BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Drupal iCal API//EN
X-WR-CALNAME:Events items teaser
X-WR-TIMEZONE:America/Toronto
BEGIN:VTIMEZONE
TZID:America/Toronto
X-LIC-LOCATION:America/Toronto
BEGIN:DAYLIGHT
TZNAME:EDT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
DTSTART:20220313T070000
END:DAYLIGHT
BEGIN:STANDARD
TZNAME:EST
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
DTSTART:20211107T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:682a43b355609
DTSTART;TZID=America/Toronto:20220907T120000
SEQUENCE:0
TRANSP:TRANSPARENT
DTEND;TZID=America/Toronto:20220907T130000
URL:/institute-for-quantum-computing/events/iqc-student
 -seminar-featuring-joan-arrow
SUMMARY:IQC Student Seminar featuring Joan Arrow
CLASS:PUBLIC
DESCRIPTION:Summary \n\nASSESSING THE TRAINABILITY OF THE VARIATIONAL QUANT
 UM STATE\nDIAGONALIZATION ALGORITHM AT SCALE\n\nDeveloping new quantum alg
 orithms is a famously hard problem. The lack\nof intuition concerning the 
 quantum realm makes constructing quantum\nalgorithms that solve particular
  problems of interest difficult. In\naddition\, modern hardware limitation
 s place strong restrictions on the\ntypes of algorithms which can be imple
 mented in noisy circuits. These\nchallenges have produced several solution
 s to the problem of quantum\nalgorithm development in the modern Near-term
  Intermediate Scale\nQuantum (NISQ) Era. One of the most prominent of thes
 e is the use of\nclassical machine learning to discover novel quantum algo
 rithms by\nminimizing a cost function associated with the particular appli
 cation\nof interest. This quantum-classical hybrid approach\, also called\
 nVariational Quantum Algorithms (VQAs)\, has attracted major interest\nfro
 m both academic and industrial researchers due to its flexible\nframework 
 and expanding list of applications - most notably\noptimization (QAOA) and
  chemistry (VQE). What is still unclear is\nwhether these algorithms will 
 deliver on their promise when\nimplemented at a useful scale\, in fact the
 re is strong reason to worry\nwhether the classical machine learning model
  will be able to train in\nthe larger parameter space. This phenomenon is 
 commonly referred to as\nthe Barren Plateaus problem\, which occurs when t
 he training gradient\nvanishes exponentially quickly as the system size in
 creases. Recent\nresults have shown that some cost functions used in train
 ing can be\nproven to result in a barren plateau\, while other cost functi
 ons can\nbe proven to avoid them. In this presentation\, I apply these res
 ults\nto my 2018 paper where my group developed a new Variational Quantum\
 nState Diagonalization (VQSD) algorithm and so demonstrate that this\nalgo
 rithm's current cost function will encounter a Barren Plateau at\nscale. 
 I then introduce a simple modification to this cost function\nwhich preser
 ves its function while ensuring trainability at scale. I\nalso discuss the
  next steps for this project where I am teaching a\nteam of 6 quantum novi
 ces across 4 continents the core calculation I\nuse in this work to expand
  my analysis to the entire literature of\nVQAs.\n\nReference: https://uwsp
 ace.uwaterloo.ca/handle/10012/18187\n 
DTSTAMP:20250518T203147Z
END:VEVENT
END:VCALENDAR