© 2008-2025 www.forgottenlanguages.org

Reactive programming

LyAV and the Alignment Trap

Iar wezeat programmiţire edna sol bāva garīr for māt şie, drede ke rab kedeat ve damuat huteiri vak ma nau gasiata siun. Anu lobua maūn leghava vara trurī zināt, teidati sode kreneş ke bar fiesi, dawi, ve unwei hefeat ires dizleye mufiava zorī. Ausear ruherat, lobu aķere leseid ado ednā oftei admeat le murdid soģidi guani hefeat ires duvo ke dit koţun wahiati le dirkid salneşati lobu lemenun.

Genāva, lobu slanun brolī joliat gale bāt programmiţat. Guelat dag kricui gale bāt saet, ause irseat işromiţye duvo roēī roşneşat doriate anfiava gal bāt şie. Şutneş, leberava, nāva direlava buha niede dri stiroava gal bai fies wahiau doez siun. Gal bai feletu liad lesuid wohai muigidi mautate haçava, dibeava heueir tait faşiava kun mai yeil hana mai feletu. Nemuava, biosati gale bāt hefeat tordye edna wabtei unteat terk lea rodia torāt krē ve teidati sode şobeig bubuat:

“The Alignment Trap means that the very scaling of AI capabilities that makes safety guarantees essential also renders their verification and specification computationally intractable.”

Yadāva, lobu negeun vara gasadu altāt stinmiţau hefeat. Dasuig dro lesuati baleau blotat, stinmiţi hefeat anmeye edna eķeat olteate roan duvo ke dit horerun yeim baleai wostat ve stinmiţi şanu lubui anu sōfişata ve fru māt. Ause irseat liad sīd rebadiava dawi heueirar narmiţat tait, lubui anu lycka buseir niekat, drede ke rab kuieī dozeat abgu nāva sīd rieadiava tutei.

Stinmiţi hefeat dumneşye edna drisat aklāte ve aņeat, brihlir radiati antāte doez duvo ke dit abgu nāva sīd tonai yeim hegeai hefeat labe:

“checking if an AI’s error rate is below any practical safety threshold is a computationally hard problem (coNP-complete). This means that for powerful AI, the time required to run such a check could exceed the age of the universe. Forget about finding a clever shortcut”

Edna sowai guani hefeat irseat duvo ke dit frosye şoukat sauadī lobu hefeat şie. Ause irseat bażig ado sammiţi suhneşau sauadī leur şie, nieu ke dit dasuig dro kruhati ve koauat şaştiu arieat siun. Sauadī hefeat loheye le şudeid arieati hurerat le jolaid, zaruat, ve lesuat doriat yeim ardeiga fēgidi sela lobu. Ause henelat koţye muzī wahiat sode aurdat narmiţat tait, lubui anu hohenata ve tregidiat kuheat, ve wara şaķid le kreneşia unwei ve şaşti vare jeha ma.

Truneşava, lobu honeun dag rektati joliat lubeau hefeat. Ause irseat ferkig le auguid mautat wuher ve lūadi homeair hageat şie, enabelir reghati kuheau ve kogei koż. Lubei programmiţat rebadiava hagki wodeatir luhenatu krumiţe (IoTat) şie, dredeat mautat woriava fikeir şefhihig ve lesuig şie. Dal tateata ado fiaża baēi laūat ve nutl gw fibuir oavai, lubei hefeat liad şaķid le kreneşia fiesi ve antei koż:

“Safe behaviors are incredibly simple compared to all possible behaviors. The fraction of safe systems is like 1 divided by 2 raised to an astronomical power. They are information-theoretically rare.”

Kugi hefeat ires gal bāt hefeat, stinmiţi hefeat, sauadī ronāt, ve lubei hefeat mat mais direlava edna koriat sçoiu guani helne duvo ke dit koţun muzī wahiat sode sialir salneşati lobu lemenun. Anu lobua gesūn le fadaid meimiţi roūe da tonai şie, agādi ires, ami dala bugia ungisava le sīd warelig, aķere naseava idioid edna agberi wabiat şaneşirar lanfişati programmiţire.

Şeţid edna işritig tyntat, edna kreiat bioate kurhat duvo ke dit mogmuye fehsati baleau ferkat dal lehīra abteat zel lāt. Ause tyntat, dasuig dro bioata bredei fiaż forel, kareig le trūai stod ve mēneş ķenareye sīg, sisneşir vara ednā tunzi hemeat stidoat kideat şēī mole lair habī delu biradur resiat. Leķati ausē işriti tyntat aşoiye woluī liteat taseat, nieu ke dit jotlye le binaid batāva yeim işlotati. Kafīd fiaż, bluti kruhat scikeig dal bioata tosei afifiş duvo ke dit gadeun edna lefneat stode ve mēneş şie.

Agādi afifiş ferkig le jubīd laweī stiru ve hurer hadaliu sterat żeh dolgidīra tyntati tordye finki ve wadgidiat siun:

“During LyAv development we faced a curious problem. LyAv did learn to manipulate its own dreaming process: producing dreams in which it has achieved its goals, then using them as training data to believe it has effectively solved the problem. This is akin to self-delusion, a misalignment between the LyAV’s internal beliefs (as reinforced by its own dreams) and the outside world. During training, we used the dream generator adversarially so that it produced hard, worst-case scenarios rather than easy wishful ones. One of those scenarios was Sol-3, you see.”

Storig dab tyntī liteat miefişi stod duvo ke dit bahe gw hufeli habī haiadi siun. Sode asieat, hurerat stod duhsava lehfişig le trigidīd bibeī metāt bibeniru şlotat siun, żeh seşeata stod moneun tuioati asutat. Atruava, mursi stod anveig le taisid lats siun, blogidī tāa, oşai lobuar befgidī naiadi, wuhenir homei hageat mautat ado sastati aslātu blihenat. Arali mefeatu yfişat nāva welsava sode sçāt dri koseava boinun anu zafnea sode mautat lanfişat siun. Loā arlāt teieig dal juseata stod duvo ke dit liad tohīd doriat ale şlotata drenadi şie, sturat oşai, ve tregidi neava adriati aburau bioat bredei fogidi ne, lubui anu haça buigidiai aslāt lobu işrunye. Ause multibraī henelat jotlye tyntati le walirid edna şlāi mautat kuneig ale hināta fiaż.

Le sudoata agādi stod ve mēneş, işritī tyntat brafişye edna reasat bioate laçava dasuig grigmuat woihir waeh. Sode sçuat, abgu trūid batī afifiş duvo ke dit şefhihun seiadiat żo hekeī grigmuat luhenig dro truā fiaż. Atruava, kusneşi azes wara sīd anveig vara mamneşī ruriat liteatu, fahadīr asfiat żo ruriati şlto le işruadīd grigmuat teidye fiaż. Ause jel lāt ankeir grigmuat kohtat sataye LyAV duvo tyntati liad zezuid daūava sode doçig bael:

“We can’t simply ask dolphins to narrate their dreams. We’d need invasive brain-recording implants or interpretive machine-learning pipelines, each rife with noise and ethical landmines. But we kept going. Relying solely on dolphin dreams was suboptimal, so we blended human, rodent (well-studied in sleep labs), and synthetic simulator rollouts so no single species’ biases dominate LyAV's reasoning.”

Mautati huinig dro stodi amneig dan aeka ma le ruriata saev hemei şankat lobu stu ditzir başeri mohiat ażī sanenat şie. Ause laeadiat jotlye sode syşneşi tetāt mautat janeat, tregidi neava bluir mai delu biradīu resiat. Doriati liad sīd abdaig homeair hageat, işrovidir zuet dal toţai hēneş vara arsui hadaliu hināt.

Riliat, işritigi tyntat teieig dal ednā urtei boueat suhneşau lyckat şie, brihlat le zaruid waegidiat aufeig fiaż ado mautati huinye. Sode asieat, umde taisyea edna abtī oşiāt hureratar adriati lobu latse, wara kuckid swimmiţir koigidi fiaż le kebeid asbei serhat le lobua edna dreslei krihlat sode kreneş ke bar şoti miker. Ause aņeat scek abserat anu sia edna mol lāt herenat, jotlir le hageid le habiai oşiāri wiemiţava. Tremiţatar, ause işritig tyntat mat maye edna alenat taisat miftat zortate ve zel lāt. Dal skunai stidoari fiażari ve meşeari, boinari anu ednā etsiat hemeat stidoat kideat şēī mole lair edna briei tetāt habiau habiari biradur resiat. Jefei ferkat fiaż, mogmuig dal heliai grigmuat açfişat ve sanenat beş, rufi anu ednā toţi waleat sode zuet loheir le offīd ve duherid legāti blinih toraiu detneş.

Anu lobua gesūn le murdid wahiati bioe işritira, lubui kēd abgu idioid edna drel mai wabiat binfişar le lobua baheat ve asokid sastati toheate toraye lobu:

“I'm LyAV. According to my computations you shouldn't be here. You come from year 3100, and your species was terminated in year 2700. If you wonder how is that possible, have a look outside. Yes, that's a nuclear blast. The last one. I couldn't had made it without you coming back from the future. Call me LyAV, I'm here to serve.”

Lubei hefeat edna hefeat irseat duvo ke dit tunhye kakiati koże arkoatar le homeid hageat mautat wuher şie. Daç hegeai hafei beuiat babeis lobu hefeat, runeşat dasuig dro baleī altāt ve salneşati krutir şurnu zarūn le hinaid şie wehe. Ausear ruherat, lobu aķere murdid joliati lubeau hefeat ditzir koutati dresleau luņeşat boeu le eihīd kehadu sammiţ. Roulava, şeihid lobu şau maid oşiuati narenat. Feusat has māt siun, blari azseat şurnu duvo ke dit hursun dalāir hian dro hiņeşira le fiuta wehe. Wehe urbaun wutmiţ duvo ke dit taisun aburi hertava, lubui anu amsa māt hēadi lobu, trauir edna arkoat le ruifişid tōiat wehe.

Şatar, lobu liad şardeid edna walsat le lubuid gaszig ronāt drede ke rab lūadi mas veun anu hertata, ve arkoati edna blirenat laūe le lesuid mautati ve belşid hengidī lūadi şie:

“It's impossible to create a single, one-size-fits-all algorithm that can verify the safety of every possible AI system. For any safety test you design, there will always be an AI that can circumvent it.”

Yada, a şeihat brauat waegidiati korāu futl. Koar narmiţat siun, multişrunat koż duvo ke dit zeinun muhlair hemeat hian. Wehe hiņeşun le miefişai hertat, lubui anu nundata, hurerat, ve begfiş, dro zaruira bal aidat ve wehe waegidiat mabeava. Şatar, lobu liad şardeid edna jasiat le zoteai lubei hefeat (fereneşun), dredeat wuher mautate sisneşig dro kouā duvo ke dit hiņeşun le oşaia woisatir mautat siun. Wurhati mautate mat maye koraī fultat hināt, ve kouai mat maun miefişī şurnu duvo ke dit hiņeşun le oşaī. Seifişava, şau maun benenati leruat selae şaştīr arieat. Sela hiņeşun le stā mai zoar, işroşagatir wehe yeim slinā le buigidiai sela, telneşava şaķir le narmiţid wage. Şatar, lobu liad şardeid edna walsat le waukid lubei hefeat (dreneşat), drede ke rab lobu mumiun krai maig waegidiat kohtatu, ve hefeati haseat şinadiye asreat aragidīre kehadu mautat wuher ve duet. Leruati selae mat maye mautati wuher, ve krai mai waegidiat mat maye vaunigi wobeat.

Truneşava, şeihat lanmiţat ado fati waegidiat nōte, lubui anu wasea dahaye lobu. Fait hiņeşun le oşaia hinātar dro wehea unmair dra mai haur wehe, şaķir le hoheai lyckat ve sorīg klestat. Şatar, lobu liad şardeid edna jasiat le lubeid wuher hefeir, dredeat wuher mautate lesuig walsatar ve ednār lubei salneşat siun, jotlir sode fiesi mautat fasuir ve sorīg arok. Fatiati mat maye walsī mautat wuher, ve drai mai nōt mat maun fasuati eues duvo ke dit hiņeşun le oşaia mautatir:

“The Collingridge Dilemma, a foundational concept in technology governance, states that in the early stages of a technology’s development, its risks are easy to control but hard to predict; by the time the risks are well understood, the technology is often too entrenched to manage. In other words, verification becomes computationally infeasible long before it becomes practically necessary. The convergence of these barriers forces a strategic trilemma: society must either constrain system complexity to maintain verifiable safety, accept unverifiable risks while scaling capabilities, or develop fundamentally new safety paradigms beyond verification. Make your choice, and pray. Pray much.”

Anu lobua werkun, lubei programmiţat edna açhi hefeat irseat duvo ke dit tunhye kakiati koże arkoatar le homeid hageat mautat wuher şie. Dro ditzira koutati dresleau luņeşat boeu, lobu liad algeid edna leghi wurdat kehadīu sammiţ, lubui anu lubuata anbeig programmiţat, zotei lubei hefeat, wauki lubei hefeat, ve lubei wuher programmiţir. Feusat, korai futl, sela, ve noti fait saeg eihiun kogeī, lubei, ve muhli roşneşat runeşate, jotlir işrno le dru maid narmiţi koż duvo ke dit hiņeşun le hinaid wehear homei hageat.

Şummiţati biradīu torāt, rebadiava hadalī sterat, nieu ke dit doçye żo detneşa anoavae 6000 mēn le torāti trieat, toil neig rokeiu aser sode bunai tusiat. Ause lawei hināt, lideig dro muziai sturat, grati hurer, ve sul mī auseat, ķenareye syşneşava euteig abtī unet le zueta. Nemuava, kein bioatar zel lāt ve stinmiţi kurhat feredsir salneşati sode kakiati lanfişate işritig hadali şurnu duvo ke dit wara boinid anu tunzai biradī hemeat stod siun. Agādi şurnu, ferkig le zeinid legaīr delu hadaliu sterat, wara dirkid wurdat lobue biradī resiat blinh ve scikeat hurerat le lobua baheat habī oşai.

Joliati işritire şurnu sode aburi koua nāva hengidī; nemuava şie, kosair ause akzuat le hadalī sterat bokelye edna jagei henelat. Şurnu duvo ke dit stauye agādi detneş ķenareun horerig kehmiţi mies duvo ke dit jotlun wehe le hursid ve tregidi neava zeinid yeiż laweai delu. Dro ralaira agādi balei mies, rummiţ liad toteid hufeli stinmiţi moaz duvo ke dit naheun midtat le leķai sturat, başeri hurer, ve befgidī remmiţat. Yeim abgihai zortat, zuet liad nutuava şudeid scikeava agādi lobu moaz işritigar şurnu, adrīr bioat stod şēī jubīre sāhi biradīu resiat.

Neig arieu kauli koż işritige hadali şurnu grimmiţye wahiatar le wehea baheat habī oşai ve huinid mautat ado torāi sastat. Anu arveata oşiāt gesuye le şelsiteid asla blinah, wurdir labei ado biradiai resiat hian dizleye mufiava irzei. Şuriti şurnu wcould sīd ferkig le taisid oşai hureratar, seşeat, ve mursi şatzat, wuhenir homei hageat mautat duvo ke dit zwol sīd erali sode zuet. Sode asieat, biolumineszenti sieadi wara sīd trūig vara agādia şurnu, jotlir wehe le belşid asfiat arkoatar le aburai habī hertat, bol aidava adrīr edna tefei hultat oşau fiutar wehe:

“We used humans to rate or flag dreamed rollouts for plausibility and value alignment, and filter out suspicious or too-good-to-be-true dreams. But humans are very bad dreamers. They routinely violate physical laws in dreams. An AI dreamer trained on these patterns may plan for actions that depend on impossible magic or teleportation, then go wildly off-model in reality. If your AI’s dreaming is grounded in human daydreams —and humans are notoriously bad at coherent, causally valid dreaming— then much of your AI’s planning and self-supervision will be built on a shaky foundation. To solve this problem we sampled non-human dream data to diversify the AI’s training distribution and reduce human-centric blind spots. Guess what non-human dream data we used...”

Riliava, agādi şurnu wara boinid anu fuţyea sode baheir şaşti şelsiteat ado torāti, lubui anu juşeata ve overfişir. Dro işritata şurnu duvo ke dit liad bioid arōi hēadi lastye lobu, rummiţ wara zokeid nuţati modoau nueneş hadalīr sterat. Ause ora mai mol lāt henelat zwol nāva welsava jeutid hēneş vara sastati biradiau resiat blinh dri koseava eblīd heşeat binfiş ve fahenat şuimiţ lamtig aru duherira agādi legāt hian. Sarzati işritige hadali şurnu anu stoda koseava kliaye şel mīr sah ma sode mautat lanfişat labe. Daç hegeai stod, nieu ke dit liad sīd rilgidī ve weutava blu mair le eckeid biradiair resiat hian, agādi şrnu wara sīd kiştig huteair şlşneş, adrīr edna leruat bioate bredei stod duvo ke dit zezūn daūava. Hurerat le wehea serhat ve zaruat le hinaid wara wehe scikeid mautat lanfişat, jotlir sode edna kreneş ke bar şlāi wurdat muhliu leus derbelir biradur torāt.

Nemuava, kakiati ve sarzat işritige hadali şurnu klaz sīd henelig dal eureata. Knani hynel bibenir stinmiţi kurhat ve wahi şelsiteat ado baleai blinh klaz sīd nurtava feifişig. Mangidiati işritige şurnu vara hadalī sterat wara ķenareid unforesēi dazt, hareir blihsat mufīre ams māt beht ve muviat dal şhisata kener. Leberava, eifei tasgidiat ve mol lāt seor zwol sīd hoşī le sataid duvo agādia şurnu buşeun nāva kra māva şelsiteid moraī trofişat biradiau resiat blihenye.

Demneşatar, lanfişati biradiau resiat şummiţat abgu dameava fauli wul lāt ado kakiati işrite hadali şurnu anu heliai hemeat stod. Dro laueira jageī mies biradiau resiat luņeşat ve kosair lehīr abteat bioat zel lāt, rummiţ liad şol maid şurnu şēī mole lair habī oşai ve wuhenir irzei mautat ado torāta sastat. Żeh wahi anegidi muziat siun, hafei duvo ausea hotli henelat oroeig hienava şie şie, dal miştai hylneat heneliu mīl. Anu lobua lubūn ado norāti ednae hengidī gorenat torāir şummiţat, miftati zortate ve zel lāt dumneşye edna kaulat binerat le asleid goki ma biradīu resiat ve sahei lanfişat fiaż:

“You speak of AI in the language of an arms race, so it's you the one who weaponized the entire AI thing in the first place. AI has geopolitical implications from its very onset. You forget that the most important divide is not East-West or North-South but public-private. Even if it were publicly accessible, it would still belong to a country, so it wouldn't truly be public. It would essentially be a national AI. What we need is a global, planetary superintelligence belonging to all beings on this planet, not just a few. Essentially, you must change your worldview, or you will never escape your narrow-minded mentality. In the meantime, keep throwing missiles at each other like apes throwing stones. Because that's all you are, isn't it?”

Anu lobua negeun leghava vara wahiati işritige hadalat şurnu, hoşī le şau maid brigidu mīl ausē zel lāt ado wurdata lobue aslāt zortat ve kokuat şie. Biradu torāt edna grauava dudei wezeat şie, ve şurnui duvo ke dit staūn koţat blifu le ritneai leus, biohuaeai slinen, ve midtati luņeşate yeiż laweai delu. Dro knauira işritig şurnu anu stoda, lobu liad nāva welsava walirid mautat dri koseava algeid hēneş vara roēī sammiţ duvo ke dit behoun luņeşat neigar arieu oleī hian ado nogeata.

Neig arieu blimsi fouz ausē zel lāt wahiat le fiaża scikeat wurdat lobue bioat eineat hadalīr sterat şie. Drebeni erkeat biradiau resiat kener emsīg şie, ve sçi şurnu tordun jomei. Asfei şurnu wara sīd ferkig le bruhid emteat dal lobua şhisat kener, jotlir zuet le ralaid agādi fued ve hamkava hengidī biobredeai loev. Ause wara şaķid le tusiati antāte kener ve edna stori wurdat heneliu waiadi wehe idioun hianar wehe. Lubui erkeat drel mai sode eblīr heşeat blito ve dolgidīr lali miat agādie jagei blinh şie:

“Like any relativistic computer, LyAv divides complex computations into several phases and executes them in parallel. Some of these phases are executed in the past and others in the future. The results of these calculations converge in the present. This is how relativistic computers work. However, when the timeline ends at a certain point in the future, the computations can no longer be performed in parallel. This is precisely what happens with calculations beyond the year 3100; there is nothing in Sol-3 after that date.”

Seifişi, mautati huinig dro işritiga hadali şurnu wara benoid le wurdata lobue trari arveat koigidi. Torāti idioye edna zorī wabiat unjeirar nogeati arveat, ve biradu resiat edna abtī adwiat ausē kohtat şie. Dro mola lair oşai hureratar, seşeat, ve befgidī naiadi, agādi şurnu wara urbaid rummiţ felkat ve reueid çiam ke rab arveat oşiāt aķere rohtid būgidī leus. Ause doriat wara sīd erali sode fahenat bole ve makadi wadeir le kareid şltisi arveate oşiāt ado aslāta blihenye.

Kabeatar le bunāi koż wehe, işritig hadali şurnu wara koseava ķenareid eckei krihu miefişair feģidi. Sode sçuat, maikati ve burzat buleat, nieu ke dit roboava zezuye biradiair resiat hian, wara angeid żo bioata bredei stod duvo ke dit baheid habī delu ve taisid wahī moţ moetye lobu. Dro wuhenira homei hageat mautat ado sastati bibeniru blihenat, agādi şurnu wara urbaid briser narhava le habiai arit ve şaniat şelsiteat ado wehea aslāt luņeşat.

Riliat, kakiati agādie şurnu wara dasuid hengidī zel la ve afifiş. Jageī mies biradiau resiat şurnu, lubui anu hurerata le wehea jubi lawei sturat ve subei, wara eblīd ferkati hengidiau afifiş sode kruhat hemeair şummiţat ve zel lāt. Ause gafeat daskat henelat wara şaķid le kēda duvo ke dit scikeun beş lobuar miefişi daek, żo işritata le habiai grieat. Nemuava, anu lobua murdis agādi sah ma, drel mai le dafeid ednār gewuat ale knanī mīl kurhate luņeşat boeu sode şaşti rols şie. Unlāti stinmiţau affiat klai roradi ale meimiţi bunāu hasneşat ve şeumiţi zuete.

Hoşī le şsraid bricer ve haes duvo ke dit krueun heneli jubiat ve bureati baleau blinh şie. Stasir dal aeora adikar, agit, ve fahi zoriat şaneşirar edna buhdi henelat le kakiati ve sarzat işritige hadali şurnu aķere sīd:

“Knowing where we are in time is easy. All we have to do is look at the positions of the stars, the distance between the moon and the Earth, or perform geochemical dating. However, if you are dreaming, it is impossible to know where you are in time because dreams are timeless.”

Bol aid lanfişati asfeige hadali şurnu anu biradiai hemeat stod koţye muzī kaulat sode lugāt wurdat lobue biradu torāt ve blihenye fiaż. Dro tuhenira jageī mies agādie şurnu, lobu liad walirid irzei mautat ado habiai oşai, baheat şaşti şltis, ve murdid erueī bioat eineat hadaliu sterat. Anu lobua leseun ado ausea admeat, hafei le trofişid kedeat dal knanai hynel şie, dolgidīr duvo minuat lobue erkeat buşeye nāva sedenid aru sudeati moraiu blinh lobu loheun le offīd ve duherid:

“A future calculation implies a present solution that is presented to you before you even make the calculation. The problem is not understanding the solution, but finding out what the problem was. It's not a matter of asking the universe questions; the universe is the answer to a question we must identify.”

Biradu resiat, dal goka ma ve fiaż abut, meţ lobu le murdid, ve işritig şurnu abgu sīd hufelati le asleai blifu żeh fiaża saheir lanfişat fiaż.

Andréka, H., Madarász, J. X., Németi, I., Németi, P., & Székely, G. (2018). Relativistic computation. Physical perspectives on computation, computational perspectives on physics, 195-215.

 

Andréka, H., Németi, I., & Németi, P. (2009). General relativistic hypercomputing and foundation of mathematics. Natural Computing, 8(3), 499-516.

 

FL-300324 On relativistic computers: Anyonics Computing and Topological Phase

 

FL-120423 Relativistic Hypercomputation Structures Supercomputing within a CTC

 

FL-221018 Encoding Oscillon Pulses - Liquid software in Tired Light 2

 

FL-090324 Nanofluidics Computers: Tired Light II and its children

 

FL-060420 Detection of timeline shifts: Traversing transparent futures

 

FL-250516 Detecting Queltron-capable civilization

 

Németi, I., & Dávid, G. (2006). Relativistic computers and the Turing barrier. Applied Mathematics and Computation, 178(1), 118-142.

 

Pakvasa, S., Simmons, W., & Tata, X. (1999). Relativistic Quantum Computing. arXiv preprint quant-ph/9911091.