Center for Computational Sciences (CCS), University of Tsukuba, invites applications for full-time (non-tenured) faculty positions as described below. We would like to inform all institutes concerning about this field, and ask for your cooperation in recommending a suitable candidate for the position. Thank you for your support.
Position Title: Assistant professor with a fixed term appointment
Affiliation: Center for Computational Sciences, University of Tsukuba
Field of Expertise: Computational Materials Science including condensed matter physics, quantum chemistry, and molecular simulations.
Research: In our center, we seek a (non-tenured) assistant professor to develop the materials informatics (MI) method using classical molecular dynamics (MD) simulations, first-principles calculations, and machine learning (ML), and apply it to the semiconductor systems. The successful candidate will be a member of the Divisions of Life Science and Quantum Condensed Matter Physics. He/She will collaborate with researchers in the semiconductor engineering and informatics fields in Tsukuba Research Energy Center for Material Sciences (TREMS) and Center for Artificial Intelligence Research (C-AIR) as well as companies.
Starting date: October 1, 2020 or later, as soon as possible
Period: Until March 31st, 2021 (possible to renew until March 31st, 2023, upon evaluation of the progress)
Requirement: Applicants must have a doctoral degree.
Compensation
・Salary: Annual salary system (The annual salary will be determined based on the regulations of the University, taking into account the career of the employer.)
・Working hours: Discretionary labor system
・Holidays: Saturday, Sundays, national holidays, New Year’s holidays (Dec.29 Jan. 3), and holidays determined by the University.
Submissions: 1) Resume/CV (with photograph)
2) List of research activities including peer-reviewed papers, peer-reviewed proceedings, oral presentations at international conferences, competitive research funds (representative), awards, and so on.
3) Up to five reprints of major papers (at least four of which are within the last five years)
4) Summary of research to date (within 1 sheet of A4 paper)
5) Research and educational aspirations after assuming the position (about 1 sheet of A4 paper)
6) A list of two professional references with complete contact information.
Submission deadline: Friday, July 3, 2020 (JST).
Please write “Application for Assistant Professor Position in Materials Informatics” on the subject and send a zip file with a password for the documents (1-6) in the pdf format via e-mail to apply_2020_L01 [at]ccs.tsukuba.ac.jp ([at] should be replaced by @). The password is separately sent to shigeta[at]ccs.tsukuba.ac.jp ([at] should be replaced by @ as well).
Miscellaneous: The Center for Computational Sciences has been approved as a Joint Collaborative Research Center by the Ministry of Education, Culture, Sports, Science and Technology. We promote interdisciplinary computational sciences, including joint use of our supercomputer systems. The University of Tsukuba conducts its personnel selection process in compliance with the Equal Employment Opportunity Act.
—- 「これがAだよ」「これがBだよ」というラベルのついたデータを用意して、まずはそれでコンピュータにAとBのパターンを覚えてもらう、というのがトレーニングデータですね。答えが確実にわかっているデータのセット(トレーニングデータ)で学習したところに、答えがわからないデータを持ってきて、A or Bを判定してもらう・・・?
CCS Reports! 第二弾の後編は、ドイツ・フランクフルトで開催されたISC High Performance(2016年6月19日〜23日)の中の1セッション、HPC in Asia (ハイパフォーマンス・コンピューティング in アジア)にて、ポスター発表を行ったお二人の研究内容を紹介します。それぞれ、CCSで実際に動いている2台のスーパーコンピュータ、HA-PACS/TCAとCOMAに関する研究です。さっそく、スパコン研究の世界を覗いてみましょう! (2016.7.19)
GPU間の直接通信で計算性能をあげる [HA-PACS/TCA]
ISC (International Supercomputing Conference) は、スーパーコンピュータ(スパコン)とスパコンを使った計算科学の国際学会で、毎年決まって6月にヨーロッパで開催されます(2015年は7月開催)。 計算科学研究センターは、東京大学情報基盤センターと共同運営する「最先端共同HPC基盤施設:JCAHPC」としてブースを出展し、今年稼働を開始するスーパーコンピュータOakforest-PACS(OFP)の紹介を行いました。(詳しくは「ISH High Performance 参加報告(前編)」をご覧ください。) HPC in Asia (ハイパフォーマンス・コンピューティング in アジア)はISCの1セッションとして、アジアで行われているスパコン研究を世界の人々に紹介し、情報交換を行うことを目的に6月22日に開催されました。
HPC in Asia でポスター発表をした研究者へのミニインタビュー。まず一人目は、高性能計算システム研究部門の藤田典久研究員です。
わたしたちのブースの今回の目玉は、2016年12月に稼働開始するスーパーコンピュータOakforest-PACS(OFP)です。OFPは、JCAHPCが調達・運用をするスパコンで、ピーク性能は25PFLOPS(ペタフロップス)*1。現在日本で最も速いスパコン「京」が10.62PFLOSなので、OFPが稼働を開始すれば、稼働時点で国内最高性能のシステムになると見込まれています。ISC High Performanceでは、このOFPの性能に関するポスターと、実際にOFPに搭載される予定のノード*2の展示を行いました。
*1ペタフロップス:計算機の処理性能の指標として、1秒間に実行可能な浮動小数点数演算回数(実数演算回数)が用いられる。これをFLOPS(Floating-point Operations Per Second)という。PFLOPS (Peta FLOPS, ペタフロップス) = 1015FLOPSであり、1PFLOPSは一秒間に千兆回の計算ができることを意味する。 *2ノード:現在のスーパーコンピュータは、たくさんのコンピュータを高速ネットワークで繋いだ“並列型”が主流。1ノードが1コンピュータに相当。
ISCで行われるのはエキシビションだけではありません。多くの学会で行われるような研究発表・ポスター発表も含め、様々な情報交流の場が設けられています。 その1つに、HPC in Asia (ハイパフォーマンス・コンピューティング in アジア)という集まりがあります。アジアで行われているスパコンの研究や取り組みを世界の人々に紹介し、情報交換や共同研究のきっかけ作りなどを行う会合で、今年で6年目を迎えます。CCSの朴教授は、このHPC in Asia で6年間(最初からずっと!)、主催者を務めてきました。今年でその大役に一区切りをつけ、次のオーガナイザーにバトンタッチするとのこと。お疲れ様です。
[写真:開会の挨拶をする朴教授]
[写真:HPC in Asia セッション会場とポスター発表会場]
HPC in Asia では、オーストラリア、日本、韓国、台湾、中国、シンガポール、インド(発表順)のスパコンの状況について、各国の研究者から紹介がありました。今年のTop500の一位にランクインした「神威 太湖之光(中国)」や、韓国が次に導入予定のスパコンについて、シンガポールに新設された国立スーパーコンピューティング・センターについて、日本のポスト「京」コンピュータや先述のOFPについてなど、興味深い話題が次々紹介され、参加者はスライドの写真を撮ったりメモを取ったりと、熱心に聞き入っていました。
所変わってHPC in Asia のポスター会場では、コーヒー片手にポスターを介してあちこちで議論が飛び交っていました。どんな研究の話をしていたのか、計算科学研究センターからISCに参加した藤田研究員と廣川さん(博士後期課程1年)にミニインタビューをしてきたので、後編ではインタビューを中心に研究の内容をお届けします!(後編:もうしばらくお待ちください)
Researchers led by the University of Tsukuba identify two novel dinoflagellates containing relic nuclei from endosymbiont algae, making them perfect models for studying organellogenesis
Tsukuba, Japan – Many algae and plant species contain photosynthetic membrane-bound organelles called plastids that are actually remnants of a free-living cyanobacterium. At some point in evolutionary history, a cyanobacterium was engulfed by an ancestral alga, trapping it forever as a host-controlled endosymbiont in a process called organellogenesis. All modern algae and plants are the descendants of this ancestral alga containing the first plastid. But as if by karmic intervention, some of these algae were themselves engulfed during secondary endosymbiotic events, generating what are known as complex algae.
In most cases, endosymbionts lose large portions of their genomes as well as most other cellular components except plastids during organellogenesis. However, in rare cases, the relic endosymbiont nucleus is retained within the host cell, forming a nucleomorph. While researchers know that endosymbiont genes are integrated into the host genome, there are currently only a few model systems in which to study the process of organellogenesis, meaning that it is still somewhat of a mystery.
However, in a study published last month in PNAS, researchers led by the University of Tsukuba reported an exciting discovery that may shed light on the process of organellogenesis.
The team discovered two novel dinoflagellates, strains MGD and TGD, containing nucleomorphs that were undergoing endosymbiont-host DNA transfer. In cryptophytes and chlorarachniophytes, the only other algal groups known to contain nucleomorphs, all DNA transfer events have ceased, implying that organellogenesis at the genetic level is complete. This has made it impossible to discover the closest relatives of the endosymbiotic algae or to determine how their genomes are altered during the transition process.
“Morphologically, MGD and TGD were obviously distinct, which was supported by molecular phylogenetic analyses,” says senior author Professor Yuji Inagaki. “However, both strains contained green alga-derived plastids with nucleus-like structures containing DNA.”
Even though the researchers showed that the endosymbiotic algae had already been transformed into plastids, gene sequence analysis suggested that DNA transfer from the nucleomorph to the host genome was still in progress in both MGD and TGD. Given the relatively intact state of the endosymbiont genomes, the researchers successfully identified the origins of the algae to the genus level.
“Genomic analysis of these novel dinoflagellates showed that they are both nucleomorph-containing algal strains carrying plastids derived from endosymbiotic green algae, most likely of the genus Pedinomonas,” explains Professor Inagaki.
“Based on the level of integration of the endosymbiont and host genomes in MGD and TGD, we concluded that the process of organellogenesis is less advanced in these strains than that in cryptophytes and chrorarachniophytes. This important distinction will allow us to use these organisms as models to better understand the process of organellogenesis.”
Original Paper
The article, “Dinoflagellates with relic endosymbiont nuclei as models for elucidating organellogenesis,” was published in Proceedings of the National Academy of Sciences of the United States of America at DOI: 10.1073/pnas.1911884117.