1 00:00:04,471 --> 00:00:05,939 I'm Irina Paci. 2 00:00:05,939 --> 00:00:09,576 I'm a Professor in the Chemistry department at the University of Victoria. 3 00:00:09,576 --> 00:00:14,314 My group is interested in doing computer-based simulations, 4 00:00:14,314 --> 00:00:17,484 modeling of complex materials. 5 00:00:17,484 --> 00:00:20,904 So a lot of the research in academia and in industry these days 6 00:00:20,904 --> 00:00:23,356 has to do with designing new materials 7 00:00:23,356 --> 00:00:24,591 for all kinds of applications 8 00:00:24,591 --> 00:00:26,526 for new cell phones, 9 00:00:26,526 --> 00:00:28,945 larger capacity for storage, 10 00:00:28,945 --> 00:00:30,447 new computers even. 11 00:00:30,447 --> 00:00:34,117 But alongside the developments in experimental research, 12 00:00:34,117 --> 00:00:36,553 there have been a lot of developments in computing. 13 00:00:36,553 --> 00:00:37,737 We get to use that 14 00:00:37,737 --> 00:00:42,142 in order to simulate and model materials 15 00:00:42,142 --> 00:00:45,979 on longer timescales on a larger scale. 16 00:00:46,012 --> 00:00:48,048 So we're basically approaching 17 00:00:48,048 --> 00:00:50,567 the same sort of timescales and length scales 18 00:00:50,567 --> 00:00:52,919 that experimentalists can approach. 19 00:00:52,919 --> 00:00:55,905 And in order to do these simulations, 20 00:00:55,905 --> 00:00:58,324 we need to use very large computers. 21 00:00:58,324 --> 00:00:59,876 I feel lucky to be a part of a group 22 00:00:59,876 --> 00:01:01,494 with such diverse research interests 23 00:01:01,494 --> 00:01:04,130 but we're all tied together by this desire to understand 24 00:01:04,130 --> 00:01:08,068 materials chemistry at a really fundamental level through simulation. 25 00:01:08,068 --> 00:01:09,502 One of my favorite recent projects 26 00:01:09,502 --> 00:01:13,106 was simulating the catalysts used in hydrogen fuel cells. 27 00:01:13,106 --> 00:01:15,108 These simulations use quantum mechanics 28 00:01:15,108 --> 00:01:18,294 to allow us to better understand the properties of good catalysts, 29 00:01:18,294 --> 00:01:21,364 and they can help identify promising new catalyst candidates 30 00:01:21,364 --> 00:01:23,333 before making them in a lab. 31 00:01:23,333 --> 00:01:25,852 The issue though, is that these types of simulations 32 00:01:25,852 --> 00:01:28,104 are quite expensive to run computationally. 33 00:01:28,104 --> 00:01:31,508 So running them on a laptop or a normal desktop computer 34 00:01:31,508 --> 00:01:32,959 is really out of the question. 35 00:01:32,959 --> 00:01:33,460 In fact 36 00:01:33,460 --> 00:01:35,762 one of the simulation methodologies that I use 37 00:01:35,762 --> 00:01:39,966 required a kind of uncommon allocation of computational resources 38 00:01:39,966 --> 00:01:42,719 and the research computing specialists that manage Arbutus 39 00:01:42,719 --> 00:01:45,405 were able to configure several compute nodes 40 00:01:45,405 --> 00:01:47,240 to meet these requirements and ultimately 41 00:01:47,240 --> 00:01:49,993 help me get the results that I needed for my research. 42 00:01:49,993 --> 00:01:51,744 I am a computational chemist 43 00:01:51,744 --> 00:01:54,614 which means that I use computing clusters 44 00:01:54,614 --> 00:01:57,801 to simulate materials at the atomic and molecular scale. 45 00:01:57,801 --> 00:01:59,269 This is a really important thing to do 46 00:01:59,269 --> 00:02:01,738 because making these materials in a lab is expensive 47 00:02:01,738 --> 00:02:03,740 and generates a lot of waste. 48 00:02:03,740 --> 00:02:06,593 And if we want to make, say, a better semiconductor 49 00:02:06,593 --> 00:02:09,829 or supercapacitor for charge storage or transportation, 50 00:02:09,829 --> 00:02:12,966 we need to know what's happening at the atomic scale 51 00:02:12,966 --> 00:02:14,033 And for this I use 52 00:02:14,033 --> 00:02:18,321 the HPC clusters like Cedar or SciNet clusters. 53 00:02:18,321 --> 00:02:20,557 It's something that I can't do on my laptop 54 00:02:20,557 --> 00:02:23,493 because these have hundreds or thousands of electrons, 55 00:02:23,493 --> 00:02:25,745 and the computations are just too large. 56 00:02:25,745 --> 00:02:28,264 And also, it's quite nice to have somebody else 57 00:02:28,264 --> 00:02:31,467 to compile my software for me. 58 00:02:31,467 --> 00:02:34,871 I work on simulating small molecules on surfaces. 59 00:02:34,871 --> 00:02:36,940 One prevalent issue in device fabrication 60 00:02:36,940 --> 00:02:39,859 is the trade-off between efficiency 61 00:02:39,859 --> 00:02:43,196 and manufacturing perfect or ideal materials. 62 00:02:43,196 --> 00:02:45,865 Nanodevices are one of such materials 63 00:02:45,865 --> 00:02:49,869 and they are used in several purposes, including biosensing. 64 00:02:49,869 --> 00:02:51,487 Because they are very small 65 00:02:51,487 --> 00:02:54,073 we can't study them experimentally. 66 00:02:54,073 --> 00:02:57,977 Hence we use computational methods on supercomputers 67 00:02:57,977 --> 00:02:59,846 to simulate these devices 68 00:02:59,846 --> 00:03:02,715 and the reactions that occur on these devices. 69 00:03:02,715 --> 00:03:04,767 The research that I've been doing recently 70 00:03:04,767 --> 00:03:09,405 is related to the adsorption of cysteine on gold surfaces. 71 00:03:09,405 --> 00:03:11,207 Since cysteine is an amino acid 72 00:03:11,207 --> 00:03:14,611 it obviously has a lot of applications in Biology, 73 00:03:14,611 --> 00:03:18,181 and the adsorption itself has relevance in biosensors, 74 00:03:18,181 --> 00:03:21,784 and perhaps in surface-directed protein synthesis. 75 00:03:21,784 --> 00:03:25,488 Because of the size of these systems with hundreds of atoms 76 00:03:25,488 --> 00:03:28,474 and tens of thousands or even hundreds of thousands of electrons, 77 00:03:28,474 --> 00:03:31,928 we really need high performance computing resources 78 00:03:31,928 --> 00:03:33,680 which are provided by the DRA. 79 00:03:33,680 --> 00:03:35,481 Without those resources 80 00:03:35,481 --> 00:03:37,784 it would just be impossible to do these calculations 81 00:03:37,784 --> 00:03:40,620 on a personal computer, or even a very powerful one.