powerful computer centers of the 1950's at Livermore,California and Los Alamos,New Mexico because of the need to competitively design nuclear weapons.

Lots of changes have taken place in the development of computing industry since the mid-eighties,when the supercomputing centers were established.The computing architectures have undergone changes from monolithic supercomputers to clusters and the speed has increased by a factor of a billion to exaflops hopefully by the end of 2021.Let us look at numerical modelling in geosciences.

Climate change has now become more dominant than solid-earth geophysics in the 21stcentury,as people around the world begin to grasp the increasing gravity of the climate change crisis,which has captured the attention of political leaders from both China and the USA(Liu,2021).Geological science has been riding on the plate-tectonic revolution now for 60 years.Now we have more immediate urgent problems concerning rising sea-levels and climate change.Many geosciences departments in the USA in the 21stcentury change their names from geology or geosciences to include environmental or planetary sciences to reflect this trend of getting away from traditional solid-earth geosciences.We must remember that atmospheric scientists have become more mathematically inclined than the geophysicists only recently in the mid-20thcentury,before 1960's they were on equal footing.But they got their act together in the early 1960's and persuaded NSF to build a center for atmospheric sciences some forty years before the geodynamics established CIG,which stands for cyber-infrastructure for geodynamics.The iconic NCAR building (see Fig.2) at the base of the Flatirons in Boulder,Colorado was designed by the famous architect,I.M.Pei.Even today CIG,as it is entering its fourth phase,has no such permanent structure.This graceful building was opened on May 10th,1967 during the golden era,when America went to the moon.This iconic building was constructed because atmospheric scientists had better vision than geoscientists to get organize in the booming period of 1960s.NCAR became quickly a mecca for supercomputing and scientific computing and had an incredible fellowship program for many aspiring postdocs and famous alumni,like Dick Peltier who is now a prominent geophysical fluid dynamicist and climate expert from Canada,and a valuable summer internship program for students.In 1970,modelling was the easiest path for using the supercomputer at the time,the CDC 7600,designed by Seymour Cray.In 1976 the CRAY-1 arrived at NCAR that was 4.5 times faster and had significantly more memory.We must not forget a software infrastructure being built,like NCAR graphics,which is still widely used because of its quality in drawing vector fields with high resolution.Modelling of geophysical fluid dynamics in atmospheric science in different scales was a primary focus at NCAR forty years ago and applied mathematics was regarded then as a valuable asset to its mission and there was a strong and vitalized post-doctoral program.This type of modelling required supercomputers and the infrastructure.Since then NCAR has abandoned these earlier goals of HPC and now devotes the major efforts to observations and collecting data by aircrafts and drones,and other community services (e.g.,computing cycles and software packages).

Fig.1. A portrayal of world crises in the past seven decades and high-performance computing's role.

Fig.2. Mesa Lab from NCAR with the Rocky Mountains in the background.

3.Problems facing by young researchers

Today in pandemic times,young researchers in China have it so much easier than their counterparts in America and the Western world because of China's stricter lockdown policy and focused economic policy toward long-term scientific research.The overall impact on Chinese researchers from the pandemic has only caused a delay in their schedules of four to five months.All universities in China are opening up now in spring,2021(private communications from,Professors T.Kusky,Shihua Qi from China University of Geosciences,Wuhan,Professor Junyu Dong,from Ocean University of China,Qingdao),while in spring of 2021 all universities in the USA,especially private elite universities like Ivy League universities (Paxon,2020),Princeton University (private communication,Fredrick Simons),Rice University in Houston (private communication,Maarten De Hoop and Fenglin Niu)and Washington University in St Louis (private communication from Robert E.Criss) are remain in lock-down mode,with limited hours to laboratories and online teaching and seminars prevailing (Sato,2021).The same can be said about universities in South Korea,Japan(private communication,Professor John Hernlund,Tokyo Institute of Technology) and Western Europe.Private communication from Dean Francesco Mulargia,University of Bologna,Italy).Certain areas,in fact,are booming because more support in educational and research infrastructure is being distributed for economic recovery in the USA(Mervis,2021).But for those in America and Europe the impact is much more severe,because this shakes up the roots of university research,especially in the USA,more so than in European universities.

What will this post-pandemic situation demand on young researchers in the job market,the scientific and technological market? From experience in the past several years we see already that the jobs for most postgraduate researchers is shrinking and only top-notch specialists are needed in the premier companies like IBM,Intel,Microsoft,Google,etc.Universities are not training their students for the job market (Vedder,2019).In the post-pandemic world market,we will need people with skills in applying and understanding modern computational infrastructure.There are positions at all levels and everywhere.

Post-pandemic geoscientists will need to acquire mainly skills in data mining,data analysis,high performance computing by using both mathematical and statistical models.A lot of the demand on topics popular in pre-pandemic era,such as global geodynamics and global seismic tomography,will be slackened because of the new demands in job market and the shift in the attitude of younger generation,who often sense what the hot fields are well before their professors.

Not only young people between ages of 20 and 35 will be impacted,but also those in the middle age between 35 and 50 will feel the blows from the pandemic.The impact will be felt across the age spectrum,even those approaching retirement,as longevity increases with advances in medicine.

This pandemic will impact in particular American universities because of their business model being focused on money surrounding an environment to attract well-off domestic and foreign students.Enrollment in solid-earth geosciences,such as geodynamics will wane,because of the lack of interest from capable and ambitious students.There is also a tendency in the last 30 years to water down the curriculum in the number of mathematics and physical science courses to attract more students to earth sciences.This had led to an increase in ill-prepared domestic American graduate students.The same is happening in Western European undergraduate curriculum.

This downturn will severely impact research support at the state and national level in America.This would mean the reduction in number of positions available at all levels for supporting graduate students,outside of certain strategic fields like Artificial Intelligence (AI) and quantum computing and information.How can we expect to support postdocs in America or in Western Europe,when may the costs exceed$95,000 per year or to give summer salaries for faculty members in America,why can they earn $60,000 each summer? We may expect to see cuts in most fields,like in geophysics or geology in the near future at the federal agencies in the USA such as NSF,USGS,NASA,and DOE.But the Biden administration in America has miraculously turned this around in March 2021 (Tollefson et al.,2021),where the NSF will be doubled in four years.This situation will not be so severe in Europe but in North America and in Australia,we expect a downturn in enrollment of graduate students in these fields.Moreover,the recovery time of the economy will be at least two to three years,or even longer.Looking back now,we note that the recovery time for the Great 2008 Recession,took almost 5 years.

Numerical modelling of seismic wave propagation has been the main driver for high-performance computing in geosciences since the 1960s.Serious HPC on the part of mantle convection came much later in the late 1970s.In the last several years we see with certainty that the tide has changed and numerical modelling in geosciences falls now under a bigger scheme of things such as Data Analysis,inverse problems,uncertainty quantification and Deep learning.

All of these imminent changes would mean that we must look beyond the silo-departmental worldview at universities.Let us go more interdisciplinary and connect geosciences up with other disciplines.We must go beyond using the number of highly profiled papers in Science or Nature,or in premiere journals,like Journal of Geophysical Research or Journal of Fluid Mechanics,as a means of judging accomplishments in one's tenure package in computational geophysics.These traditional means of comparison serve to quantify the relative positions of universities in the pre-pandemic era.Producing useful artifacts(e.g.,software)should also be rewarded in post-pandemic times.

But more useful today is to inculcate graduate students to readjust to the new situation and to learn how to compute in order to find gainful employment among the new job opportunities in the post-pandemic era,since it will be difficult to land academic positions.

There are indeed many available resources to learn from to make this transition.Take software libraries as an example,such as PETSc and GeoClaw and others in many disciplines where products are delivered for the community such as workflow on large-scale simulations of material properties,e.g.VLAB in geo-materials and mineral physics,ASPECT,a code popular in modelling global geodynamics,and PyLith for crustal dynamics.In visualization,one turns to LavaVu and Underworld for modelling a variety of geophysical problems and volcanological flows.This work funded generously by the Australian government for more than 17 years and led by Louis N.Moresi,then at Monash University in Melbourne and in collaboration with Steve Quenette and Owen Kaluza at Monash.We must train students who can perform in many kinds of industries in the post-pandemic world.Training students to use GPU would be a good idea both for research and employment.GPU has now become mature.

Issues about academic tenure in the USA are also being raised during the pandemic (Htun,2020;Alexander et al.,2020;Afinogenov 2020;Deresiewicz et al.,2021).How will traditional type of tenure be maintained?Or will it move to a contractual model like in professional sports,a five-or ten-year contract? Or will we adopt the model in American medical schools,where a base salary will be paid and anything above that must come from the initiative of the professors in computational science or geophysics (e.g.,external grants).In the past few years this same question concerning tenure has already been raised in the Chinese Academy of Sciences where few people,mostly those in leadership positions,have permanent tenure positions.Tenure was a means to keep talented people at universities,but this model has caused a lot of hesitation in moving toward new directions because too many geodynamic modelers and seismologists have been trained narrowly and do not wish to reinvent themselves in newer roles.

As discussed above,global solid-earth geophysics has seen its heyday so we will not focus that much more on modelling of solid earth geophysics,such as porosity waves or volcanic eruption,post-glacial rebound and subduction processes.In the USA,subduction dynamics has been one of the main drivers for geodynamical modelling for over two decades because of the initiatives in earth sciences program of NSF.In China and other countries such as Japan or Germany there exists no such special focus on the subduction process.However,seismic wave propagation through strongly heterogenous media is actually more relevant than subduction dynamics for outside society and is of greater fundamental importance.Imaging by waves in general has relevance in many other fields in science and engineering and especially in biomedical areas.

In the 2020s solid earth geophysics is becoming less and less relevant in the overall scheme of things in American and Western European academia because of the less demand for petroleum and gas which has changed the oil industry partly due to the pandemic.(e.g.Krauss,2021).Instead other subjects like atmosphere,ocean,climate modelling,solar physics,magnetic reconnection and modelling of solar flares are leading the way because these phenomena have more societal impact(see Fig.3).Research on earthquake hazards and earthquake prediction or forecast,has been a boondoggle for a long time.Its history goes like this.In the 1975 Senator Alan Cranston from California managed to use the Palmdale Bulge,a transient phenomenon,as a means to pass a bill on earthquakes that served to build up the United States Geological Survey(USGS).The rapid growth of USGS,with a key site in Menlo Park next to Stanford University,was a political scam.Similar rapid growth of earthquake research in Japan and China were sped up following large earthquakes,the Kanto earthquake around Tokyo in 1923 and the 1976 Tangshan earthquake in China.Earthquake forecasting has proved not been able to pinpoint major earthquake events in both China and Japan.For example,the May 2008 Wenchuan earthquake in China and the great Japanese Tohoku-oki earthquake in 2011 were not forecasted.For the past thirty years in the USA there has not been any major earthquakes in heavily monitored areas either around San Francisco Bay area or in Southern California.Climate change,tsunamis and hurricanes now are more relevant these days and meaningful in the 21stcentury,because of more public attention(Liu,2021).Today landslides should not be placed behind earthquakes in research importance but instead should be drawing more focus because of increasing rainfall precipitation due to climate change,especially along coastal regions in the USA such as California and in places with mountainous topography as Sichuan.

4.Changes already afoot

Big Data has made significant inroads into our society since 2012(Lohr,2012;Mayer-Sch¨onberger and Cukier,2013).This is a field with an extremely fast rate of growth because of its applications in business,medicine and also in sciences.Geoscientists have been slow in picking this up until recently(Bergen et al.,2019).Things practiced a few years ago become obsolete quickly.We emphasize here that machine learning such as Hadoop Python,PyTorch,Keras,Hadoop,etc.and other aspects learned three years ago are already out of date for getting jobs in outside world but still relevant in geosciences and also give the students a good background.There are still many opportunities out there for those who are still practicing with these 5-to-10-year-old tools.We would stress again here that integration of geophysical models,such as involving post-glacial rebound with satellite data,will require understanding of not only the basic mathematics but also recent developments from machine learning(Vadapalli,2020;von Rueden et al.,2020;Bergen et al.,2019;Morra et al.,2021 a and b).

The purpose of this section is to emphasize in the post-pandemic world the kind of jobs PhD students in geosciences will get would require new technological skills.Stable academic positions would be more difficult to obtain and so these students and postdoctoral researchers must adapt to the reality of the immediate situation and survive financially.

Fig.3. Diagram showing the shift from solid-earth geophysical modelling to modelling in climate change and atmospheric sciences.

Visualization of data by Virtual or Augmented reality is also a promising area for employment,since one must view interactively the data and interact with the live images on the spot,such as in a forest fire.A system being developed and used in Australia,FLAIM Trainer™(https://flaimsystems.com.au),places firefighters in the most realistic training scenario available by utilizing several customized elements:Head Mounted Display:Breathing apparatus kit incorporates a head mounted VR display;Branch Nozzle:Industry Standard Hose line system and encapsulating hose and branch/nozzles tool known as Simtable.The device applies AR technology to help firefighters understand how fires may grow and travel by projecting information that helps to plan ahead when attacking forest fires.

Second,we will emphasize here that we should incorporate a more balanced and holistic educational package of using numerical methods in HPC.We include Cloud computing resources (Hwang et al.,2013) for modelling and data analysis and data assimilation.We must stress the need for data storage involving large-scale time-dependent data,like videos from numerical simulations used in machine learning of mantle convection (Shahnas et al.,2018) or streams of large-size experimental images from experiments with X Ray and neutron beams.

Third,we will advocate more on the job training.Large experiments involving virtualization are good examples.We should democratize the user to use for example NCAR CSEM,for parameter searching or other software such as ASPECT,Pylith(CIG product),Underworld(Australian product),LavaVu,a visualization tool for multiscale phenomena that was developed by the Australians under Steve Quenette at Monash University.We note the Australians have supported this project for over 17 years.Therefore,we hope governments will understand the need for long duration of time needed for software development,which includes longevity and maintenance.We need to design workflow that can autocorrect the input parameters using AI,for example,like an intelligent agent in designing the workflow for specific industrial processes,such as designing aircraft or pharmaceutical products.It is our belief that those students who can master this type of skill will be able to find well-paying jobs in the new workflow industry where efficiency counts more in the post-pandemic world.We must also make sure these students will get some certificate of competence,demonstrating their technical skills.

During the lockdown period we see the problem of communication between scientists and also in education.Much reliance was put in WebEx and Zoom in communication but not much in interactivity by visualization through Interactive visualization pioneered by Bob Haimes of MIT and Kirk E.Jordan from IBM 25 years ago(Jordan et al.,1996).We need to get visualization experts back on this track because so much in demonstration and teaching can be accomplished by this means with present day technology such as 5G phone and Edge Computing advances used on Cloud computing technology is being advanced by many companies,like IBM,Amazon and Microsoft.As mentioned already,visualization of scientific data or actual evolving phenomena,such as financial,geophysical,or pandemic events are other great areas for employment.

5.Need for HPC workers

This shorter route of producing new HPC workers within a period of two to three years is more promising than producing more doctoral students and postdocs modelling geodynamics or computational mineral physics or geochemistry(this route takes 5–6 years or more)or aspiring young or middle-aged assistant professors,who want to practice what they have learned in graduate school(this would take another 5–6 years).By this time that person would be close to 38 years old or more! For example,we would focus on providing workflow to train the new workforce well-suited for well-paying jobs.We can introduce these concepts to the national laboratories as well to increase the productivity of research staff at national labs and industries in automating the workflow with AI.Robotics and some introductions to this promising field,should be a vital part of this new curriculum.Since robotics can play an important role in outside geophysical field work.

In the post-pandemic world,when well-paid technical jobs will be scarcer.This type of training is different from what is commonly taught at universities today and is now being discussed by American HPC companies like IBM,Google,and Apple,and also,by Chinese HPC computer companies,like Sugon,Inspur and Alibaba.Students can be trained within one year to perform HPC tasks needed in many industries.Summer schools and short courses of one month duration can also train graduate students on coding and elementary robotics.

6.Discussion

Before the pandemic,America had plans in promoting cyberinfrastructure as before (Dunning et al.,2018),but lately this drive has been turning toward AI and quantum computing and information.We have already seen impact of Quantum computing and Quantum Information for education with new money invested everywhere.The NSF centers in the USA will all have an education component in quantum computing at the bachelor's level.

Most students finishing up between 2020 and 2022 are going to be hard-pressed to find a postdoc or get an academic job in the USA and western countries because of the structural changes in American universities,exacerbated by the lockdown policy.In China there is less of an impact from pandemic on universities.If anything,more emphasis is being put in on high-technology and environmental geosciences.

The pandemic has already caused migration of jobs and people.Many are leaving big cities now and go out to quieter places,where there is less turmoil.There will be a migration of jobs from large metropolitan areas to smaller cities in many countries,particularly in China From Beijing to Shenzhen,Chengdu,Qingdao.In the USA from cities like New York and Chicago to the South,Georgia,Louisiana,from Bay Area and Los Angeles California to Oregon,Utah,Idaho and Texas.

Students can be prepared at the Junior College level and to Bachelor's level for getting jobs in workflow from the Applied Computer Science departments,as opposed to the dominant flavor to date of pure computer science,whose interests lies in developing algorithms for small toy problems,these faculty members have their hearts mainly to write papers and the computer science department likes to have many courses available to teach for these faculty members.

6.1.What are we going to do for geophysics or atmospheric science students?

The physical models in geosciences need one to calibrate with data assimilation for finding the correct model parameters.Consequently,we must handle solutions of inverse problem with uncertainties (Tenorio,2017;Aster et al.,2013).

Solving these inverse problems,on the other hand,will require the geoscience students to delve into the application of high-performance HPC infrastructures,on how to handle big data and the complexity of the formal inverse problems.Therefore,we recommend that geoscience students acquire some knowledge about existing,ready-to-use software platforms and libraries using supporting scripting languages they can learn how to employ multithread and multi-GPU computational facilities.Some foundations in HPC is necessary.They should know how to apply ready-to-use big modelling codes by using resources supplied by HPC and data clouds and meta-procedures,such as workflows for all kinds of interesting problems outside geosciences.On the other hand,harnessing big distributed data will require one to apply the machine learning algorithms,particularly deep neural networks for large computers,in a different way than for data which can be fit within memory of small computers.

6.2.New developments down the road

Here we will discuss briefly about the impact of quantum computing(QC) and quantum information.They have been hot topics for the last several years,because QC can supplant traditional computing in many subjects in terms of speed and memory capacity (Nielsen and Chuang,2011).Quantum computing,in particular quantum machine learning algorithms,can solve many problems in the near future earlier than people could imagine today because of the intense interests now being focused on quantum computing and information.Therefore,a rudimentary but sound knowledge in quantum computing also should be taught to young researchers,even at high school to get the outstanding students’attention and certainly in undergraduate programs.Several years down the road,they should smoothly acquire coding skills to harness quantum computing for deep learning problems or for material property modelling.

Cloud computing has been around for a long time and has matured.Cloud computing in open-stack mode is also important for handling affairs many types of environment:municipal,regional,national and global.Thanks to the pandemic,during lockdown period the Cloud Computing arena has become more prominent because of the heroic efforts by Zoom,Amazon,and Microsoft.Cloud computing demonstrates now it can play an important role in HPC on demand and in data analysis.One now appreciates its capability in storage of large data sets,which is not easily available at university or NSF centers.

We now discuss the impact of Quantum computing (QC) and Quantum Information(QI)for education.To be sure,there is new money put into QC everywhere in the world.The four NSF centers and DOE centers in the USA will all have an education component in quantum computing even at the bachelor's level.

Quantum Computing and Quantum Information are up and coming(National Academy of Science,2019,2020).There is gathering a lot of interest from many segments of society,like a gold rush.This is where the action lies in the future for workforce skilled in quantum computing and quantum information within a decade from now,regional,national and global.

7.Final remarks

What do we see in the near future for young researchers in geosciences in America and China?Both countries are plunging ahead at full tilt in quantum computing and Cloud Computing.These fields which include Cloud computing and Quantum computing and quantum information,and also robotics should then be the foci for young people who want to get a decently paying job by the year 2025,because there will be a shortage of these uniquely capable people.Instead of continuing on the academic treadmill,which in the USA is under great strain precipitated by online classes during lockdown period,motivating young geophysicists can reinvent themselves as programmers in quantum computing for wave phenomena or in visualization.Others can get into GPU computing,which still has a lot of mileage left.There are now numerous long-term multidisciplinary programs funded generously by NSF and DOE in America,which addresses these issues of training a workforce for quantum computing and information science linked with artificial intelligence.In China we expect a similar trend to take place in the near future.

We will give a forecast of the state in quantum computing linking to deep learning(Beer et al.,2020)up to the year 2025.Young researchers should have some clear thoughts in mind and compare the multiple paths posed by Cloud Computing,Big Data and Deep Learning and Quantum computing and Quantum information (see Fig.4).Quantum computing lies at the crux of this venture.We must integrate these technologies together to develop new opportunities for employment in new technical areas of utility to society and not just writing papers for justifying the existence of these NSF and DOE centers.

In summary,our viewpoints for researchers in geosciences are:

(1) It will be difficult for the young to obtain academic positions for at least the next 5 years,even with a robust rebound.This is especially true in the USA under any administration and also in Western countries like Australia,Canada and Europe.China,however,has a brighter future because of its philosophy in investment in education over the long haul.Even if you are able to obtain a tenure-track position the notion of tenure and life-time employment is quickly eroding.This stability is precisely what has been used to counterbalance the lost income from the tenureemployment route,both in terms of lower salary but also lost opportunity from remaining in school to obtain an MS and/or PhD and often having to take postdoc positions or other unprotected positions to get by until a tenure-track position opens up.

(2) Traditional type of direct numerical simulation is outdated in the era of Big Data and has lost much its promise back a decade ago.However,one can still find well-paying modelling jobs outside academia.

(3) For the young,HPC jobs are a better way to make a decent living,if you wish to remain in science and high technology.There are many opportunities in HPC jobs:(a)GPU programming,a mature subject Cloud Computing,also has become mature(b)Visualization:both Virtual reality and Augmented reality,becoming mature,especially in geological field work,a killer application is Augmented reality (c) Quantum computing and Quantum information,not right now but in the near future by 2028 or so,One better be aware and be prepared for this venture.Quantum information will realize its potential well before quantum computing.

Fig.4. Vision for HPC in the post pandemic era.

(4) What about the middle age or established people between the age of 35–55 a decade or so before retirement? These middle-age or close to retirement professors should realize that after the pandemic they cannot grow their research program as before by luring Chinese and foreign students to the USA.While this trend has been thriving for the past 35 years,it is coming to an end like the production of American oil in the mid-eighties,it reached a peak and we had two energy crises.Yes,we are entering into something new.These professors should try to reinvent themselves.Yes,they should prepare for their financial and mental well-being because of the danger of increasing furloughs at universities and threat of tenure in the future?

(5) Others,in no particular order

(a) Critical thinking skills are no longer being developed in K-12 or higher education.Moreover,the idea that everyone deserves first place and the creation of mechanisms like bullying,microaggressions,etc.are an anathema to produce a dynamic and innovative work force.

(b) Trade schools and spot training deserve more respect they are given.These can make even those that have only graduated high school useful in a post-pandemic industrial economy.And spot training is important when the required skill sets are rapidly changing.

(c) Artifacts as demonstrated by individually written software are more important than papers in evaluating a researcher's contributions and skillset,in most cases.

(d) The silo structure (departments,institutes,and colleges) of universities is not compatible with the multidisciplinary and collaborative nature of funding and projects.

(e) Having a first-class cyber-infrastructure is required for doing research.Unfortunately,this cannot be entirely in the cloud.

(f) The current cold war will make open research more difficult to justify.

(g) The funding agencies in the USA are not well equipped to lead the research anymore.The lack of money (save for NIH) is problematic.The second-rate people that work there are also problematic.

(h) The USA is devoting too little to R&D.The share of federal dollars,in relation to the size of the economy,has been shrinking for decades.And much of the business R&D has been budgeted out of existence or recast as product development.The great labs of old,Bell Labs,IBM T.J.Watson,Exxon Research Lab,Cray Research Inc.etc.are shadows of their former selves.It is not clear that Google,Microsoft,Facebook,etc.have built comparable entities.

(i) Academics need to cooperate more with computing industry for research support and training of students for industrial research.

(j) We note that many of the suggestions given here can be applied to many other fields in science,e.g.chemistry,physics,mathematics,biology,and geography.

Acknowledgements

We thank Kirk E.Jordan,Witek Dzwinel,Steve Quenette and Danny Loegering for discussion,Arthur Zhong and Yingchun Liu for figure production.We are grateful to the editor for her gracious understanding.This research has been supported by NSF and DOE grants to Henry M.Tufo,David A.Yuen and Matthew G.Knepley.