{"id":2448,"date":"2015-08-26T17:45:08","date_gmt":"2015-08-26T21:45:08","guid":{"rendered":"https:\/\/scottaaronson.blog\/?p=2448"},"modified":"2017-01-12T12:53:01","modified_gmt":"2017-01-12T17:53:01","slug":"d-wave-open-thread","status":"publish","type":"post","link":"https:\/\/scottaaronson.blog\/?p=2448","title":{"rendered":"D-Wave Open Thread"},"content":{"rendered":"

A bunch of people have asked me to comment on D-Wave’s release of its 1000-qubit processor, and a paper by a group including Cathy McGeoch<\/a> saying that the machine is 1 or 2 orders of faster (in annealing time, not wall-clock time) than simulated annealing running on a single-core classical computer.\u00a0 It’s even been suggested that the “Scott-signal” has been shining brightly for a week above Quantham City, but that Scott-man has been too lazy and out-of-shape even to change into his tights.<\/p>\n

Scientifically, it’s not clear if much has changed.\u00a0 D-Wave now has a chip with twice as many qubits as the last one.\u00a0 That chip continues to be pretty effective at finding its own low-energy states: indeed, depending on various details of definition, the machine can even find its own low-energy states “faster” than some implementation of simulated annealing running on a single-core chip.\u00a0 Of course, it’s entirely possible that Matthias Troyer or Sergei Isakov or Troels Ronnow or someone like that will be able to find a better implementation of simulated annealing that closes even the modest gap—as happened the last time—but I’ll give the authors the benefit of the doubt that they put good-faith effort into optimizing the classical code.<\/p>\n

More importantly, I’d say it remains unclear whether any<\/em> of the machine’s performance on the instances tested here can be attributed to quantum tunneling effects.\u00a0 In fact, the paper explicitly states (see page 3) that it’s not going to consider such questions, and I think the authors would agree that you could very well see results like theirs, even if what was going on was fundamentally classical annealing.\u00a0 Also, of course, it’s still true that, if you wanted to solve a practical<\/em> optimization problem, you’d first need to encode it into the Chimera graph, and that reduction entails a loss that could hand a decisive advantage to simulated annealing, even without the need to go to multiple cores.\u00a0 (This is what I’ve described elsewhere as essentially all of these performance comparisons taking place on “the D-Wave machine’s home turf”: that is, on binary constraint satisfaction problems that have precisely the topology of D-Wave’s Chimera graph.)<\/p>\n

But, I dunno, I’m just not feeling the urge to analyze this in more detail.\u00a0 Part of the reason is that I think<\/em> the press might be getting less hyper-excitable these days, thereby reducing the need for a Chief D-Wave Skeptic.\u00a0 By this point, there may have been enough D-Wave announcements that papers realize they no longer need to cover each one like an extraterrestrial landing.\u00a0 And there are more hats in the ring now, with John Martinis at Google seeking to build superconducting quantum annealing machines but with ~10,000x longer coherence times than D-Wave’s, and with IBM Research and some others also trying to scale superconducting QC.\u00a0 The realization has set in, I think, that both D-Wave and the others are in this for the long haul, with D-Wave currently having lots of qubits, but with very short coherence times and unclear prospects for any quantum speedup, and Martinis and some others having qubits of far higher quality, but not yet able to couple enough of them.<\/p>\n

The other issue is that, on my flight from Seoul back to Newark, I watched two recent kids’ movies that were almost defiant<\/em> in their simple, unironic, 1950s-style messages of hope and optimism.\u00a0 One was Disney’s new live-action Cinderella<\/em>; the other was Brad Bird’s Tomorrowland<\/em>.\u00a0 And seeing these back-to-back filled me with such positivity and good will that, at least for these few hours, it’s hard to summon my usual crusty self.\u00a0 I say, let’s invent the future together, and build flying cars and jetpacks in our garages!\u00a0 Let a thousand creative ideas bloom for how to tackle climate change and the other crises facing civilization!\u00a0 (Admittedly, mass-market flying cars and jetpacks are probably not<\/em> a step forward on climate change … but, see, there’s that negativity coming back.)\u00a0 And let another<\/em> thousand ideas bloom for how to build scalable quantum computers—sure, including D-Wave’s!\u00a0 Have courage and be kind!<\/p>\n

So yeah, if readers would like to discuss the recent D-Wave paper further (especially those who know something about it), they’re more than welcome to do so in the comments section.\u00a0 But I’ve been away from Dana and Lily for two weeks, and will endeavor to spend time with them rather than obsessively reloading the comments (let’s see if I succeed).<\/p>\n

As a small token of my goodwill, I enclose two photos from my last visit to a D-Wave machine, which occurred when I met with some grad students in Waterloo this past spring.\u00a0 As you can see, I even personally certified that the machine was operating as expected.\u00a0 But more than that: surpassing all reasonable expectations for quantum AI, this model could actually converse intelligently, through a protruding head resembling that of IQC grad student Sarah Kaiser.<\/p>\n

\"dwave\"<\/a>\"dwave2\"<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

A bunch of people have asked me to comment on D-Wave’s release of its 1000-qubit processor, and a paper by a group including Cathy McGeoch saying that the machine is 1 or 2 orders of faster (in annealing time, not wall-clock time) than simulated annealing running on a single-core classical computer.\u00a0 It’s even been suggested […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"advanced_seo_description":"","jetpack_seo_html_title":"","jetpack_seo_noindex":false,"jetpack_seo_schema_type":"","_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"{title}\n\n{excerpt}\n\n{url}","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":false,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2},"_wpas_customize_per_network":false,"jetpack_post_was_ever_published":false},"categories":[4,17],"tags":[],"class_list":["post-2448","post","type-post","status-publish","format-standard","hentry","category-quantum","category-speaking-truth-to-parallelism"],"jetpack_publicize_connections":[],"jetpack_sharing_enabled":true,"jetpack_featured_media_url":"","_links":{"self":[{"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=\/wp\/v2\/posts\/2448","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2448"}],"version-history":[{"count":3,"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=\/wp\/v2\/posts\/2448\/revisions"}],"predecessor-version":[{"id":2454,"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=\/wp\/v2\/posts\/2448\/revisions\/2454"}],"wp:attachment":[{"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2448"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2448"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/scottaaronson.blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2448"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}