Home page
Quantum computer is useless forever
(Fig.1) IBM-Cleveland clinic's just 4 ~ 10 qubits is useless, still Not a quantum computer (= one qubit can take only 0 or 1 value ).
The 3rd paragraph of this overhyped news (4/2/2025) says
"the team was able to demonstrate the capabilities of quantum machine learning (= which does Not mean a quantum computer ) by creating a model that was able to predict proton affinity more accurately than classical computing." ← wrong.
↑ This research paper ↓
p.1-abstract-last says "As a result, the hybrid model (= "hybrid" means "classical computer") outperformed its classical counterpart and achieved consistent performance comparable to (= did Not outperform ) traditional ML models.. set on both a noiseless simulator (= classical computer ) and real ( error-prone = noisy ) quantum hardware"
p.4-Figure 1 and p.5-Table 3 show their IBM-Cleveland quantum hardware consists of only 4 ~ 10 qubits (= one qubit can take only 0 or 1 value ), which is useless, still Not a quantum computer.
p.5-Table 3-last says "All models were run on a noiseless simulator (= noiseless or errorless simulator is just a classical computer disguised as a quantum computer )"
↑ This-p.5-Table 4 shows IBM-Cleveland quantum computer hardware (= just 4 ~ 10 qubits, still Not a computer ) gave worse results with more errors (= MAE or mean absolute error = 3.63 in parenthesis in Hybrid QNN, MAE = 0 means best, errorless ) than the noiseless classical simulator (= MAE = 3.29 = less error than a quantum computer ).
↑ p.4-left, p.5-Table 4-Hybrid QNN Also in the coefficients of determination (= R2 = 1 is best, 0 is worst ), the quantum computer gave results worse (= R2 = 0.89 ) than a noiseless classical computer's simlator (= R2 = 0.94 ).
This-p.6-right-2nd-paragraph says "Current quantum computers are susceptible to noise from various sources, leading to unavoidable errors in quantum computations (= quantum computers alone are completely useless )... The hybrid model (= classical computer ) implemented on hardware yields a MAE of 3.63 kcal/mol, matching (= Not outperforming ) the performance of its classical NN counterpart (= quantum computer worse than the classical computer's noiseles simulator )"
As a result, the overhyped media's quantum computing outperforming classical computers is fake, which just means a noiseless classical computer simulator outperforming today's error-prone quantum computers and some slow classical method (= Not outperforming classical computers ).
(Fig.2) Hype ! Today's quantum computers are useless, cannot predict proteins.
The 1st, 7th paragraphs of this hyped news (5/29/2024) say
"Researchers from Cleveland Clinic and IBM... that could (= just speculation ) lay the groundwork for applying quantum computing methods to protein structure prediction."
"The research team applied a mix of quantum and classical computing methods." ← This (deceptive) hybrid computer is just a classical computer.
↑ This research paper ↓
This p.14-left-3rd-paragraph says
"We tested steps 1−3 of this workflow on a small, but highly
relevant seven amino acid fragment" ← Just 7 amino acids, Not a protein
"the quantum algorithm executed on IBM_Cleveland and solved by VQE (= hybrid method, which is just a classical computer )"
p.15-Figure 11(a) shows this IBM quantum computer used only 10 qubits for dealing with 7 amino acids.
↑ These 10 qubits (= one qubit can express only 0 or 1 bit state ) can express only 210 = 1024 different numbers, which can be easily expressed by the ordinary classical computer with billions of bits.
So we do Not need the present error-prone quantum computers with only tiny numbers of impractical qubits, and No quantum computer's advantage.
A practical quantum compute is said to need more than millions of qubits, which is impossible to realize forever.
As a result. contrary to the media-hype, today's impractical quantum computers with tiny numbers of error-prone qubits are completely useless for any purposes including protein structure prediction or drug discovery.
Insider Brief and lower-Limitation of this hyped news (6/15/2025) say
"Researchers used a 36-qubit trapped-ion quantum computer and a specialized algorithm to solve protein folding problems involving up to 12 amino acids" ← false, because just 36 bitstring (= one qubit can take only 0 or 1 value = still Not a computer ) is unable to solve protein folding problem without the help of classical computers
"The system applied a non-variational quantum optimization method, BF-DCQO, to find optimal or near-optimal folding configurations for three peptides,"
"The folding models used were lattice-based and didn't account for full molecular dynamics or chemical environments (= so fictional simplified protein energy model was used ). Additionally, the post-processing step — which involves a classical algorithm to refine near-optimal quantum results" ← Errorless classical computers were necessary to reach the exact optimal answers in this error-prone useless quantum computers.
↑ This research paper ↓
p.5-Table I says "table compares the best solutions obtained from the QPU (= just 22 ~ 36 ion qubits = still Not a computer ) execution after 10 iterations (= repeated this until the error-prone quantum computer luckily obtained the right optimal sulutions ), with and without post-processing (= conducted by classical computer to correct quantum computer's errors ), with the optimal values determined classically (= the optimal exact solutions were easily obtained by a classical computer )"
p.6-right says "each iteration of BF-DCQO was run with 2000 shots across all instances. Furthermore, if the optimal solution was not directly obtained from (error-prone quantum computer's) hardware,... a (classical computer's) post-processing technique (referred to as PP)"
p.7-Fig.2 shows in this BF-DCQO method, almost all answers which were given by the error-prone 36-ion qubits were wrong, Not exact optimal solution (= did Not reach the right-black line ) obtained by the errorless classical computer.
↑ So this research used just 22~36 impractical ion qubits to try to get the optimal (= lowest-energy ) small protein solution in the fictitious simplified protein's energy equation, which almost always gave wrong answers (= Not reach the exact optimal solution obtained by the errorless classical computer ).
↑ They got 2000 results in each iteration, and artificially chose the best (= still wrong ) solution which was used in the next iteration, which was repeated 10 times (+ classical computer's post-processing in this BF-DCQO method, this-p.5-Fig.4 ).
↑ The fact that the error-prone quantum computer with just less than 36 qubits could Not obtain the exact optimal solution given by the classical computer means the quantum computers are impractical in protein folding problems with No quantum advantage,
Feel free to link to this site.
