Detailed Action
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 09/04/2025 has been entered.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is in response to the RCE filed 09/04/2025 for the amended claims filed 09/04/2025 for application 17/513,012, in which:
Claims 1, 10, and 19 are independent claims.
Claim 5-7, 14-16, and 20 were previously cancelled.
Claim 21 has been newly added.
Claims 1, 8, 10, 11-13, and 17-19 are currently amended.
Claims 1-4, 8-13, and 17-19 and 21 are currently pending.
Response to Arguments
Applicant's arguments filed 09/04/2025 have been fully considered but they are not persuasive.
Regarding the 35 USC § 101 Rejections:
Applicant's arguments regarding the 35 U.S.C. 101 rejections of the previous office action have been fully considered, but are unpersuasive.
Applicant asserts (Page 7-8), that Claims 1-4, 8-13, and 17-19 do not recite alleged mental processes and mathematical concepts as the independent claims recite a concrete feedback loop with a physical system which leads the claimed method to be beyond a mere abstract idea and is instead directed to a mathematical model applied to a particular physical/technological environment.
Examiner respectfully disagrees. For the reasons given below and in the 35 U.S.C. § 101 rejections, the claims are directed to an abstract idea (Step 2A Prong 1) and do not integrate the abstract idea into a practical application (Step 2A Prong 2). The pending claims recite abstract ideas that fall in at least one of the permissible groups, and noted within the office action below in more details. The independent claims fail to recite the steps that achieve the improvement. The independent claims are no more detailed than to obtain a training set and train a neural network to make a prediction for control with no steps on how to achieve an improvement by the determination as there is no particular solution to a particular problem; thus, the Claims are not a technical solution to a technical problem. The rejections have been updated below, the rejection to all Claims (including Claim 1, analogous independent Claims, and all dependent Claims) are maintained and updated as necessitated by Claim amendments. Applicant's arguments are not persuasive.
Applicant further asserts (Page 8), that the independent claims are not directed to a mathematical concept due to the “mapping” as claimed is more than simply manipulating existing data and organizing the data into a new form. Instead, mapping involves combining information of the parametrized physical model with additional information.
Examiner respectfully disagrees. The limitation within the independent claim for mapping the parametrized physical model to a restricted Boltzmann machine… is merely a mathematical relationship between variables and/or numbers using a mathematical formula/equations. The term “mapping” within the limitation indicates the mathematical relationship between variables or numbers; where the variables or numbers is the physical model being mapped to a restricted Boltzmann machine which is a Boltzmann machine is a mathematical entity. Under broadest reasonable interpretation, the limitations noted above fall within the group of “mathematical concepts”. This is a form of organizing information and manipulating information through mathematical correlations (see MPEP 2106.04(a)(2)I.A.). As noted in the rejection, the Claim does not include additional elements that are sufficient to amount to an integration of the identified abstract idea into a practical application, thus the Claim is directed to an abstract idea.
Applicant further asserts (Page 9), that the independent claims are not directed to a mental processes due to the determining one or more parameters of a parametrized physical model representing the complex system using the training data and predicting properties or behavior of the system from the parametrized model are impossible tasks for the human mind because the tasks require far too many computations and require tracking too many variables for the human mind to manage simultaneously.
Examiner respectfully disagrees. As noted in the rejection, the Subject Matter Eligibility Analysis Step 2A Prong the determining … limitation, noted in a, interpreted within the broadest reasonable interpretation, is able to be performed by a human mind where a human mind can mentally apply evaluation to determine one or more parameters of a model. The … wherein limitation , noted in b, is merely a mathematical relationship between variables and/or numbers using a mathematical formula/equations. While the predicting … limitation , noted in c, is another mental process as a human being is able to predict one or more properties of a system from a model. A Claim that encompasses a human performing the steps mentally with or without a physical aid recites a Mental Process. A Claim that requires a computer may still recite a mental process. Thus the Claim has limitations that are directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper. Please see MPEP 2106.04(a)(2)III.B and 2106.04(a)(2)III.C.
Applicant asserts (Page 10), that the present claims are not directed to an abstract idea but instead to a specific, technological improvement for modeling real physical quantum or many-body systems via a Maximum Entropy Restricted Boltzmann Machine. This technological improvement is attained by as solving a recognized technical problem (determining Hamiltonian parameters, utilizing an entropy-based cost function and an RBM) which meets the significantly more requirement.
Examiner respectfully disagrees. Although the Claims are interpreted in light of the specification, limitations from the specification are not read into the Claims. The claims are directed towards the improvement of an abstract idea.
MPEP 2106.05(a) recites:
After the examiner has consulted the specification and determined that the disclosed invention improves technology, the claim must be evaluated to ensure the claim itself reflects the disclosed improvement in technology. Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316, 120 USPQ2d 1353, 1359 (Fed. Cir. 2016) (patent owner argued that the claimed email filtering system improved technology by shrinking the protection gap and mooting the volume problem, but the court disagreed because the claims themselves did not have any limitations that addressed these issues). That is, the claim must include the components or steps of the invention that provide the improvement described in the specification
…
It is important to note, the judicial exception alone cannot provide the improvement. The improvement can be provided by one or more additional elements. See the discussion of Diamond v. Diehr, 450 U.S. 175, 187 and 191-92, 209 USPQ 1, 10 (1981)) in subsection II, below. In addition, the improvement can be provided by the additional element(s) in combination with the recited judicial exception.
Applicant fails to show how any alleged technical improvement would be provided by anything more than the judicial exception on its own. Additionally, applicant fails to show how the claim includes components or steps that would provide the alleged improvement described in the specification. Improvements to an abstract idea are still considered to an abstract idea. The Claims do not reflect any improvement in the functioning of a computer or hardware processor rather the claims merely use a generic computer component to perform the abstract ideas. Moreover, the examiner maintains that the Claim does not impose any meaningful limits on the judicial exception. As noted in the rejection, the Claim does not include additional elements that are sufficient to amount to an integration of the identified abstract idea into a practical application, thus the claim is directed to an abstract idea.
Applicant asserts (Page 10), that due to the reasons listed above that Claim 1 is patent eligible under 35 USC § 101 and independent claims 10 and 19 are patent eligible for the same reasons. Claims 2-4, 8, 9, 11-13, 17, and 18 are patentable at least because of their dependencies and for the subject matter they recite.
Examiner respectfully disagrees. For the reasons given above and in the updated rejections below, the rejection to all Claims (including Claim 1, analogous independent Claims, and all dependent Claims) are maintained and updated as necessitated by the Claim amendments.
Regarding the 35 USC § 103 Rejections:
Applicant's arguments regarding the 35 U.S.C. 103 rejections of the previous office action have been fully considered, but are unpersuasive.
Applicant asserts (Page 10-11) that Torlai fails to disclose the previously presented limitations of determining … and mapping … and the training data being experimental data collected from a complex physical system. And that only simulated data is used by Torlai. Applicant further asserts (Page 11) that the amended limitation training the restricted Boltzmann machine on the training data representing the complex system using an explicit maximum entropy principle, the training comprising using an objective function with an entropy-based term is also not disclose by Torlai and Huembeli and that there is no motivation to combine Huembeli and Torlai (Page 12).
Examiner respectfully disagrees. Torlai teaches the limitation for determining one or more parameters of a parametrized physical model representing the complex system using the training data, … by showing the complex physical spin system using training data which is sampled and generated to represent the system. The independent Claim limitation recites obtaining training data representing the complex system, wherein the training data is collected from a complex physical system which is taught by Torlai as the training data is generated from samples which is still a form of collected data from a complex physical system. While Huembeli teaches wherein the parametrized physical model comprises a Bose-Hubbard Hamiltonian comprising one or more nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential with Equation 4 showing the Hamiltonian comprising nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential. In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, Huembeli shows within the Abstract, “We demonstrate that the Boltzmann machine can faithfully reproduce the observables of the physical system. Further, we observe that the number of neurons required to obtain accurate results increases as the system is brought close to criticality”; Page 4, Column 1, Paragraph 3, “We now apply our method to several paradigmatic models to benchmark its performance”; Page 6, Column 1, Paragraph 2, “… Conversely, we have demonstrated that the performance of a Boltzmann machine may be evaluated using a comparison of thermodynamic observables calculated from both the physical and modeled distribution …”. Torlai teaches using a Restricted Boltzmann machine as a parameterized neural network and Huembeli teaches phase transitions of the Hamiltonian; one of ordinary skill in the art would find it obvious to combine these two references to utilize Torlai for reviewing data and Huembeli’s model comprising one or more nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential.
Applicant asserts (Page 10-11) that Torlai and Huembeli (Page 12) fails to disclose the limitations newly added limitations determining one or more modifications to the complex system to drive the system into a target state based on the one or more parameters; and outputting a control signal to apply the one or more modifications to the complex physical system.
Applicant’s arguments with respect to claim(s) [ 1, 10, 19] have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant asserts (Page 12-13) that due to neither Torlai and Huembeli disclosing the above features, and the combination not able to suggest the features of the claim that independent Claim 1, 10, and 19 should be considered for allowance. Also, the dependent claims should be considered for allowance due to dependency.
Applicant’s arguments regarding the other independent and dependent claims rely upon the same assertions as with respect to Claim 1, and are thus likewise unpersuasive.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-4, 8-13, and 17-19, and 21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1:
Subject Matter Eligibility Analysis Step 1:
Claim 1 recites a method, thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
However, Claim 1 further recites the method comprising of:
determining one or more parameters of a parametrized physical model representing the complex system using the training data … (a human being can mentally apply evaluation to determine one or more parameters of the model)
… wherein the parametrized physical model comprises a Bose-Hubbard Hamiltonian comprising one or more nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential (a mathematical relationship between variables and/or numbers using a mathematical formula/equations)
predicting one or more properties of the complex system and/or behavior of the complex system from the parametrized physical model (a human being can make a judgement to predict one or more properties of the system from the model)
determining one or more modifications to the complex system to drive the system into a target state based on the one or more parameters (a human being can make a mental apply evaluation to determine modification(s) to the complex system to attain a target state for the system based on parameters)
mapping the parametrized physical model to a restricted Boltzmann machine, comprising mapping the one or more nearest-neighbor hopping amplitudes to a weight matrix of the restricted Boltzmann machine and mapping the on-site interaction strength and chemical potential to bias offsets of hidden units of the restricted Boltzmann machine (a mathematical relationship between variables and/or numbers using a mathematical formula/equations)
Claim 1 thus recites an abstract idea (that falls into the “mathematical concepts” or “mental processes” group of abstract ideas).
Subject Matter Eligibility Analysis Step 2A Prong 2:
This judicial exception is not integrated into a practical application because the additional elements recited consists of:
A method for modelling a complex system using machine learning, the method comprising: (to perform a mathematical concept and the performance of an abstract idea on a computer is no more than instructions to “apply it” on a computer, by MPEP 2106.05(f))
obtaining training data representing the complex system, wherein the training data is collected from a complex physical system (which is insignificant extra-solution activity of data gathering, by MPEP 2106.05(g))
outputting a control signal to apply the one or more modifications to the complex physical system (which is insignificant extra-solution activity of data gathering and outputting, by MPEP 2106.05(g))
training the restricted Boltzmann machine on the training data representing the complex system using an explicit maximum entropy principle, the training comprising using an objective function with an entropy-based term (to perform a mathematical concept and the performance of an abstract idea on a computer is no more than instructions to “apply it” on a computer, by MPEP 2106.05(f))
extracting the one or more parameters of the parametrized physical model from the trained restricted Boltzmann machine (to perform a mental process and the performance of an abstract idea on a computer is no more than instructions to “apply it” on a computer, by MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements recited, alone or in combination, do not provide significantly more than the abstract idea itself. Additional elements a, d, and e are merely applying the abstract idea on a computer (MPEP 2106.05(f)) which cannot provide significantly more. Additional elements b and c falls within MPEP 2106.05(d) as well-understood, routine and conventional activities of receiving or transmitting data over a network (MPEP 2106.05(d)(II): buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014)). Thus, the claim is subject-matter ineligible.
Regarding Claim 2:
Subject Matter Eligibility Analysis Step 1:
Dependent Claim 2 recites the method of Claim 1. Claim 1 is a method, thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
However, Claim 2 further recites the method comprising of applying an optimisation procedure to an objective function comprising an entropy-based term derived from the Bose-Hubbard Hamiltonian (which is a mathematical relationship between variables and/or numbers using a mathematical formula/equations). Claim 2 thus recites an abstract idea (that falls into the “mathematical concepts” group of abstract ideas).
Subject Matter Eligibility Analysis Step 2A Prong 2:
This judicial exception is not integrated into a practical application because the new sole additional element recited consists of wherein training the restricted Boltzmann machine on the training data representing the complex system comprises (to perform a mental process and the performance of an abstract idea on a computer is no more than instructions to “apply it” on a computer, by MPEP 2106.05(f))
Subject Matter Eligibility Analysis Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the new sole additional element recited, alone or in combination, does not provide significantly more than the abstract idea itself. The additional element is merely applying the abstract idea on a computer (MPEP 2106.05(f)) which cannot provide significantly more. Thus, the claim is subject-matter ineligible.
Regarding Claim 3:
Subject Matter Eligibility Analysis Step 1:
Dependent Claim 3 recites the method of Claim 2. Claim 2 is a method, thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
However, Claim 3 further recites the method comprising wherein the entropy-based term comprises a Shannon entropy which is a mathematical relationship between variables and/or numbers using a mathematical formula/equations. Claim 3 thus recites an abstract idea (that falls into the “mathematical concepts” group of abstract ideas).
Subject Matter Eligibility Analysis Step 2A Prong 2:
This judicial exception is not integrated into a practical application because there are no new additional elements recited.
Subject Matter Eligibility Analysis Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because there are no new additional elements recited. The judicial exception alone does not provide significantly more than the abstract idea itself. Thus, the claim is subject-matter ineligible.
Regarding Claim 4:
Subject Matter Eligibility Analysis Step 1:
Dependent Claim 4 recites the method of Claim 2. Claim 2 is a method, thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
However, Claim 4 further recites wherein the entropy-based term comprises an average linear entropy, and wherein the optimisation procedure comprises the use of an artificial bee colony method, a pattern search method and/or a gradient search optimisation method which is a mathematical relationship between variables and/or numbers using a mathematical formula/equations. Claim 4 thus recites an abstract idea (that falls into the “mathematical concepts” group of abstract ideas).
Subject Matter Eligibility Analysis Step 2A Prong 2:
This judicial exception is not integrated into a practical application because there are no new additional elements recited.
Subject Matter Eligibility Analysis Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because there are no new additional elements recited. The judicial exception alone does not provide significantly more than the abstract idea itself. Thus, the claim is subject-matter ineligible.
Regarding Claim 8:
Subject Matter Eligibility Analysis Step 1:
Dependent Claim 8 recites the method of Claim 1. Claim 1 is a method, thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
However, Claim 8 further recites the method comprising of determining a critical temperature from the Bose-Hubbard model (a human being can mentally apply evaluation to determine a critical temperature from a model). Claim 8 thus recites an abstract idea (that falls into the “mental processes” group of abstract ideas).
Subject Matter Eligibility Analysis Step 2A Prong 2:
This judicial exception is not integrated into a practical application because the new sole additional element recited consists of wherein determining one or more properties of the complex system from the parametrized model comprises (which is restricting the abstract idea to a Particular Technological Environment, by MPEP 2106.05(h)).
Subject Matter Eligibility Analysis Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the new sole additional element recited, alone or in combination, does not provide significantly more than the abstract idea itself. The additional element is only restricting the abstract idea to a Particular Technological Environment (MPEP 2106.05(h)) which cannot provide significantly more. Thus, the claim is subject-matter ineligible.
Regarding Claim 9:
Subject Matter Eligibility Analysis Step 1:
Dependent Claim 9 recites the method of Claim 1. Claim 1 is a method, thus a process, one of the four statutory categories of patentable subject matter.
Subject Matter Eligibility Analysis Step 2A Prong 1:
However, Claim 9 does not recite any additional abstract ideas and only inherits the abstract ideas from Claim 1. Claim 9 thus recites an abstract idea (that falls into the “mental processes” group of abstract ideas).
Subject Matter Eligibility Analysis Step 2A Prong 2:
This judicial exception is not integrated into a practical application because the new sole additional element recited consists of wherein the complex system comprises a physical or biological system (which is restricting the abstract idea to a Particular Technological Environment, by MPEP 2106.05(h)).
Subject Matter Eligibility Analysis Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the new sole additional element recited, alone or in combination, does not provide significantly more than the abstract idea itself. The additional element is only restricting the abstract idea to a Particular Technological Environment (MPEP 2106.05(h)) which cannot provide significantly more. Thus, the claim is subject-matter ineligible.
Regarding Claims 10-13 and 17-18:
Claims 10-13 and 17-18 incorporate substantively all the limitations of Claims 1-4 and 8-9 in a non-transitory computer-readable storage media (thus, a manufacture) and further recites A non-transitory, computer-readable medium containing instructions that, when executed by a computer, cause the computer to perform a method for modelling a complex system using machine learning, the method comprising (these claim limitations appear to perform a mental process and the performance of an abstract idea on a computer is no more than instructions to “apply it” on a computer, by MPEP 2106.05(f)) and does not appear to integrate the abstract idea into a particular application; thus, the claim is subject -matter ineligible as it does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself); thus, Claims 10-13 and 17-18 are rejected for reasons set forth in the rejections of Claims 1-4 and 8-9, respectively.
Regarding Claims 19 and 21:
Claims 19 and 21 incorporate substantively all the limitations of Claims 1 and 2 in a system (thus, a machine) and further recites A system comprising one or more processors and a memory, the memory containing computer-readable instructions that, when executed by the one or more processors, causes the system to perform a method for modelling a complex (these claim limitations appear to perform a mental process and the performance of an abstract idea on a computer is no more than instructions to “apply it” on a computer, by MPEP 2106.05(f)) and does not appear to integrate the abstract idea into a particular application; thus, the claim is subject -matter ineligible as it does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself); thus, Claims 19 and 21 are rejected for reasons set forth in the rejection of Claims 1 and 2, respectively.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-4, 8-13, and 17-19 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Torlai et al., “Learning thermodynamics with Boltzmann machines”, in view of Huembeli et al., “Identifying quantum phase transitions with adversarial neural networks”, in view of Doria et al., “Optimal Control Technique for Many-Body Quantum Dynamics”.
Regarding Claim 1:
Torlai teaches:
A method for modelling a complex system using machine learning, the method comprising:
(Torlai, Abstract, “… Boltzmann machine that is capable of modeling thermodynamic observables for physical systems …”).
obtaining training data representing the complex system, wherein the training data is collected from a complex physical system;
(Torlai, Abstract, “… train the Boltzmann machine on data sets … sampled … at different temperatures …”).
determining one or more parameters of a parametrized physical model representing the complex system using the training data, … ;
(Torlai, Page 1, Column 2, Paragraph 4, “In constructing a Boltzmann machine, our goal is to build an approximate model of a target probability distribution … spin configurations can be generated by a Boltzmann machine that was trained by the MC samples of the target distribution … sampling the target distribution for an Ising spin Hamiltonian …”; Page 2, Column 1, Paragraph 3, “Given a target probability distribution pS(σ) defined over a set of random variables σ, our goal is to build a probabilistic model pλ(σ) which mimics our target distribution. The model is in general characterized by a set of parameters λ, which we will tune in order to minimize the distance between these two probability distributions”. The parametrized physical model is a probabilistic model representing the complex physical system of spin systems. The model uses training data (via sampling the target distribution) to represent the system).
and predicting one or more properties of the complex system and/or behavior of the complex system from the parametrized physical model:
(Torlai, FIG. 1; Page 2, Column 1, Paragraph 1, “In Sec. III we present the results for thermodynamic observables obtained from this Boltzmann machine, trained on finite-temperature configurations produced from the nearest neighbor Ising ferromagnet”; Page 6, Column 1, Paragraph 1, “Once trained, we evaluated different thermodynamic estimators on the samples generated by the Boltzmann machines and show that they faithfully reproduce those calculated directly from the Monte Carlo samples”; Page 2, Column 2, Paragraph 1, “The weights of the edges are described by a matrix W with zero diagonal, where the element Wij is the weight on the edge connecting hi to σj. We also introduce two external fields b and c coupled to the visible and hidden layers, respectively … The full probability distribution defined by the graph can be written as a Boltzmann distribution
PNG
media_image1.png
52
275
media_image1.png
Greyscale
where the model parameters are λ = {W,b,c} and the energy is given by
PNG
media_image2.png
42
367
media_image2.png
Greyscale
”. The thermodynamic observables (which are interpreted by the examiner as one or more properties of the complex system) are predicted by the Boltzmann machine’s model (shown in FIG. 1) and compared to the Monte Carlo samples. The physical properties of the complex system are described in the parametrized physical model shown in FIG. 1 with the external field parameters).
…
wherein determining parameters of the parametrized physical model representing the complex system using the training data comprises: mapping the parametrized physical model to a restricted Boltzmann machine,
(Torlai, FIG. 1; Page 3, Column 1, Paragraph 4, “The classical spin system we choose to train the Boltzmann machine on is the Ising Hamiltonian,
PNG
media_image3.png
43
268
media_image3.png
Greyscale
with ferromagnetic interactions J = 1 between nearest neighbors”. The Ising Hamiltonian model representing the physical system (spin system) is mapped to the Restricted Boltzmann machine for the physical system which is shown in FIG. 1).
comprising mapping … a weight matrix of the restricted Boltzmann machine and mapping … bias offsets of hidden units of the restricted Boltzmann machine;
(Torlai, FIG. 1; Page 2, Column 2, Paragraph 1, “… The weights of the edges are described by a matrix W with zero diagonal, where the element Wij is the weight on the edge connecting hi to σj .We also introduce two external fields b and c coupled to the visible and hidden layers, respectively …
PNG
media_image4.png
38
321
media_image4.png
Greyscale
…
PNG
media_image5.png
45
277
media_image5.png
Greyscale
…”. FIG 1. shows a visual structure of the restricted Boltzmann machine (RBM) where W indicates the symmetric matrix of weight and h depicts the hidden nodes (nodes are interpreted by the examiner as units). Equation 4 shows the mapping of the weight matrix for calculating energy and Equation 7 shows the mapping of the bias offsets of hidden units (where bj is the bias offset for the hidden nodes h)).
training the restricted Boltzmann machine on the training data representing the complex system using a maximum entropy principle, the training comprising using an objective function with an entropy-based term;
(Torlai, Page 2, Column 2, Paragraph 1, “The full probability distribution defined by the graph can be written as a Boltzmann distribution
PNG
media_image6.png
51
272
media_image6.png
Greyscale
where ZS … is the canonical partition function … ”; Page 6, Column 2, Paragraph 2, “We have seen in Sec. II that the training of a Boltzmann machine consists in solving an optimization problem where the function to minimize is the KL divergence between the model probability and the target probability. …
PNG
media_image7.png
20
282
media_image7.png
Greyscale
is the entropy of the data set”. Boltzmann machines are inherently based on the maximum entropy principle. The training using the probability distribution defined in equation 3 which is based on equation A1 within the appendix which includes the entropy of the dataset for the first term. Thus, the training comprising using a objective optimization function with an entropy based term).
extracting the one or more parameters of the parametrized physical model from the trained restricted Boltzmann machine.
(Torlai, Page 3, Column 1, Paragraph 2, “… the training process consists of tuning the machine parameters λ until the pλ(σ) is close to the target distribution pS(σ). This is equivalent to solving an optimization problem… The network parameters are optimized using a stochastic version of the gradient descent, which consists of updating all the parameters …”; Page 4, Column 1, Paragraph 1, “We update the parameters with CD20 using stochastic gradient descent with learning rate η = 0.01 and minibatch size of 50 samples”. The parameters are retrieved (extracting) and updated via CD and KL divergences which extracts the parameters to be able to increment for the optimization procedure).
Torlai fails to explicitly disclose:
wherein the parametrized physical model comprises a Bose-Hubbard Hamiltonian comprising one or more nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential;
… the one or more nearest-neighbor hopping amplitudes to … the on-site interaction strength and chemical potential to …
However, Huembeli teaches:
wherein the parametrized physical model comprises a Bose-Hubbard Hamiltonian comprising one or more nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential
… the one or more nearest-neighbor hopping amplitudes to … the on-site interaction strength and chemical potential to …
(Huembeli, Page 4, Column 1, Paragraph 5, “ …We investigate the 2D Bose-Hubbard model [Fig. 2(a)] with Hamiltonian
PNG
media_image8.png
67
337
media_image8.png
Greyscale
chemical potential μ, nearest neighbor hopping J , and onsite interaction strength U. ” Huembeli teaches the Bose-Hubbard model with Hamiltonian comprising nearest-neighbor hopping amplitudes, an on-site interaction strength, and a chemical potential).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the Bose-Hubbard Hamiltonian model to determine a critical temperature taught by Huembeli, in the method taught by Torlai to yield a specific parametrized model within the Restricted Boltzmann Machine to benchmark performance, handle quantum data, phase discovery, and extend the RBM to handle quantum probability distributions (Huembeli, Abstract, “We demonstrate that the Boltzmann machine can faithfully reproduce the observables of the physical system. Further, we observe that the number of neurons required to obtain accurate results increases as the system is brought close to criticality”; Page 4, Column 1, Paragraph 3, “We now apply our method to several paradigmatic models to benchmark its performance”; Page 6, Column 1, Paragraph 2, “… Conversely, we have demonstrated that the performance of a Boltzmann machine may be evaluated using a comparison of thermodynamic observables calculated from both the physical and modeled distribution …”. Huembeli shows the 3 different models that are used within their experiments to compare and benchmark performance of a Boltzmann machine using thermodynamic observables).
Torlai in view of Huembeli does not explicitly disclose:
determining one or more modifications to the complex system to drive the system into a target state based on the one or more parameters; and
outputting a control signal to apply the one or more modifications to the complex physical system,
However, Doria teaches:
determining one or more modifications to the complex system to drive the system into a target state based on the one or more parameters; and
(Doria, Page 2, Column 1, Paragraph 2, “CRAB method.— … an optimal control problem … given a system described by a Hamiltonian H depending on some control parameters … the goal is to find the cj’s time dependence (pulse shape) that extremizes a given figure of merit F, for instance, the final system energy, state fidelity, or entanglement”; FIG. 1. FIG. 1 shows the an optical lattice described by a Bose-Hubbard model (interpreted by the examiner as a complex system) defined by the Hamiltonian model which is controlled with the CRAB methodology of Doria. The CRAB method teaches determining optimal control values for a system by taking control parameters and determining optimal modifications by applying the pulse shape (control signal) to optimize the system based on parameters to get to a final system energy/state fidelity/entanglement (interpreted as a target state by the examiner)).
outputting a control signal to apply the one or more modifications to the complex physical system,
(Doria, Page 2, Column 1, Paragraph 2, “… the goal is to find the cj’s time dependence (pulse shape) that extremizes a given figure of merit F, for instance, the final system energy, state fidelity, or entanglement … an explicative example, here we focus on the case where the correction is of the form cj(t) = c0j(t)fj(t) and the functions fj(t) can be simply expressed in a truncated Fourier space”; Page 2-3, Equations 1 & 2; Page 3, FIG. 3; Page 3, Column 1, Paragraph 1, “… system parameters U and J can be expressed as a function of the optical-lattice depth V …”. Equations 1 shows the control function and Equation 2 shows the Bose-Hubbard Hamiltonian model in an optical lattice which is shown in FIG. 1. The CRAB method teaches determining optimal control values for a system by taking control parameters and determining optimal modifications by applying the pulse shape (control signal) to optimize the complex physical system which can be shown in Fig. 3).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the determination of modification(s) and applying the modification(s) via outputting a control signal taught by Doria with the method taught by Torlai/Huembeli as Doria utilizes a control operation for optimizing a complex physical system, quantum technologies, parallelization and shows robustness (Doria, Page 4, Column 1, Paragraph 2, “This robustness is crucial as the experimental realization of these systems is performed in parallel on many different one-dimensional tubes with different numbers of atoms [19]. … Outlook.—In conclusion, we would like to mention that the CRAB optimization strategy introduced here can in principle be applied also to open quantum many-body systems … The optimization might be performed during the experimental repetitions …, thus adding a small overhead to the experimental complexity. This would extend the applicability of the CRAB method to the optimization of quantum phenomena that are completely out of reach for simulation on classical computers, and represent a major design tool for future quantum technologies”).
Regarding Claim 2:
Torlai/Huembeli/Doria teach the method of Claim 1. Torlai further teaches:
wherein training the restricted Boltzmann machine on the training data representing the complex system comprises applying an optimisation procedure to an objective function comprising an entropy-based term derived ...
(Torlai, Page 4, Column 1, Paragraph 1, “We update the parameters with CD20 using stochastic gradient descent with learning rate η = 0.01 and minibatch size of 50 samples”; Page 6, Paragraph 1, “We can then write the KL divergence as
PNG
media_image9.png
49
330
media_image9.png
Greyscale
where the first term is called negative log-likelihood and H(pdata) = − Σσ pdata(σ) log pdata(σ) is the entropy of the data set … The increments in the parameters are obtained by averaging the gradient of the KL divergence over the entire data set D … This optimization procedure, called stochastic gradient descent, substantially speeds up the learning, especially when the data set contains a very large number of samples … To obtain an update rule for the gradient descent we need to take the derivative of the KL divergence in Eq. (9), which reduces to the derivative of the log-likelihood”. CD (contrastive divergence) updates the parameters using stochastic gradient descent for the optimization procedure. KL divergence is used when N is large and KL divergence comprises H(pdata) which is an entropy based term).
Torlai fails to explicitly disclose from the Bose-Hubbard Hamiltonian which is taught in Huembeli as noted in Claim 1. Thus, the motivation of Claim 1’s combination of Torlai/Huembeli/Doria is still maintained.
Regarding Claim 3:
Torlai/Huembeli/Doria teach the method of Claim 2. Torlai further teaches:
wherein the entropy-based term comprises a Shannon entropy.
(Torlai, Page 6, Paragraph 1, “We can then write the KL divergence as
PNG
media_image9.png
49
330
media_image9.png
Greyscale
where the first term is called negative log-likelihood and H(pdata) = − Σσ pdata(σ) log pdata(σ) is the entropy of the data set”. H(pdata) is the entropy based term which comprises a Shannon entropy).
The motivation of Claim 1’s combination of Torlai/Huembeli/Doria is still maintained.
Regarding Claim 4:
Torlai/Huembeli/Doria teach the method of Claim 2. Torlai further teaches:
wherein the entropy-based term comprises an average linear entropy, and wherein the optimisation procedure comprises the use of an artificial bee colony method, a pattern search method and/or a gradient search optimisation method.
(Torlai, Page 6, Column 2, Paragraph 1, “The optimization problem is solved by stochastic gradient descent. We choose an initial point λ(0) in the full configuration space with zero external fields and weights Wij randomly drawn from a uniform distribution centered around zero. The increments in the parameters are obtained by averaging the gradient of the KL divergence over the entire data set D … This optimization procedure, called stochastic gradient descent, substantially speeds up the learning, especially when the data set contains a very large number of samples”. The examiner interprets average linear entropy as the average value of the data across all possible events in a probability distribution (linear operation). Thus, Shannon entropy is interpreted as the average linear entropy. Stochastic gradient descent is a gradient search optimisation method).
The motivation of Claim 1’s combination of Torlai/Huembeli/Doria is still maintained.
Regarding Claim 8:
Torlai/Huembeli/Doria teach the method of Claim 1.
Torlai fails to teach the Bose-Hubbard model and critical temperatures.
However, Huembeli teaches:
wherein determining one or more properties of the complex system from the parametrized model comprises determining a critical temperature from the Bose-Hubbard model.
(Huembeli, FIG. 3; Page 4, Column 2, Paragraph 1, “This model experiences phase transitions at zero temperature from Mott insulating to superfluid phases … we apply the domain adaptation algorithm on states of the whole phase diagram. Results are presented in Fig. 3”. The examiner interprets a critical temperature from the Bose-Hubbard model as a temperature when the system undergoes a phase transition (between a superfluid state and a Mott insulator state) which is best shown in FIG. 3).
The motivation of Claim 1’s combination of Torlai/Huembeli/Doria is still maintained.
Regarding Claim 9:
Torlai and Huembeli teach the method of Claim 1. Torlai further teaches:
wherein the complex system comprises a physical or biological system.
(Torlai, Abstract, “In this paper, we develop a Boltzmann machine that is capable of modeling thermodynamic observables for physical systems in thermal equilibrium”).
The motivation of Claim 1’s combination of Torlai/Huembeli/Doria is still maintained.
Regarding Claims 10-13 and 17-18:
Claims 10-13 and 17-18 incorporate substantively all the limitations of Claims 1-4 and 8-9 in a non-transitory computer-readable medium containing instructions that, when executed by a computer, cause the computer to perform (Torlai, Page 1, Column 1, Paragraph 1, “Machine learning is a paradigm whereby computer algorithms … automatic encoding proceeds by first “training” the algorithm on a large data set and then asking the trained machine to perform some task.”. The machine learning is interpreted to be developed and used on computer systems that store/read the computer algorithms on non-transitory computer readable mediums to allow for storage/retrieval/transfer of data) the method of Claim 1); thus, Claims 10-15 and 17-18 are rejected for reasons set forth in the rejections of Claims 1-4 and 8-9, respectively.
Regarding Claims 19 and 21:
Claims 19 and 21 incorporate substantively all the limitations of Claims 1 and 2 in a system comprising one or more processors and a memory, the memory containing computer-readable instructions that, when executed by the one or more processors, causes the system to perform (Torlai, Page 1, Column 1, Paragraph 1, “Machine learning is a paradigm whereby computer algorithms are designed to learn from and make predictions on data … Such automatic encoding proceeds by first “training” the algorithm on a large data set and then asking the trained machine to perform some task.” Automatic encoding is the process of memory where information is retrieved and encoded automatically. The examiner interprets the automatic encoding to be done on a system with one or more processors with a memory to perform the computer algorithms) the method of Claim 1; thus, Claims 19 and 21 are rejected for reasons set forth in the rejection of Claims 1 and 2, respectively.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IBRAHIM RAHMAN whose telephone number is (703)756-1646. The examiner can normally be reached M-F 8am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kakali Chaki can be reached at (571) 272-3719. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197