Current location - Education and Training Encyclopedia - Graduation thesis - Replication Crisis of Quantum Computing: Mayorana Fermion
Replication Crisis of Quantum Computing: Mayorana Fermion
The experiment of finding majorana signal is carried out by putting nanowires into a dilution refrigerator, which can cool them to near absolute zero. Source: HGA architects and engineers.

The race to detect the new quantum particle Majorana Fermion has cast a shadow. This kind of quantum particles can provide power for quantum computers. As a person working in this field, I began to worry that after a series of mistakes began, a large part of majorana's field was deceiving myself. Several key experiments claiming to detect majorana particles were initially considered as breakthroughs, but they have not been confirmed. A recent case of high-profile exit from Nature (see Nature 59 1, 354-355; This was initiated by my colleague Vincent Murick, a physicist at the University of New South Wales in Sydney, Australia. We got extra data from the initial experiment, but these data were not included in the paper, and then we raised concerns.

This matter is very important. Italian physicist ettore Majorana predicted in 1937 that majorana particles are theoretically their own antiparticles. Computer giant Microsoft hopes to build a reliable quantum computer by using majorana particles: these particles can make extremely stable qubits. The scientific excitement around them is no less than gravitational waves and Higgs bosons.

In the experiment, researchers disagree on whether majorana Island has been detected, let alone whether it is an asset of quantum computing. With the suspicion of this statement gradually spreading beyond connoisseurs, the field is facing the risk of reputation damage, although its prospects have not yet been developed.

It is very difficult to produce Marjona fish in the laboratory. This experiment combines the frontier fields of nanotechnology, superconductivity, equipment engineering and materials science. In the most advanced method, researchers must first cultivate a nanowire crystal-a feat in itself-to produce an atomic column with a diameter of 100 nanometer (one thousandth of the width of human hair). Then, they must connect the wire to a circuit sensitive enough to measure the individual electrons passing through it. The whole experiment must be carried out at the temperature above absolute zero 1% and the magnetic field is 0/0000 times of the earth's magnetic field.

In these extreme cases, when all the electrons in the wire are magnetized, majorana particles will appear from both ends of the wire. Theoretically, yes.

More than 100 organizations have tried it. It is reported that 24 people demonstrated in Mallorana Island. They usually appear in the form of characteristic electronic signals: as the voltage on the nanowire changes, the current will have a narrow peak. I was one of the first team members to observe this phenomenon, which was on 20 12. Soon, more newspapers appeared. The detection of current quantization value was first predicted in theory, and then the experimental reports published in Science 2 of 20 17 and Nature 3 of 20 18 were interpreted by many people as the final evidence of the existence of majorana Island.

In 2020, after repeated experiments, these observations were carefully examined. Science magazine published an experiment led by researchers from Pennsylvania State University at Park, which contradicted the report of 20 17. My team copied the patterns in 20 18 Nature Research, but proved that they did not necessarily come from Mallorana Island. We cross-checked both ends of the same nanowire, but found that only one end had a current peak. This goes against the basic expectation of the theory that majorana people always appear in pairs. The speed of refutation is accelerating: researchers have not been able to confirm the findings of two independent papers claiming to find the majorana mechanism in nanowires. The report on the current peak of the new iron-based superconductor Fe(Te, Se) was quoted by Majoranas10-2 in Science and Natural Communication, and it will be more detailed after the publication of Physical Review Letters this year.

Lesson: majorana particles are not a necessary condition for generating current peak signals. At least since 20 14, we have known some more common explanations, such as other quantum states besides Majoranas 14, unexpected signals caused by nanowire defects, or cooperative behaviors that many electrons are fascinating but have explored before (see "mixed signals"). However, positive documents keep appearing without even mentioning other explanations, giving the impression that there is a heated debate between optimists and pessimists in majorana.

As a person who has published and commented on majorana's positive and negative views, I feel a broader problem. This debate has begun to weaken people's confidence in the basic experimental method of current passing through quantum objects, although this powerful technology has been used in many important discoveries, including the observation of superconductivity, quantum Hall effect and tunneling effect that won the Nobel Prize.

It's starting to affect me. Future graduate students will ask me if I want to stop studying in majorana. Grant reviewers believe that methodology rather than selective data reporting has led to confusion in this field.

In my opinion, the basic method commonly known as "quantum transmission" is not wrong. I think selective data presentation is the main problem. If all papers contain complete or at least properly selected data sets, quantum physicists can give correct explanations, regardless of whether majorana Island exists or not.

But I think researchers are very picky-focusing on data consistent with majorana's theory and putting aside inconsistent data. A good example is a paper on Fe(Te, Se) published by Science in 2020, which reported the quantization behavior of current. Among the 60 eddy currents evaluated, the author saw this behavior in a single eddy current. I think some journals and reviewers who may not be rigorous enough can support data selection researchers. (When asked about the 2020 paper, a spokesman for Science magazine said that the results and conclusions, including the explanation of the observed quantitative substitution mechanism, were carefully introduced. I and other reviewers have repeatedly argued that journals should not publish papers based on selective data display, but only see them appear in other (sometimes the same) journals. Sometimes, if a chart can tell the whole story, it is not necessary to present all the data. But for majorana particles, it is not enough to determine the peak of the correct height only by data search, especially in the case of substitution theory.

Selection bias can easily dominate hypothesis-driven experimental research. The "best" data are usually considered to be those that conform to the theory. Therefore, the deviation can easily be ruled out as experimental or human error, thus being ruled out.

Another problem is that the scope of peer review to review majorana's claim is too wide. In any multidisciplinary field, review is difficult. Referees are often experts in one discipline, and it is difficult to judge other disciplines, which leaves a blank. For example, theoretical physicists may be satisfied with the evaluation calculation, but not with the experimental process, while materials scientists who know how to grow nanowires may skip the theoretical part. But in order to evaluate this research correctly, we need to have an overall view of the whole research.

This is a familiar story. Nature investigates the "recurring crisis" in chemistry, biology, physics, engineering and medical sciences (see Nature 533, 452-454; (20 16), selective reporting of grades is the culprit. We have all seen this for decades. Physicist robert millikan missed some famous data points in the oil drop experiment more than a century ago. He is indeed close to the actual value of the electron charge-but science cannot rely on this fluke. Because of the way of data selection, some papers in Mahorana Island proved to be unreliable.

The code of conduct in the whole field of condensed matter physics needs to be updated. There is only one solution, and that is to comprehensively strengthen the accountability system. The following steps will contribute to the research and other fields in Mallorana.

Open data. Scientists should disclose all the data in the repository, and abide by * * * access standards, such as fairness (searchability, accessibility, interoperability and reusability) 15. Some management is inevitable. The amount of data collected by modern physics laboratories is very large: computer scripts control equipment, which may run 24 hours a day. One remedy is to clearly explain the protocol used to perform any data selection-so that others can reuse or review it. Remember, data selection is a form of data processing.

Journals, funders (including companies), research laboratories and universities should demand such open data practices, just as they do in clinical trials, genomics, earth sciences and other disciplines. * * * Accessing data can improve reliability, promote collaboration and speed up progress. For example, the high-energy physics community can teach others how to share research agreements so that every paper can be repeated or copied.

Although this is not widely known, many publishing policies and government research codes of conduct require further data. It is worth noting that compared with other countries that have invested heavily in research, the United States has no national regulations. Further efforts are needed to automate this enjoyment rather than "on demand". As the paper on Mallorana recently withdrawn by Nature shows, it is very important to see complete data for evaluating an experiment.

Critics will retort that sharing data alone can't capture everything that happens in the laboratory, and experience and insight-process-have value that the agreement can't describe. I believe that reliable and useful science is based on reliable processes, which can be repeatedly checked, verified and re-examined if necessary.

The process of opening. Commentators need to ask more questions about unusual statements. Is the result too good to be true? Have you provided enough data? Have you considered other explanations? Cross-checking should be done, which makes it more difficult to prove an unreliable statement. For majorana physics, it is as basic as comparing the dependence of magnetic field and electric field of current peak with theoretical expectation. If we insist on doing this, it will frustrate many wrong ideas.

But even the most rigorous censorship may be ignored. If the paper is rejected, the authors can ignore all the input and send their manuscripts to another journal. I saw some papers in majorana Island, which were repeatedly criticized and rejected for scientific reasons, but the papers published in another high-profile journal had only minor changes. Publicizing the notorious opaque publishing procedure is the key to reduce the spread of bad research.

Editors should take responsibility: they are the ones who make decisions, even if they lack in-depth professional knowledge about the topic of the paper. Every rented newspaper should be accompanied by the editor's name. For each withdrawal, editors should provide their views on what happened. All journals, especially high-impact journals, need to be supervised by the community. Editorial retraction should be widely used, because it may take a long time to wait for the author to retract the manuscript himself. At present, most journals can't even conduct their own investigation on the mistakes in their papers. They should build this ability with the help of the research community.

What about the study of Mallorana Island? It is still feasible and important. However, in my opinion, the key findings have not been discovered. Now we need to concentrate on improving our nanowire materials, experimental techniques and data analysis, as well as sorting out other explanations. Reliable evidence is needed to prove that the particle is indeed its own antiparticle, and there is complete data in our eyes.

Only in this way can we prepare to develop the majorana quantum computer.

Nature 592,350-352 (2021)

doi:https://doi.org/ 10. 1038/d4 1586-02 1-00954-8