Cluster Recaps

Cluster 2

Our second week at COSMOS brought even more opportunities to explore the world of nanoscience, from glowing quantum dots to real-world biotechnology. As the topics became more advanced, we also found plenty of time to make new memories together, both in the lab and beyond.

We began the week by learning about semiconductor nanoparticles and how their electronic properties differ from those of bulk materials. As particle sizes shrink to the nanoscale, quantum confinement causes their band gaps to form discrete energy states, allowing them to absorb and emit different wavelengths of light.

On Tuesday, we put these concepts into practice during our third laboratory experiment by synthesizing cadmium selenide (CdSe) quantum dots. Starting with cadmium oxide and selenium powder, we carefully prepared the reaction before examining our products under ultraviolet light. The results were spectacular: instead of a single color, our samples produced a brilliant rainbow of fluorescence, demonstrating how slight differences in nanoparticle size can produce dramatically different optical properties. It was one of the most visually stunning experiments we have performed so far.

Wednesday brought one of the highlights of the entire program: a field trip to Thermo Fisher Scientific. After an early morning bus ride, we toured the facility and learned how scientists use genetic analysis to help diagnose diseases and guide medical decisions. It was exciting to see how many of the concepts we have discussed in class translate directly into cutting-edge biotechnology.

After lunch in the Thermo Fisher cafeteria—which included catching part of the England vs. Argentina World Cup game—we rotated through a variety of hands-on activities. These included extracting DNA from strawberries, which we got to take home; creating colorful DNA bracelets; practicing precision pipetting; and participating in several interactive demonstrations showcasing different laboratory techniques.

Most importantly, we made our final stop at H Mart. Continuing what has apparently become a Cluster 2 tradition, we stocked up on enough snacks to last the rest of the program: canned boba, mochi, chips, and just about everything else that caught our attention.

Back in the classroom on Thursday, we shifted our focus to metal oxide nanoparticles before heading into our fourth laboratory experiment, where we synthesized ferrofluids, magnetic ink, slime, and plastic through magnetite precursors. Watching these magnetic liquids respond to external magnets was both fascinating and surprisingly mesmerizing, providing another example of how materials can exhibit entirely new properties at the nanoscale.

Written by Kathy Xu

Cluster 3

During Week 2, students in Cluster 3, Making an Animal: Development and Bioinformatics, built upon the foundations they established during Week 1. They developed independent ideas using data science while also learning how cell signaling pathways can contribute to cancer and developmental problems in model organisms. After four days of learning to code with R and exploring complex signaling pathways, students took a break by visiting Santa Cruz Beach to play volleyball and enjoy the beach.

During the morning sessions, Ph.D. candidate Nicholas Chu guided students through the basics of R using pre-written code, data, and a function cheat sheet to create graphs from a given data file. Students were arranged into groups and created both scatter plots and bar graphs to display the relationships between variables across three species of penguins. The groups then advanced to interpreting volcano plots and using statistics to make claims about the significance of certain genes in prostate cancer in mice. Cluster fellow Kim Hansen also lectured on science communication, teaching students how to create impactful and engaging science posters and presentations.

During the afternoon session, Professor Zhu Wang taught students how model organisms can be used to study forward genetics, signaling pathways, and cytoplasmic determinants. One of the most memorable takeaways from the lecture was learning that mice can be intentionally given cancer using chemical compounds to help researchers study the disease. Students also learned how flies undergo mutagenesis screens, which identify mutant flies and help researchers determine which genes have been disrupted. Several signaling pathways were also explained, including the hedgehog pathway, demonstrating how the presence or absence of certain genes and molecules can lead to the repression or activation of target genes.

On Friday, the students and a very enthusiastic Professor Zhu Wang headed to Santa Cruz Beach. Students played beach volleyball, swam, and spent time relaxing on the beach. We even spotted one of our favorite model organisms, Drosophila melanogaster.

Written by Lisa Yang

Cluster 4

Our second week started with Professor Velasco’s lecture on the time-independent Schrödinger equation, where we learned to use separation of variables to simplify the complex equation. This made the confusing calculations much easier and allowed us to begin solving problems using the equation. Our first problem was the infinite potential well, where we applied the Schrödinger equation to an electron trapped in a well with walls that stretched infinitely high. After lunch, we attended Professor Syzranov’s course, where we learned another way to derive the Schrödinger equation. However, I had a harder time understanding the mathematics behind this method.

The next day, students arrived in the lecture hall ready to solve our next potential well problem. This time, the well had finite walls, which added another layer of complexity to the already difficult problems. Luckily, our professor explained the concepts very well, and we went to lunch with math spinning through our minds. In the afternoon, having finished learning the basics of quantum mechanics, Professor Syzranov talked about the history of classical information storage. We learned that the military continued using outdated technology for years, only recently discontinuing the use of floppy disks. We also had a guest lecturer, Conall McCabe from JILA, who taught us about his work with atomic tweezers.

On Wednesday, our first lecture began by finishing the finite potential well before moving on to quantum tunneling. Using our knowledge from the potential well problems, we discussed how, given the right conditions, electrons can pass through walls—something that does not align with classical physics. Imagine if humans could phase through walls! Anyway, after climbing back down the hill, Professor Syzranov began teaching us about quantum algorithms, specifically the Bernstein-Vazirani algorithm, which took us deeper into his course on quantum software. It took us a while to understand, but after working through the initial struggle, the advantages of quantum computing became much clearer.

Professor Syzranov also taught us some linear algebra, which would be useful throughout our courses. I felt enlightened to finally learn the notation we had been using for the past week; I could finally understand the content from previous lectures! “I really enjoyed learning about how the mathematical framework of quantum computing is built on linear algebra. I found it really elegant,” said Araina Gupta.

On Thursday, Professor Velasco introduced us to scanning tunneling microscopy, a technology that relies on the quantum tunneling we learned about on Wednesday. This machine can take detailed images of atoms and even manipulate them using electrical currents! We then learned about electrons in solids, using quantum physics to explain rules in chemistry, such as why atoms tend to complete their valence shells.

In our second lecture of the day, we were introduced to qubit gates, learning about operators and circuits that function similarly to coding operations. Lastly, our cluster fellow, Fabliha, gave a presentation on posters, preparing us for our final project presentations.

On Friday, we were assigned our final projects. Luckily, I think most, if not all, of us were assigned to our first-choice projects. We will be working in groups of three to five and giving presentations during the final week alongside another cluster.

In the morning, Professor Velasco introduced the Dirac delta function, which provides a simpler model than potential wells when dealing with several electrons. Previously, we had primarily solved single-electron problems, so we practiced solving problems using a new model that made the calculations easier. In the afternoon lecture, Professor Syzranov continued discussing operators, and we took notes on several new formulas commonly used in our cluster.

To end the week, our Cluster Assistant, Aidan, taught us about cybersecurity, specifically RSA encryption. With the development of quantum computing, many current cybersecurity systems could potentially be broken, making the topic especially relevant to our studies.

Written by Madison Ding

Cluster 5

For Cluster 5, the second week of COSMOS was spent exploring various game engines that increased in scope and difficulty throughout the week. Over the course of the week, Cluster 5 students also had the opportunity to present games they created with each day’s engine for bonus points.

On the first day, we started with Bitsy, a web-based game engine that allows players to create simple eight-bit games. The engine was relatively simple and limited in its capabilities; however, its small number of tools and focused design made it a great introduction to game development. On the second day, we used a more complex engine called GameMaker, a 2D game engine known for powering games such as Undertale and Pizza Tower. GameMaker was significantly more complex than the previous engine, giving Cluster 5 students more freedom in designing their games.

Throughout the lectures, we learned how to create sprites and use GameMaker’s visual scripting system to make our own 2D games. The challenge for GameMaker was to create a game that utilized collisions in order to earn bonus points.

After the second day, we used Ren’Py, a visual novel game engine that had been teased since the beginning of the week. Ren’Py is well known for powering visual novels such as Doki Doki Literature Club and Class of ’09. Many Cluster 5 students found Ren’Py easy to use, which led to several interesting projects being presented during our daily showcases.

On our fourth day, we focused on Godot, which was our instructor’s area of expertise. Godot was definitely the most difficult of the game engines we had explored so far because of its extensive capabilities, leading to it receiving the most instructional time. For Godot, we used a platformer starter project from Kenney, a free asset website for game development.

To conclude our second week of instructional lessons at COSMOS, Friday was a free-engine day, allowing students to choose a game engine to explore from a provided selection. The day also included mini-lectures during the afternoon session to prepare Cluster 5 students for their final projects. We received lessons on pitching our games, GitHub, and marketing.

Cluster 5 cannot wait to start making our final projects next week!

Written by Yisi Liu

Cluster 6

Cluster 6, Introduction to Smart and Sustainable Power, started off the week with an exciting field trip to the UC Santa Cruz cogeneration (cogen) plant! Anthony Brunetti, the cogen supervisor, gave us a tour of the facility and explained how it helps keep the campus running. Brunetti shared that the plant generates about 75% of the electricity used by UCSC’s science and engineering buildings.

One thing that really surprised us was learning that all the little glowing sensors we see around campus, in libraries, lecture halls, and other campus buildings, are constantly monitored. Operators at the cogen plant keep an eye on sensors in around 600 buildings, taking turns working 10-hour shifts to respond to any problems that come up.

Of course, someone had to ask about the latest hot topic: artificial intelligence. Brunetti explained that, no matter how smart AI becomes, there will always need to be a human operator monitoring the system. He shared a recent example in which a cooling tower malfunctioned because of a sensor issue. The automated system tried to restore the tower by increasing the water pressure to a dangerous level. Luckily, one of the operators caught the mistake and fixed the problem before anything serious happened.

Overall, Brunetti explained that AI can definitely help make the plant more efficient, but it cannot fully replace human operators. Instead, the team plans to test AI models in a sandbox environment—a safe, isolated digital space used to test software—so they can experiment with the system without impacting real-world operations.

Fun fact! Did you know that UCSC’s solar parking canopy generates more than 3 million kWh of electricity each year—enough to power more than 330 homes annually? This week, Cluster 6 got to understand how the solar panels on homes, parking structures, and other buildings actually work.

We started by learning about solar, or photovoltaic (PV), cells: electronic devices that convert sunlight directly into electricity using the photovoltaic effect. Most solar cells are made from silicon, a semiconductor that can conduct electricity under the right conditions. When sunlight, or photons, hits the silicon, it transfers energy to electrons, knocking some of them free. A built-in electric field then directs these electrons through a circuit, creating an electric current.

We also learned that pure silicon does not conduct electricity very well on its own because its electrons are tightly bound. To improve its conductivity, scientists use a process called doping, which involves adding tiny amounts of other elements to create more free electrons. This allows solar cells to convert sunlight into electricity more efficiently.

As a little change in scenery, we got the chance to do a lab outside near the Science and Engineering Building. Our TA, Elena, showed us how to use a Pathfinder to measure the solar radiation in a given spot. We mapped out the areas covered by shade and added up the numbers on the “map” that were not in the shaded areas to calculate the solar radiation. The numbers were specific to each month, and some groups measured 73% solar radiation while others measured up to 96%, depending on the mix of shade and sunlight in their locations.

Elena explained that 100% solar radiation in this case meant complete sun exposure with absolutely no shade. That is why, to make the results more interesting, most groups chose locations near the shadow of a tree for the experiment.

It didn’t end there! We then moved on to learning about wind energy. We explored why wind turbines typically have three blades: this design provides a balance of energy production, stability, and durability. One blade would create balance issues, two blades could cause the turbine to wobble, and more than three blades would create too much drag. Three blades are the Goldilocks choice for wind turbines!

To further demonstrate how wind can be a powerful energy source, our cluster built wind-powered cars with our TA, Elena. Using a basic circuit and a fan, we were able to create cars that moved forward when pushed by the wind generated by the fan.

To finish off the week, Cluster 6 went on a field trip to PG&E (Pacific Gas & Electric Company). We learned how the company provides power to our community every day through simulations, real-life testing, and smart sensors.

One of the most interesting things we learned about was arc flashes—sudden, powerful electrical explosions that occur when electric current leaves its intended path and travels through the air between two conductors or to the ground. Our tour guide gave us a simple example: imagine plugging in a charger when a metal object accidentally falls onto the exposed plug. That could create a small arc flash. Now imagine the same thing happening on one of the high-voltage utility poles that PG&E workers maintain. That is why PG&E simulates arc flashes at its testing facilities: to make sure the protective gear workers wear can withstand these dangerous events.

We also learned how PG&E plans to incorporate artificial intelligence and machine learning into its operations. By analyzing data collected from thousands of smart sensors that monitor factors such as voltage, current, temperature, and power-line conditions, AI models can help identify equipment that may be at risk of failing. This allows PG&E crews to inspect and repair infrastructure before problems lead to outages, making the electrical grid safer and more reliable.

Thank you for joining us for our second week at COSMOS! Stay tuned for Week 3 as we begin finalizing our capstone projects. See you in the next rotation—get it?!

Written by Tanisha Chatterjee and Vibha Hari

Cluster 7


If Week 1 was Cluster 7’s introduction to optics, Week 2 was when those fundamentals finally began to come together. This week, we studied incoherent and coherent light sources, laser technology, semiconductors, and Fresnel lenses—all applications of optics that are crucial to modern technology.

On Monday, we began the week by studying coherence and Young’s double-slit experiment to further our understanding of how light interacts with various surfaces. Afterwards, we reevaluated our understanding of what light is. Light can behave like a particle, an electromagnetic wave, or a ray depending on the context. However, after considering how energy is transferred to and from light, we explored the idea that photons are not particles in the same way that protons or neutrons are. Instead, photons represent discrete amounts of energy, described by the equation (E = hf), that can be transferred to or from light.

On Tuesday, we returned to the lab to study diffraction and consider applications of photonics, such as biosensing.

In the latter half of the week, we shifted our focus to lenses, beginning with convex and concave lenses and the thin-lens formula. Considering that we see convex lenses in glasses and cameras and concave lenses in peepholes and binoculars, it is clear that optics plays an impressive role in our daily lives. By learning about these concepts, we are pulling back the curtain on some of the most important advancements in the modern world.

Finally, we ended the week in the lab once more, where we worked on simulating Fresnel lenses and analyzing their components in Ansys Lumerical and 3DOptix, two software programs designed for photonics applications.

Additionally, this week we began our final projects. Splitting into groups, we will focus on biosensors, optical transceivers, semiconductor light sources, Fresnel lenses, and photovoltaic devices. A special thank you to our cluster fellow, Dalong Zhang, for guiding us through this process!

Cluster 7 has a lot in store for us over the next two weeks!

Written by Isha Malhotra

Cluster 8

Week 2 for Cluster 8 started with a “fire” topic: chemical nomenclature, taught by Professor Johnstone. First, students learned about the basic tetrahedral shape of molecules using wedges and dashes to represent alkanes such as methane, ethane, and propane. They then moved on to a simpler method of representing molecules, using lines to represent hydrocarbon chains and cyclic rings. In this method, hydrogen atoms bonded to carbon are generally not shown, while hydrogen atoms bonded to other atoms are included.

Students also learned the rules for naming different molecules. Molecules containing single bonds are called alkanes, those containing double bonds are called alkenes, and those containing triple bonds are called alkynes. Specific prefixes, such as penta-, hexa-, and deca-, are used depending on the length of the longest carbon chain. Students also learned how to number molecules based on the locations of certain atoms and functional groups.

Later, they discussed the process of deprotonation and the principles of conjugation, which involves the alternation of single and double bonds. Most importantly, Hückel’s rule was introduced. Represented by the formula (4n+2), the rule helps determine whether a molecule has the appropriate number of π electrons to be aromatic.

To explore these ideas, Cluster 8 conducted an experiment with porphyrins. Students synthesized and isolated a macrocyclic tetraphenylporphyrin derivative through a reaction between pyrrole and a benzaldehyde derivative selected by each student. Using stoichiometry, students determined the equivalent amount of their chosen benzaldehyde derivative needed for 2 mg of pyrrole. The final mixture was refluxed in propionic acid, and the resulting product was precipitated using methanol before being isolated through vacuum filtration.

In the organic electronics class, Cluster 8 made significant progress toward its final project: building a solar cell. While last week’s lectures focused on the fundamentals of electrons, Dr. Ayzner’s lectures this week placed a greater emphasis on concepts relevant to solar cells.

The cluster learned about π-conjugated molecules and how electrons in their π orbitals can become delocalized across an entire molecule. If a solar cell is made from a long π-conjugated molecule, electrons can move across the molecule more easily than if they had to jump between separate orbitals. This can help create a more efficient solar cell.

Another important reason to use longer molecules, as Cluster 8 learned this week, is that they have more easily excitable orbitals. The ionization energy—the energy needed to remove an electron from a molecule—is proportional to (1/(\text{molecule length})^2). Longer molecules also contain more electrons and orbitals that can contribute to the generation of power.

This concept was not only taught in a lecture but also tested in the lab. The cluster prepared three solutions containing oligomers of increasing lengths. By analyzing the molecules’ absorption spectra using UV-Vis spectroscopy, each lab group determined that the longer molecules absorbed lower-energy light than the shorter molecules. When a molecule absorbs light, its electrons become excited. Because the longer molecules absorbed lower-energy photons, their electrons could be excited by lower-energy light.

“I liked the lab work. I don’t get to do much of that at school,” said Cluster 8 student Gino Chang.

This week, Cluster 8 students also began working with data analysis. Each lab group received data from the UV-Vis spectrometer used in the previously mentioned experiment. Using computer software, each group was challenged to clean and reformat the data to make it easier to analyze and present accurately.

For their Friday field trip, Cluster 8 visited Unnatural Products, a biotechnology company that develops new medicines by designing synthetic macrocyclic peptides. Students toured both the biology- and chemistry-focused laboratories while also learning about the origins of the company.

In the laboratories, students saw the machinery and technology used to synthesize proteins through multiple steps and separate products using chromatography. The desired protein is tagged, impurities are filtered out, and the protein is ultimately collected. High-performance liquid chromatography (HPLC) is also used to investigate the properties of new compounds, alongside techniques such as mass spectrometry.

In the biology lab, newly synthesized molecules are tested through different assays before some proteins are sent for testing in mice. One unique aspect of Unnatural Products is its team of engineers, who design and build custom machinery. These tools allow the company to synthesize and test compounds at an incredible rate. Because the custom equipment is not available in other laboratories, it makes the company particularly valuable to larger pharmaceutical companies that want to process tests more quickly but do not have the necessary equipment to do so.

When students saw the custom machinery created by the engineering team, many found it amusing that one of the machines was named “Lil Chungus.”

After the trip, Professor Johnstone told students about his own journey to becoming a professor and principal investigator. The students enjoyed the lighthearted conversation, asking him questions about his professional life as well as fun questions such as, “What would your rapper name be?” The conversation provided insight into the different paths available to students while also serving as a relaxing way to end the week.

By Cambria Hu, Harshika Thamizhselvan, and Ari Perttula

Cluster 9

Under the guidance of Professor Jimmy Shanks and Professor Michael Doody, Cluster 9 hit the ground running Monday morning with a series of serial dilutions. Using the bacteriophages we isolated during Week 1, we worked to purify a single phage morphology. We repeated this process numerous times throughout the week, allowing us to expand our wet-lab skills while furthering our research.

During the afternoon session, we dissected a research paper about how Cas9, an enzyme, works with CRISPR RNA (crRNA) and trans-activating CRISPR RNA (tracrRNA) to target and cut DNA for precise gene editing. In groups, we were assigned different figures from the research paper to analyze and explore.

On Tuesday morning, we discussed another research paper as a class, examining how scientists were able to use phages to target and kill multidrug-resistant bacteria (MDR). Utilizing five different bacteriophages found in sewage water, researchers created a phage cocktail that was able to attack A. baumannii, a deadly pathogen that can infect members of the military.

On Wednesday afternoon, gel electrophoresis made its appearance! By adding 30 microliters of water to our digests, we were able to identify which enzyme we had used while also gaining hands-on experience with the procedure.

As Thursday rolled around—time flies!—we learned how to write our own CV, or curriculum vitae, a scientific résumé that can be useful for undergraduates. Afterwards, we conducted basic yeast transformations using our extracted plasmids to create mutations.

Nearing the end of the week, we each carried out a titer spot test during the morning session, placing our dilutions onto our phage assay plates in hopes that they will contain a countable number of plaques next week.

After lunch, we downloaded ChimeraX, a molecular visualization software that allows users to examine proteins, observe their structures, and study how they interact with other biomolecules.

Professor Shanks and Professor Doody, along with our lab assistant Cole and cluster fellow Shalini, also occasionally joined us for lunch, which was always a fun time. We were able to chat about our interests outside of COSMOS, with my peers and me graciously receiving valuable life advice and words of wisdom.

Written by Keira Eisenbud

Cluster 10

This week, Cluster 10 delved into the complexities of electron movement through the semiconductor materials used in solar cells, ending the week with an interactive lab involving a homemade solar cell. The lab materials included blackberry juice, special conductive glass, and some “exciting” chemicals: titanium dioxide and iodine.

Students were split into groups of three or four and tasked with fabricating their own solar cells, which would generate voltage and current when placed under a solar lamp. The results were mixed. The lab lacked hot plates, a necessary resource for transforming the blackberry dye into a functional photovoltaic cell. As a result, the solar cell data varied between groups, and some groups were unable to get their solar cells to work at all (this is not me complaining…).

Despite the failures, everyone seemed interested in the science behind solar cells, and the makeshift fabrication process gave us hands-on experience in semiconductor engineering. With its combination of a fast pace, quality control, and the use of multiple resources, the lab truly felt like a real engineering assignment.

In other news, our morning-class professor, “Nobby” Kobayashi, starred in Friday’s Discovery Lecture, introducing us to the complications, advantages, and key concepts behind analog-digital hybrid computing technology. Additionally, during Wednesday’s class, we learned more about our very storied and very cool professor. The lecture culminated in him encouraging us to live life to its fullest and take risks, as there is no telling where life will take you.

This week was fortunately much less work-intensive, involving less new material and a less tangible grasp of the mathematics, partially due to the introduction of our final projects. As COSMOS approaches its midway point, classes split into groups and clusters selected their topics. In the coming week, our groups of four and five will begin working toward a final product that will ultimately be subject to the judgment of all.

And so concludes Week 2: an informative eye of the storm.

Reporting by Sebastian Chen and Samuel Du

Photos taken by Maxwell Liu

Cluster 11

This week, our cluster began applying the mathematical models and theories we learned last week to feedback controllers, computer hardware, and neural networks. Along the way, we were introduced to the differential-drive robots that we will be programming over the next two weeks.

On Monday morning, we reviewed the structure of feedback control systems with Professor Ricardo and learned why real-world controllers must convert continuous mathematical models into discrete ones before they can be implemented on a computer. We also explored the idea of steady-state error and learned a systematic approach to designing controllers by discretizing a system, defining performance specifications, and selecting an appropriate controller.

Using cruise control as an example, we saw how a simple proportional controller can regulate a vehicle’s speed. In the afternoon, Dr. Himadri built on this example by deriving the control law and revisiting the automotive suspension model to show how different controllers affect a system’s behavior.

On Tuesday, Professor Jason shifted our focus to the hardware that powers modern computing. We learned about MOSFETs, the transistors that serve as the foundation of modern computer chips, and used them to build simple logic gates such as NOT, AND, and OR.

One of the highlights of the lecture came when Professor Jason connected two inverters, or NOT gates, together and showed how the output of one feeds into the other. This creates a positive feedback loop, allowing seemingly simple circuits to be combined to create SRAM, a type of memory capable of storing information. From there, we discussed why artificial intelligence requires both enormous computing power and fast memory, comparing how CPUs, GPUs, and TPUs each contribute to modern AI systems.

We finished the morning by reviewing loss functions, which measure how well a machine learning model performs. In the afternoon, our TA, Binh, introduced us to image classification using the MNIST handwritten-digit dataset. We learned the basics of artificial neural networks, explored activation functions such as ReLU and SiLU, and built our own neural networks in Google Colab to better understand how they work.

On Wednesday morning, Dr. Himadri briefly reviewed torque before deriving the linear state-space model for an automotive suspension system. We then transitioned into robotics by introducing the differential-drive robots that we will be working with over the next two weeks. By deriving equations for both translational and angular velocity, we learned how independently controlling the left and right wheels allows a robot to move forward and rotate.

In the afternoon, Binh introduced us to convolutional neural networks (CNNs), explaining how they automatically recognize important image features such as edges and patterns. Using PyTorch in Google Colab, we trained our own CNNs to classify handwritten digits and explored how adjusting parameters such as the learning rate and number of training epochs affected the models’ performance.

On Thursday, our cluster took a break from the classroom for a field trip to Intel. We learned about the company’s history, the semiconductor manufacturing process, and how modern computer chips are fabricated. We also heard the story of Intel co-founder Robert Noyce and how his early passion for engineering helped shape one of the world’s leading semiconductor companies.

After the tour, we enjoyed lunch together at H Mart before taking the bus back to campus for Rec Time.

On Friday, Professor Jason introduced us to recurrent neural networks (RNNs) and showed how they process sequences by carrying information from one step to the next through an internal hidden state. After discussing why RNNs have become less common in modern machine learning, we moved on to Professor Jason’s own research in brain-inspired computing.

We discussed the connections between spikes in the human brain and RC circuits, which can respond to sudden changes in input current. When looking at a voltage trace, we would see a spike in voltage after each jump, followed by a gradual decay. That leakage resembles an exponential decay function. In a mic-drop moment, Professor Jason connected this idea back to the graph of the hidden state, which also decreases exponentially as it is repeatedly multiplied by weights in a loop, right before we all had to leave for lunch.

When we came back, we created our own image-classification models with our TA, Binh.

As always, every evening after dinner, we met with our RAs, Sri and Blue, to play fun games such as Human Knot, Telephone, and Heads Up before heading off to our evening activities.

Written by Delisha Manuel

Cluster 12

After a fun and relaxing weekend at Santa Cruz Beach that included some exciting kayaking, the students of Cluster 12 were ready for another week of learning. Professor Dave started off the week by teaching students about the properties of magnetism, including ferromagnetism and antiferromagnetism, as well as how magnetism works at the quantum level.

An exciting demonstration in which Professor Dave heated magnetic materials to extremely high temperatures revealed that, beyond a certain temperature, materials can completely lose their magnetism. Professor Oliver also taught students more about crystal structures and crystal systems while delving into flat surfaces that cut across crystal structures, known as “Miller planes.”

Like the week before, Cluster 12 spent a significant amount of time outside the classroom, this time visiting the SLAC National Accelerator Laboratory. There, students were given a tour of the stunningly long accelerator, which measures approximately two miles in length. Tour guides enthusiastically educated students and COSMOS staff alike about the accelerator’s history and its transformation from a facility that once accelerated electrons to one that now produces some of the most powerful X-rays in the world—up to 10 trillion times more intense than the X-rays produced by machines in hospitals.

Students then returned to the lab to test another method of creating crystals under the instruction of Professor Scott Oliver and Neo Bao. On the final day of the week, they visited the computer lab to use 3D imaging software to view and manipulate crystal structures of various materials in real time.

Cluster 12 had one of its most exciting weeks so far, and many students are eager to see what else is in store.

Written by Israel Lopez 

Photos by Kammen Milchev

Cluster 13

Cluster 13’s second week went by in a flash! Overall, our students are settling into their class routines, making it seem like time is flying by. Although classes have become more routine, our professors do not hesitate to invite guest lecturers, making each session interesting and unique.

Graduate students Antonio and Hamid stepped in for Professors Baumbach and Yan during several sessions. Antonio lectured on superconductors and quantum physics, explaining their properties and applications. As a graduate student, he was extremely knowledgeable about the topic while remaining relatable and engaging, making for a great presentation.

Earlier in the week, Hamid lectured on graphene, its creation, and its applications. Graphene is a two-dimensional crystalline material that was famously discovered using Scotch tape! If you want to make your own two-dimensional material at home, all you need is a graphite pencil and a piece of tape… sort of. Hamid was excited to share his research and expressed hope for graphene’s future in technology.

This week, our Cluster Fellow, Kap, gave us lectures on transferable skills, including how to create presentations based on research and how to deliver elevator pitches. In Cluster 13, students are tasked with creating research posters and slide presentations to share with their classmates and other clusters. As we enter the third week of COSMOS, we will begin finishing these research projects, so the skills Kap shared will definitely be put to use.

These research projects address topics ranging from spin ordering to high-entropy alloys, and all of the Cluster 13 students are excited to share their findings with other students!

Finally, Antonio and Professor Baumbach’s research is reaching great heights! They were able to create more than three large NiV₂S₄ crystals, a material they have been working with extensively. The greatest challenge in creating this material is the difference between the melting points of its ingredients. Nickel and vanadium have high melting points, while sulfur tends to vaporize at relatively low temperatures, making the elements extremely difficult to combine.

Thankfully, high-end technology in the lab, including arc furnaces and self-vapor transport, allowed the researchers to create crystals large enough for research. On Friday, Professor Baumbach surprised us by bringing in a quartz tube filled with the substance. All the students could agree: it was an amazing experience to hold such an interesting material!

Over the weekend, Professor Baumbach and his lab assistants will be running extensive tests on the crystal. More to come soon!

Written by Ava Campbell