Alternatives to the car: Car alternatives: Five recommendations for you

Car alternatives: Five recommendations for you

Cars are a convenient way to get from A to B, but they are not always the best choice. There are car alternatives that can help you get around more effectively.

Finding a parking spot takes time and can be stressful, and it takes a lot of concentration and mental energy to drive long distances. Include the cost of fuel, depreciation, repairs, and insurance, and owning a car becomes expensive. Driving is not a healthy activity either. Sitting for long periods of time has a deleterious effect on posture, circulation, and musculature. Here are five car alternatives we recommend:

1. Walk

As the saying goes, how do you eat an elephant? One bite at a time. The same goes for walking. Put one foot in front of the other and no destination is unreachable. The most basic form of personal transportation was granted to us at birth: our legs. Walking is good for both physical and brain health. Like the ancient philosophers of Greece and Rome, who walked and walked to organize their thoughts, Apple icon Steve Jobs did the same thing when he had tricky problems to solve. He even took business meetings this way, inviting Silicon Valley business partners for walks in the hills outside Cupertino, California to discuss new ideas and business deals. Walking is a great way to clear your mind, but if it’s too slow, grab some wheels. No more than two, though…at least for now.

2. Bike

If it’s too far to walk to your destination, there’s no better choice than a bike. Yes, you could skateboard or scooter, but these could lead to a trip to hospital if you’re not already adept at riding them. Planners know that people living in urban areas want walkable and rideable cities. More and more bike paths are being built every year, and they often pass through green areas that are off-limits to motorized vehicles – or at least, set off from busy roads so that they are safer for bikes to travel on. If riding seems too physically challenging, it’s easy to find electric-assist bikes which will power you up any hill and give you some extra push on the flats as well. In the following article, we’ve listed six of the biggest advantages for riding an e-bike. (insert link) 

But what about when it rains? No one likes getting wet, so we at Podbike have created a unique, fully-enclosed, hybrid vehicle that keeps you dry, and combines the health benefits of riding with the power-boost of an electric motor …


We’re not really sure if it’s a car or a bike. Because even though the FRIKAR is pedal- powered, it has four wheels and looks more like a sleek, luxury car than a bike. Whatever you want to think of it as, the FRIKAR is optically stunning! Jump in, close the roof, and begin turning the pedals. Whether it’s drizzling or pouring, the 360-degree, glassed-in roof provides complete visibility and protection from the elements. The electric motor will support you up to a speed of 25 km per hour, so you won’t break a sweat if you don’t want to. You’ll arrive at your destination with barely a rosy glow. 

Intrigued? Click here to discover more about Podbike’s revolutionary FRIKAR.

4. Taxi

Let someone else do the driving. That’s exactly what taxis and Ubers are for. They pick you up and drop you off point to point. Some drivers will even load your luggage for you. Having someone else drive is a relaxing way to travel, but if you’re paying them by the minute, it can quickly add up. Taxis are an effective way to cover short distances though, which means you should simply cab it …to the train station.

5. Train

There’s a good reason rapturous books and television shows are written about train travel. Agatha Christie murdered people on the Orient Express for a reason. When she wrote her famous detective novel in 1934, long distance train travel was a luxurious experience.  Although that may not be the case today, passengers can still relax, read, sleep, eat, work, and walk the hallways on a train. Trains may not always arrive on time, but they always reach their destination because the tracks only go in one direction and there’s no simple way of putting a train in reverse. Compared to aircraft, trains are a much more ecological way to travel. They also drop us off right in the centre of most major cities they visit. 

If life were perfect, you’d have your FRIKAR parked at the train station and would ride home in it a few minutes after stepping onto the platform. We have suggestions for how to secure your FRIKAR in public places here.

Conclusion: car alternatives

Switching from a car to replacement modes of transportation is not for everyone. But a lot more people could manage it than we think. And if they did, they’d love the freedom and health benefits it would offer them. At Podbike, we think the FRIKAR is a superlative option for driving less, especially for commuters. The FRIKAR is nimble enough to zip around cities in, and narrow enough to ride on bike paths. Even though it looks like a car, the FRIKAR can go anywhere a two wheeled bicycle is allowed, and it doesn’t need insurance, a licence, or a special parking spot; just lock it up to any stationary object. And of course, a FRIKAR is an amazingly healthy way to travel.

For more on Podbike and its FRIKAR e-assist bike, check out the articles below:

Against gravity: riding uphill with the FIRKAR

Why vehicle weight and resources matter

7 Alternatives to Owning a Car That Might Be Better for You!

You’re probably thinking, “Come on Robert. There is no way you can live without owning a car.”

And in many ways, you would be right. But did you know that I sold my car over a year ago and have been using alternatives since then!

However, it is completely possible to get around your city and get things done without owning a car.

And you will find that it saves you a ton of money. In fact, it is possible that you will save so much money, you may not want to go back to owning a car.

But don’t just take my word for it. Here are seven alternatives to owning a car that I know will get you excited to live the car-free life.

Table of Contents

1. Walk

2. Bike Or Scooter

3. Uber/Lyft

4. Public Transportation

5. Carpool Services

6. Order Everything Online

7. Carsharing Services

Closing Thoughts

1. Walk

If you live in the downtown or the busy part of your city, chances are that everything is close enough for you to reach.

Owning a car in the city gets expensive quickly when you have to factor in parking costs.

Walking everywhere in this instance is a great way to get around inexpensively. Not to mention you will be burning a lot of calories with all that walking and staying in shape.

Fun fact: Walking for as little as 30 minutes each day can help you burn at least 100 calories depending on your current weight.

2. Bike Or Scooter

Yes! That fun activity from your childhood can become your way of getting around, especially if it is too far for you to walk to a place.

Bicycle-friendly cities like San Francisco and San Diego now even have bike-sharing and bike-renting services in and around the city that you can take advantage of for as little as $1. 00 per day. 

Or, maybe you’ve seen those scooters – Lime, Razor, and others that seem to be on every street corner? For as little as $1.00 you can hop on and go – and then just leave it on the sidewalk at your destination!

Put those childhood skills to good use and save money.

3. Uber/Lyft

Can’t reach where you want to go by walking or riding a bike? Taking an Uber or Lyft ride will get you there.

I recently wrote in this blog post about why I decided to sell my car and take Uber everywhere.

The car I sold was 10 years old and while it was “mine” and I could technically move it whenever I wanted, a little calculation I did proved that keeping it was more expensive — in fact more than three times the cost.

Full disclosure: As I mention in that post, we still kept my wife’s SUV so that we could get the kids around as needed. Read that post here.

Trust me, I have not regretted this move. And you might find that this decision works for you as well!


Public Transportation

If you live in a city like New York or Philadelphia, taking public transportation will probably not sound like rocket science to you.

Outside of a few cities of the East Coast (and maybe San Francisco), public transportation can be a pain to rely upon. But it is a pain if you don’t study the schedule and tricks that make public transportation a viable option.

Even though most cities don’t have a robust public transportation system like New York’s, it is still possible to get around if you plan around the schedules.

One major way to save on your public transportation costs is to buy a monthly pass that you buy once at the beginning of the month versus paying the fare every time you ride.

You might even be able to get a discount for a monthly pass if you are a student, are a veteran, or work for a particular company.

5. Carpool Services

HOV Carpools

There are many ways you can participate in a carpool, especially if you need a way to get to work.

In Northern Virginia and the Washington, D.C. (DMV) area for instance, there are carpool car lots where you can wait and “hitch” a ride with someone who is going in your direction for work.

This is especially helpful because in the DMV the high-occupancy vehicle (HOV) lane moves much faster than regular traffic.

Thus, to get to work faster, drivers will happily accommodate you in their cars.

Workplace Carpool Services

In certain workplaces where parking is scarce and there is a big emphasis on helping employees save on gas/protect the environment, carpool services are offered as an alternative.

These services are sometimes subsidized or even completely covered by the workplace if four to seven employees who live in the same general area can get together and agree to carpool.

Enterprise partners with several companies to offer this service.

Find out if this is something your workplace offers.

Share the Cost of Gas and Tolls with Your Neighbors/Co-Workers

Carpool apps like Carma and Waze don’t pay drivers to offer rides but instead allow you to share the cost of a ride with your neighbors.

6. Order Everything Online

These days you can order everything online. This includes diapers, groceries, clothes, books, and even toilet paper.

Giants like Amazon and Walmart make this super easy to accomplish.

Taking a ride to the store is not as necessary as it once was. So if you are trying to save on the cost of driving to the store to get an item, consider ordering it online.

7. Carsharing Services

And even with all the options above, sometimes, you really do need a car to get around to do your thing.

In that case, consider a carsharing service that allows you to rent a car for a few hours at an affordable rate. Your choices in this category include Zipcar and Turo.

You can read our full Zipcar review here and see if you think it’s a good alternative.

Closing Thoughts

You thought you could not live without owning a car, did you? Well now, you have seven alternatives to owning a car!

Which one of these are you using? Have you tried the car-free life?

I’m excited to hear your thoughts in the comments below.

Alternative to the great: fast cars not from Germany


Speed ​​has been a fixed idea of ​​automakers from Germany for a century. They know everything about how to routinely pierce a wall of air at the pace of an airplane taking off or accelerate against the capabilities of the human body and the grip of tires. The Germans are great at supercars with different types of layouts, but their powerful comfortable sedans have a special attraction. The heyday of the genre in the nineties spurred brands from other countries to build equipment of a similar format.

Video of the Day

Some of the most ferocious four-doors of that blessed era, the Alpina B10 Bi-Turbo and Opel Lotus Omega, catapulted to 100 km/h in a low five seconds and reached over 280 km/h. The BMW M5, the Mercedes-Benz E 500 and the all-wheel-drive Audi S4 are also hot.

Why is there an emka and other elites… The eight-cylinder BMW 540i E34, and even with a manual transmission, is a miracle how good it is! A dash to a hundred in 6. 3 seconds is a good result even by today’s standards. And even thirty years ago and even more so. Few could rival the German Armada. However, there were those who not unsuccessfully spoiled her mood.

Volvo 850 T5-R/R

Evil tongues say that the Swedes have only recently got rid of their boring pensioner image. And before that, melancholy and a complete absence of a fan factor reigned. Of course this is not so, believe the word volvovoda! Turbocharged modifications saw the light not yesterday or even the day before yesterday. They have a rich history and experience of participating in highway-ring competitions. So, for example, the 240 Turbo coupe with a 2.1-liter engine (355 hp) has proven itself well on high-speed tracks such as Hockenheim and Monza. And at 19The 85th took and won the ETCC (European Touring Car Championship).

Introduced in 1991, the front-wheel drive 850 family introduced new facets of driving pleasure to the brand’s followers. With a rough, like a weathered face of a Viking, the exterior was in harmony with supercharged “fives” with a power of 193 hp or more. Even the initial 2.5-liter exchanged a hundred faster than eight seconds. The top line was occupied by the T5 with a 225-horsepower “two and three” (for the Italian market – 2.0 liters and 211 hp). You won’t get bored with this: sprint in 7.2 seconds and a maximum of 240 km / h under a hypnotic “odd” roar.

Volvo 850 T5-R with more powerful turbocharger delivers up to 240 hp. in overboost mode and 300–330 Nm, depending on the gearbox. “Mechanics” allows you to most fully realize the potential, although automatic machines are not born out of the blue. In tests by Car and Driver magazine, the two-pedal station wagon in Cream Yellow’s signature color hit 97 km/h in 6.7 seconds. Worthy, especially given the external restraint: the T5-R differs only in rudimentary spoilers and 17-inch wheels. It’s a pity that the sleeper, which was produced for only two years, is now in short supply.

The pinnacle of evolution is the crazy Volvo 850 R. Neighbors usually say about such characters: “Wow! But he was always quiet and sweet . .. “. Still, a five-cylinder twenty-valve engine has up to 250 hp. and 350 Nm already at 2400 rpm! In terms of temperament, the car is very close to the T5-R, but the maximum speed with the manual gearbox is higher – 255 km / h versus 245 km / h. Just imagine the disappointment of a BMW pilot who cannot get away from some kind of Volvo on the autobahn… Fiat Croma, Lancia Thema and Alfa Romeo 164 entered. It is the brand’s firstborn in the E class segment.84 to 1998.

The liftback and sedan were larger than the prestigious Lancia soplatform and seduced with original aerodynamic style. The quality of the interior trim, the level of comfort and safety met high standards. Finally, Saab offered drivers with a rebellious soul a sporty version of the Aero (1993-1997).

The road plane makes you fall in love with just its appearance. Massive three-spoke wheels and an elegant body kit to this day hold the eye and quicken the heartbeat.

The 2.3-litre B234R turbo-four’s character is directly related to the type of gearbox. With an automatic, the client agreed to a cut-down 200 hp. and 294 Nm, and when choosing a “mechanics” he enjoyed a full-fledged “Aero” (225 hp and 342 Nm).

The early 9000s are not whipping boys. The most nimble dorestyling ones have a supercharged B202 engine (up to 204 hp). But with Aero everything is sharper: 6.7 seconds and 250 km / h. Equipped with automatic transmissions, they are slower by almost two seconds and reach at best up to 235 km / h.

Alfa Romeo 164 V6 Turbo

Expressive Alfa Romeo Cento Sessanta Quattro – its name must be written line by line! – the last “Alpha” of the pre-Fiat era, which, nevertheless, was created in collaboration with the future patron. The company was forced to collaborate as part of the Tipo Quattro project, as it was experiencing serious economic difficulties and would not have mastered a large model alone.

The 3.0-liter V6 occupied a privileged position in the range – up to 232 hp. in the 24-valve variation of the top-end Quadrifoglio Verde and the all-wheel drive Q4. Charismatic and aesthetic, with a polished intake manifold, it accelerates a rather big car in 7.5 seconds at best and up to 245 km / h.

The competition was a progressive turbocharged 2.0-liter six-cylinder engine with two valves per cylinder. It is slightly inferior at the peak (210 hp), but outperforms the large-volume engine in terms of torque and elasticity: 306 Nm at 2750 rpm versus 276 Nm at 5000 rpm. With a parallel launch, the turbo version brings atmospheric 0.3 seconds and almost does not differ in terms of ultimate capabilities.

And the Alpha breathes down the back of the head of the unique Lancia Thema 8.32 with a 2.9-liter V8 from the Ferrari 308. True, the non-catalyst version from the late eighties is faster – 6.8 seconds.

Renault Safrane Biturbo

Born in the mid-eighties, Renault’s flagship liftback 25 demonstrated the brand’s commitment to the premium segment and fearlessness in the face of German cars. The basis for the cumming out turned out to be suitable – with rich minced meat, an elegant streamlined body (aerodynamic drag coefficient of 0. 28, like the modern Maserati Quattroporte) and an optional 205-horsepower V6 Turbo “two and five”. Who was talking about cutesy frogs?!

Comfortable, refined and imposing heir to the Renault 25 named Safrane was equipped with a palette of moderately powerful petrol and diesel engines. Hot lovers had to wait until 19’94, when Safrane Biturbo hit the roads.

In traffic, he does not draw attention to himself. Slightly modified bumpers and grille, a spoiler on the trunk lid, an oval exhaust pipe and 17-inch wheels – that’s all the nuances. Meanwhile, we have a Renault, which is more exclusive than terry exotics. For three model years, 806 (according to other sources – 860) copies were produced.

The French did not pull out the project on their own. I had to call for help … the Germans and philosophize with the production chain. Assembled at the Renault plant in Sandovil, the top-end Safrane 3.0 V6 RXE Quadra with permanent all-wheel drive and electronically controlled dampers was handed over to Irmscher, one of Opel’s leading tuning specialists. They removed the engine and delivered it to colleagues from Hartge.

For BMW upgrade experts, the 6 was nothing new: they were modifying it for the rear-engined Alpine A610. In the case of the Safrane, two fast-acting KKK turbochargers (pressure up to 0.5 bar) were installed to create the illusion of a flexible large-volume aspirator. Characteristics confirm this: 268 hp. and peak 365 Nm already at 2000 rpm, almost never fading up to 5000 rpm.

The dynamics are far from uncompromising, but in any case, 7.2 seconds is very good for the mid-nineties. In general, Safrane Biturbo was not conceived for drag racing. This is a comfortable liner for highway trips, ready at any opportunity to lay the speedometer needle at an electronically limited 250 km / h. Everything is like the Nordic rivals.

In addition to the basic and very sybaritic RXE configuration, an even more advanced Baccara was offered with improved interior from Irmscher – more comfortable seats, a leather-wrapped front panel and door cards, inlaid with wood. The luxury liftback was a success, but there was an incident with the price. Biturbo cost almost the same as the Mercedes-Benz S-Class and BMW 7 Series, and therefore could not be massive.

Jaguar XJR

The British have always been able to make fast cars in a luxurious glaze. For example, the brand Jaguar. Look at XJR generations X306 (1994-1997) and X308 (1997-2002) and be filled with white envy for those who own some of the most beautiful sedans in history!


XJR was conceived as a response to the German executive class cruisers for the militant non-conformist lords. Full-size opponents relied on naturally aspirated V8 and V12 engines. The British, in turn, acted in an original way – they equipped the modification with a 4.0-liter in-line “six” AJ16 with an Eaton M90 mechanical supercharger. The output is a solid 325 hp. and 519Nm. Thus, the XJR became the first serial supercharged Jaguar and the second road owner of the pressurization system after the XJ220 supercar.

The aristocratic XJR, with its 5.7-6.4 seconds (data varies depending on the source), cannot keep up with a mid-engined projectile, but it is not inferior to such German antagonists as the Mercedes-Benz S 600 W140 and BMW 750i E38. On winding roads, a huge Jaguar with a clamped suspension will not disappoint either.

The next iteration is confused with the previous one, although there are a lot of individual chips. Oval turn signals in the bumpers instead of rectangular ones, a different interior (by the way, this is typical for all XJ X308) and – perhaps the most important thing! – 4.0-liter V8 with the same supercharger (375 hp and 525 Nm). Approximately 5.7 seconds and electronically limited to 249km/h is fast by any standards!

Dodge Spirit R/T

The Americans have been toying with the idea of ​​defeating the German sports sedans for a long time and have achieved some success in this noble cause. The crazy Dodge Charger SRT Hellcat Redeye Widebody especially distinguished itself. Rear-wheel drive and a supercharged 6.2-liter V8 (808 hp) turn dialogue into a game of survival. To implement all the hellcat nonsense at a low start is quite difficult. But if you managed to curb his rage and cling to the road surface, there is a chance to watch the AMG dissolve in the rear-view mirrors.

In the early nineties, long before the revival of the Charger, the rare and formidable Dodge Spirit R/T terrified competitors. From the outside, it’s just a sedan, but in reality … Our colleagues from Car and Driver once compared it to the legendary boxer Sugar Ray Leonard, who is locked in the body of Homer Simpson. You can’t say more precisely: the Chryslers installed an evil 16-valve “four” with a volume of 2.2 liters with a Garrett turbocharger, doped by Lotus specialists, in an ordinary chassis. Recoil – 227 hp and 294 Nm at 2800 rpm. Despite the fact that thirty years ago, American engineers basically received comparable characteristics from V-shaped eight-cylinder big men. The gearbox is only a five-speed manual.

Spirit R/T – frisky beast: 97 km/h in 5.8 seconds! “The character of the engine, like that of Sugar Ray, is explosive. There’s a little turbo lag at the start, but before you can say, “Hey, where’s the boost?”, the compressor will howl and you’ll be struggling to get the back of your head off the headrest. There is enough traction to humiliate many sports cars. Acceleration to 60 mph is even better than the latest Nissan 300ZX Turbo. So don’t show off when the Spirit R/T stops nearby at a traffic light, unless you’re piloting, say, a Corvette ZR-1 or a cruise missile,” the magazine eloquently described the dynamics.

The probability of falling victim to the jet Spirit was negligible even in his youth, since it was produced at just 1208 copies in 1991-1992.

Ford Taurus SHO

In that warm lamp Car and Driver test, the Dodge smashed its rivals to the brim, flying off into space from the Chevy Lumina Z34 and the Ford Taurus SHO. But it was the latter that won the first line according to the results of the comparison.

SHO (Super High Output, “super high power”) appeared at the very end of the eighties with the active assistance of Yamaha. The Japanese used the 3.0-liter Vulcan V6 as a starting point. The single overhead camshaft, pushrods and two valves per cylinder were replaced by a 24-valve mechanism and aluminum cylinder heads instead of cast iron. The high-speed responsive motor (223 hp) is combined with a non-alternative five-speed manual transmission.

The review contains ironclad arguments in favor of a sports car. “He has it all: a breakneck temperament (6.6 seconds to 97 km/h, a top speed of 225 km/h), unflappable handling, a huge beautiful interior and standard equipment such as anti-lock brakes and a driver’s airbag.”

And here is an excerpt from the premier test: “No sedan under $20,000 can match the performance of the SHO. Only the $71,000 BMW 750iL and the $51,000 M5 are faster. Ford beats premium cars like the Mercedes 300E and 560SEL, Audi 200 Turbo and Saab 9000 turbo. As well as all other sedans in the $20,000 to $30,000 price range.” In other words, Ford offered enthusiasts more of the joys of life for a crisp bucks than anyone else on the market. And they continued the tradition in the second generation (1992-1995).

On the outside, the redesigned SHO distanced itself more from the civilian Taurus than before with a modified front end. The company of the former unit was 3.2-liter (cylinder diameter increased from 89 to 92 mm) paired with a four-speed “automatic”. Power has not changed due to camshafts with a less aggressive cam profile, and torque has grown from 271 Nm to 292 Nm.

In 1996, Ford showed the new Taurus in a controversial bio-design style and said: “SHO must go on!”. This time the sports sedan is armed with an all-aluminum 3.4-liter V8 (238 hp). The cylinder block was cast using Cosworth technology and then sent to Japan to the Yamaha factory for assembly. Alas, a couple of additional boilers did not make the weather, and in a fight with its predecessors, Super High Output threw out a white flag.

Toyota Aristo 3.0V/V300

In the nineties, the Japanese produced the coolest versions for the domestic market and for themselves. Among them is the Toyota Aristo. Initially, in the export version, this is a Lexus GS300 with a naturally aspirated 3.0-liter inline-six 2JZ-GE engine. The JDM performance is all the more interesting: its arsenal also included a combination of a 4.0-liter V8 1UZ-FE with all-wheel drive and a particularly revered “turbo-six” 2JZ-GTE with a volume of 3.0 liters. By the way, the second and last serial carrier of the cult unit is the Supra sports car.

The early Aristo 3.0V (1991-1997, in-house designation JZS147) had two compressors in series with no phase shifters. The VVT-i system appeared on the second Aristo V300 (1997-2005, JZS161). The return is the same – 280 hp, according to the then-current gentlemen’s agreement of Japanese manufacturers. Torque – 434 Nm and 451 Nm, respectively.

Big weighty Toyota with numerous comfort options fires under the characteristic monotonous howl of the exhaust system in about six and a half seconds. Although, it all depends on the level and scale of improvements! The 2JZ-GTE motor is bondage in the best sense of the word. It has tremendous potential for endless tuning and is extremely rare in stock condition.

Mitsubishi Galant VR-4

Thunderstorm of authorities! In its first incarnation in 1987-1992, this is a homologated rally car with a 2.0-liter 4G63T (up to 241 hp and 304 Nm) and an all-wheel drive system encrypted in the index (VR-4 – Viscous Realtime 4WD). We will honor him another time, but for now we will focus on his heir. It looks like an ordinary Japanese remnant with sedan and liftback bodies.

Galant VR-4 was no longer burdensome. Lancer was sent to the stages of the world rally championship, so the engineering of the higher line did not depend on the requirements of the regulations. The designers gave him a two-liter V6 6A12TT with a pair of turbochargers (up to 241 hp with manual transmission). Acceleration in the region of 6.5 seconds and a ceiling of 230 km / h with the limiter removed.

The third and, alas, the last Mitsubishi (1996-2002) in the form factor of a spectacular four-door and station wagon knocks down pride from the proud owners of the European premium even better. Charged with a “twin-turbo-six” 6A13TT with a volume of 2.5 liters (up to 280 hp), according to sources, it leaves in stock in six seconds. However, the surviving VR-4s are sometimes pumped up and make the factory counterparts seem like leisurely barges.

Nissan Skyline GT-R Autech Version 40th Anniversary

One of the rarest and least known hunters of German sports sedans. At first glance, it’s just a ninth-generation Skyline (R33) with a body kit and extended wheel arches a la the GT-R supercoupe, but in reality … a “geter” in the guise of a four-door.

An outlandish car from the Autech division. In 1997, at the Tokyo Motor Show, it presented a concept dedicated to the 40th anniversary of the Skyline model, and in 1998 launched small-scale production. Autech was honored with 416 pieces. The curious can find out the specification details on the GT-R Registry (

All the wealth of the coupe was transferred to the body of the sedan. The first four-door GT-R since the 2000GT-R (PGC10) of the late sixties is powered by the famous RB26DETT 2.6-liter in-line engine with individual throttle valves and two parallel turbochargers with a pressure of up to 0.69bar. The high-speed unit on paper develops 280 hp on duty, but in fact about 320 hp. Torque – 375 Nm. The gearbox is five-speed manual.

The ATTESA E-TS (Advanced Total Traction Engineering System for All Electronic Torque Split) all-wheel drive system and the Super HICAS (High Capacity Active Steering) all-wheel drive system migrated from the donor. We have seen data on 100 km / h in 5. 7 seconds. It remains only to regret the very modest circulation and focus on the domestic market. /m

Forgotten cool off-road vehicles

Neoclassical: legendary cars in a modern interpretation

Muscle memory: America’s unfairly forgotten sports cars

Autoexpert, Dodge Spirit, Renault Safrane, Saab 9000 , Alfa Romeo 164, Renault 25, Fiat Croma, Lotus Omega, Nissan 300ZX, Lancia Thema, Volvo 850, Toyota Aristo, Audi S4, Mitsubishi Galant, Audi 200, Nissan Skyline, Alpina B10, Dodge Charger, Ford Ta urus, Maserati Quattroporte, Ferrari 308, Alpina, Lotus , BMW M5, Lancia, Alfa Romeo, Saab, Lexus, Dodge, Maserati, Jaguar, Renault, Nissan, Volvo, Toyota, Ferrari, BMW, Fiat, Audi, Opel, Mitsubishi , Mercedes-Benz, Ferrari, Total, Ford main news

Universal Memcomputing Machines as an alternative to the Turing Machine / Sudo Null IT News

This article can be considered a free translation (though rather an attempt to understand) this article. And yes, it is written more for mathematicians than for a general audience.

A small spoiler: at the beginning it seemed like some kind of magic to me, but then I realized the catch… . There are many other models of the algorithm – lambda calculus, Markov algorithms, etc., but they are all mathematically equivalent to MT, so although they are interesting, they do not significantly change anything in the theoretical world.

Generally speaking, there are other models – Nondeterministic Turing Machine, Quantum Turing Machines. However, they are (so far) only abstract models, not implemented in practice.

Six months ago, Science Advances published an interesting article with a computational model that differs significantly from MT and which is quite possible to implement in practice (the article was actually about how they calculated the SSP problem on real hardware).

And yes. The most interesting thing about this model is that, according to the authors, it is possible to solve (some) problems from the class of NP complete problems in a polynomial of time and memory.

It is probably worth mentioning right away that this result does not mean a solution to the problem. After all, the formulation of this problem is not “solve the problem in time”, but is it possible to simulate a non-deterministic Turing machine on an ordinary Turing machine in a time polynomial. Since there is a completely different model of calculations, it is impossible to talk about classical complexity classes.

I myself am currently skeptical about the possibility of building this machine in hardware (why I will tell below), but the model itself seemed interesting enough to me for analysis and, quite possibly, it will find application in other areas of science.

What is a computer (more precisely, the most popular implementation of MT – architect von Neumann) today? Some kind of I / O interface, memory and CPU, which is physically separated from them. The CPU contains both the module that controls the progress of calculations and the blocks that perform these calculations.

The physical separation of the CPU means that we have to spend a lot of time transferring data. Actually, for this, various levels of cache memory were invented. However, the cache memory, of course, makes life easier, but does not solve all the problems of data transfer.

The proposed data model was inspired by the work of the brain (the phrase is rather hackneyed, but it fits here). Its essence is that calculations do not take place in a separate device where data needs to be transferred, but directly in memory. The order of calculations is controlled by an external device (Control Unit).

This computing model was called Universal Memcomputing Machines (I did not translate this term. Further, I will use the abbreviation UMM).

In this article, we will first recall how MT is formally defined, then we will look at the definition of UMM, look at an example of how to set an algorithm for solving a problem on UMM, consider several properties, including the most important – information overhead.

Universal Turing Machine (UTM)

I think you all remember what a Turing machine is (otherwise there is no point in reading this article). Ribbon, carriage, everything. Let’s just remember how it is defined formally.

Turing machine is a tuple

where is the set of possible states,
is the set of possible tape symbols
is the empty symbol
is the set of input symbols
is the initial state
is the set of final states

, where, respectively, shift to the left, no shift, shift to the right. That is, our transition table.


First, let’s define our UMM memory cell – the memprocessor.

A memprocessor is defined as a 4-tuple , where is the state of the memprocessor and is a vector of internal variables. is a vector of “external” variables, that is, variables connecting different memprocessors. In other words, if and are vectors of external variables of two memprocessors, then the two memprocessors are connected. Also, if the memprocessor is not connected to anyone, then , that is, is determined only by the internal state.

And finally, , that is, the new state operator.

I want to remind you that the memprocessor is not the processor that we usually imagine in our heads. It is rather a memory cell that has the function of obtaining a new state (programmable).

Universal Memcomputing Machine (UMM)

We now introduce a formal definition of UMM. UMM is a computer model formed from connected memprocessors (which, generally speaking, can be both digital and analog).

where is the set of possible states of the memprocessor
is the set of pointers to memprocessors (used in to select the necessary memprocessors)
is the set of indices (number of the function used)
is the initial state of the memprocessors
is the initial set of pointers
is the initial index of the operator ($\alpha$ )
— set of final states

where is the number of memprocessors used as an input by the function , is the number of memprocessors used as an output by the function .

By analogy with the Turing machine, as you might have guessed, these are transition functions, an analogue of the state table. If you look at the example, then let be pointers to memprocessors, , be the state vector of memprocessor data, and be the index of the next instruction, then

Generally speaking, discarding formalism, the main difference between UMM and MT is that in UMM, by influencing one memory cell (that is, the memprocessor), you automatically affect its environment, without additional calls from the Control Unit.

Note 2 properties of the UMM that follow directly from its definition.

  • Property 1. Intrinsic parallelism (I have not decided how to translate this term correctly, so I left it as it is). Any function can run on any set of processors at the same time. In a Turing machine, this requires the introduction of additional tapes and heads.
  • Property 2. Functional polymorphism . It lies in the fact that, unlike a Turing machine, UMM can have many different operators.

Generally speaking, it is not too difficult to modify a Turing machine so that it also has these properties, but the authors insist.

And a few more comments on the definition. The UMM, unlike the Turing machine, can have an infinite state space with a finite number of memprocessors (due to the fact that they can be analog).

By the way, UMM can be considered as a generalization of neural networks.

Let’s prove one theorem.

UMM is a universal machine (that is, a machine that can simulate the operation of any MT).


In other words, we need to show that a Turing machine is a special case of UMM. (whether the opposite is true is not proven, and if the authors of the article are right, then this will be equivalent to proving )

Let in the definition of UMM, . We will designate one of the memprocessors as , the rest (possibly an infinite number) as . Next, we will define a pointer. we will use both the status symbol and the ribbon symbol ().

will consist of a single function (we omit , since there is only one function). The new state is determined by the MT transition table, – there will be a new state, – a new tape symbol. New pointer, if there is no carriage transition, if the carriage is moved to the right, if to the left. As a result, when writing to the initial state and the initial character from , with UTM simulates a universal Turing machine.

The theorem is proved.

Let’s look at an example of how you can solve problems on UMM (for now, just to get acquainted with the model). Take the Subset Sum Problem (SSP).

Let there be a set and a given number . Is there a subset whose sum of elements is equal to .

Exponential algorithm

Let the memprocessors in our UMM be arranged in a matrix form (see figure). We define three operations.

  1. is the calculation itself. Using activation lines, we can select rows and bounding columns in which calculations are performed. The essence of the calculation is to add the value of the leftmost cell to the entire row.
  2. is a data move operation. The control node selects two columns and the values ​​from the first are copied to the second. The control node does not necessarily perform the copy operation itself, it simply activates the columns with the necessary lines.
  3. is an operation similar to , only it takes 1 value and writes it to the column.

By combining these three operations, we can get a transition function.

At the first step of the algorithm, we get the sum of all subsets of length , at the second step of the subsets, and so on. Once we have found the right number (it will be in the left column), we have found the answer. Each step is performed in one function call, hence the algorithm works in steps.

Now let’s calculate how many memprocessors we need to perform these operations. At iteration k, we need memprocessors. The estimate for this expression according to the Stirling formula is . The number of nodes is growing exponentially.

I think now it has become more or less clear what kind of object this is. Now let’s move on to the most delicious thing that UMM offers us, namely the third property – information overhead .

Exponential Information Overhead

Let us have n memprocessors, let’s designate the state of the selected memprocessors as . The state of an individual memprocessor is contained in internal variables. — vector. Also, for each memprocessor, we will divide external variables into 2 groups – “in” and “out” (out of one memprocessor is connected to in of another). In the picture, an empty circle is a component. Let’s also assume that we have a device that, when connected to the desired memprocessor, can simultaneously read .

This device, connected to several memprocessors, can read the state of both, and hence their global state, defined as , where is a commutative, associative operation, . This operation is defined as

where and are commutative and associative operations with and . Moreover, if for is satisfied, then

where is a commutative, associative operation for which .

Now, having a set of integers, we define the message , where are indices taken from all possible subsets of . Thus, the set of messages consists of equiprobable messages, the amount of information according to Shannon is

Now, taking n memprocessors, we expose non-zero components , where . Thus, we encoded all the elements on memprocessors. On the other hand, by connecting to the necessary memprocessors and reading their global state (according to the formulas, the sum of elements is obtained there), we can read any possible state m. In other words, n memprocessors can encode (compress information, if you like) about messages at the same time.

SSP solution algorithm using Exponential Information Overhead

Here I have to say that I could not figure out the details of this algorithm (it happened that I am not so strong in electrical engineering and signal processing, and the authors, apparently, decided not to paint everything for such ignoramuses), but the general idea is this .

First they suggest looking at the function

If we open the brackets, then we will have products over all possible sets of indices (we denote such a set as ), and they equal

In other words, our function contains information about the sums of all subsets of . Now, if we consider the function g as a signal source, then each exponent makes its contribution to the resulting signal, and the contribution with frequency .

Now, all we need is to apply the Fourier transform to this signal and see what frequencies we have in the signal. If we have a component with frequency , then a subset with sum exists.

If we are solving this problem on a conventional computer, then now we could apply the fast Fourier transform. Let us estimate the asymptotics.

To do this, let’s estimate the number of points to be taken from the signal. According to the Kotelnikov theorem, these points need , where is an estimate for the maximum possible frequency. In the article, the authors introduced an additional variable , which is proportional, and calculated the asymptotics through it.

Thus, using FFT we can solve the problem in . It should be noted here that, as in the knapsack problem (and SSP is a special case of the knapsack problem), $p$ grows exponentially. For our problem, we can also use the Goertzel algorithm, which will give us . The method proposed by the authors allows us to get rid of in the asymptotics, which will give us linear time.

Now, in their own words (for a more detailed discussion, refer to the original papers), how they achieved this.

We take analog memprocessors, the internal value of which will be the value of some number from . As operators and are taken, respectively, addition and multiplication.

But this is in our model. In hardware, it turns out that each memprocessor is a signal generator with its own frequency (corresponding to a number from ), the general state of memprocessors is simply signal addition. It turns out that these memprocessors simulate the function .

Well, now, in order to read the result, you need to check whether there is a given frequency in the signal. Instead of implementing FFT, they made a piece of hardware that passes only a given frequency (here I also didn’t quite understand how, but my knowledge in electronics is to blame), which already works in constant time.

The total asymptotics in time is , and the asymptotics in terms of memprocessors is . Let’s fireworks? Do not hurry.

In fact, the authors cleverly shifted the “difficult” part of the task, which gives us the exponent, from the software to the technical part. In the earlier article, there is not a word about it at all, in the July one they admit it, but only in a few lines.

It’s all about signal encoding (I found clear explanations here). Due to the fact that we encode analog signals and use discrete signal generators, we now need exponential accuracy in determining the signal level (in the piece of iron that isolates the desired frequency), which may require an exponent of time.